lidong.wu@utdallas
description
Transcript of lidong.wu@utdallas
lidongwuutdallasedu
Small Sensor Big Data
Ding-Zhu DuUniversity of Texas at Dallas
lidongwuutdallasedu
Small Sensor and Big Data
Lidong WuUniversity of Texas at Dallas
Sensor
Drowning in Vast Amount of DataDigitized World
BigData
Outline
I Data Collection in Sensor System
II Data Analysis on Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
Outline
I Data Collection in Sensor System
II Data Analysis on Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
Have you watched movie Twister
sensorBucket ofsensors
tornado
Where are all the sensors
Smartphone with a dozen of sensors
Where are all the sensors
Wearable devices - Google Glass Applersquos iWatch
Buildings
Where are all the sensors
Transportation systems etc
Where are all the sensors
Sensor Web
Large of simple sensors Usually deployed randomly Multi-hop wireless link Distributed routing No infrastructure Collect data and send it to base station
Applications of Senor Web
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
lidongwuutdallasedu
Small Sensor and Big Data
Lidong WuUniversity of Texas at Dallas
Sensor
Drowning in Vast Amount of DataDigitized World
BigData
Outline
I Data Collection in Sensor System
II Data Analysis on Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
Outline
I Data Collection in Sensor System
II Data Analysis on Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
Have you watched movie Twister
sensorBucket ofsensors
tornado
Where are all the sensors
Smartphone with a dozen of sensors
Where are all the sensors
Wearable devices - Google Glass Applersquos iWatch
Buildings
Where are all the sensors
Transportation systems etc
Where are all the sensors
Sensor Web
Large of simple sensors Usually deployed randomly Multi-hop wireless link Distributed routing No infrastructure Collect data and send it to base station
Applications of Senor Web
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Sensor
Drowning in Vast Amount of DataDigitized World
BigData
Outline
I Data Collection in Sensor System
II Data Analysis on Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
Outline
I Data Collection in Sensor System
II Data Analysis on Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
Have you watched movie Twister
sensorBucket ofsensors
tornado
Where are all the sensors
Smartphone with a dozen of sensors
Where are all the sensors
Wearable devices - Google Glass Applersquos iWatch
Buildings
Where are all the sensors
Transportation systems etc
Where are all the sensors
Sensor Web
Large of simple sensors Usually deployed randomly Multi-hop wireless link Distributed routing No infrastructure Collect data and send it to base station
Applications of Senor Web
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Outline
I Data Collection in Sensor System
II Data Analysis on Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
Outline
I Data Collection in Sensor System
II Data Analysis on Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
Have you watched movie Twister
sensorBucket ofsensors
tornado
Where are all the sensors
Smartphone with a dozen of sensors
Where are all the sensors
Wearable devices - Google Glass Applersquos iWatch
Buildings
Where are all the sensors
Transportation systems etc
Where are all the sensors
Sensor Web
Large of simple sensors Usually deployed randomly Multi-hop wireless link Distributed routing No infrastructure Collect data and send it to base station
Applications of Senor Web
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Outline
I Data Collection in Sensor System
II Data Analysis on Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
Have you watched movie Twister
sensorBucket ofsensors
tornado
Where are all the sensors
Smartphone with a dozen of sensors
Where are all the sensors
Wearable devices - Google Glass Applersquos iWatch
Buildings
Where are all the sensors
Transportation systems etc
Where are all the sensors
Sensor Web
Large of simple sensors Usually deployed randomly Multi-hop wireless link Distributed routing No infrastructure Collect data and send it to base station
Applications of Senor Web
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Have you watched movie Twister
sensorBucket ofsensors
tornado
Where are all the sensors
Smartphone with a dozen of sensors
Where are all the sensors
Wearable devices - Google Glass Applersquos iWatch
Buildings
Where are all the sensors
Transportation systems etc
Where are all the sensors
Sensor Web
Large of simple sensors Usually deployed randomly Multi-hop wireless link Distributed routing No infrastructure Collect data and send it to base station
Applications of Senor Web
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Where are all the sensors
Smartphone with a dozen of sensors
Where are all the sensors
Wearable devices - Google Glass Applersquos iWatch
Buildings
Where are all the sensors
Transportation systems etc
Where are all the sensors
Sensor Web
Large of simple sensors Usually deployed randomly Multi-hop wireless link Distributed routing No infrastructure Collect data and send it to base station
Applications of Senor Web
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Where are all the sensors
Wearable devices - Google Glass Applersquos iWatch
Buildings
Where are all the sensors
Transportation systems etc
Where are all the sensors
Sensor Web
Large of simple sensors Usually deployed randomly Multi-hop wireless link Distributed routing No infrastructure Collect data and send it to base station
Applications of Senor Web
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Buildings
Where are all the sensors
Transportation systems etc
Where are all the sensors
Sensor Web
Large of simple sensors Usually deployed randomly Multi-hop wireless link Distributed routing No infrastructure Collect data and send it to base station
Applications of Senor Web
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Transportation systems etc
Where are all the sensors
Sensor Web
Large of simple sensors Usually deployed randomly Multi-hop wireless link Distributed routing No infrastructure Collect data and send it to base station
Applications of Senor Web
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Sensor Web
Large of simple sensors Usually deployed randomly Multi-hop wireless link Distributed routing No infrastructure Collect data and send it to base station
Applications of Senor Web
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Applications of Senor Web
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
observerAn example of sensor web
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Whatrsquos Sensor
Small size Large number Tether- lessBUThellip
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Whatrsquos limiting the task
Energy Sense Communication scale
CPU
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Challenge
Target is Covered
Sensor system is Connected
Coverage amp Connectivity
Golden Rule then we say
System is alive
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
sensor
target
communication radius
sensing radius
Rc
Rs
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Coverage amp Connectivity
Communication Range
Sensing Range
d le Rs
d le Rc
sensor
target
communication radius
sensing radius
Rc
Rs
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Min-Connected Sensor Cover Problem
Figure Min-CSC Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
[Resource Saving]
communicationnetwork
sensing disks
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Previous Work for PTAS
Itrsquos NP-hard
Ο(r ln n) ndash approximation given by Gupta Das and Gu [MobiHocrsquo03 2003] where n is the number of sensors and r is the link radius of the sensor network
Min-Connected Sensor Cover Problem
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Main Results
Random algorithm
Ο(log3n log log n)-approximation n is the
number of sensors
Partition algorithm
Ο(r)-approximation r is the link radius of the
network
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
With a random algorithm which with probability 1- ɛ produces an Ο(log3n log log n) - approximation
1
Algorithm 1
Group
Steiner Tree
2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
1 2Min-CSC Min-CTC GST
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Sensor Cover Problem
A uniform set of sensors and a target area
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target area and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
1 2Min-CSC Min-CTC GST
How to map to GST
Min-Connected Target Coverage Problem
A uniform set of sensors and a target POINTS
Find a minimum of sensors
to meet two requirements
[Coverage] cover the target POINTS and
[Connectivity] form a connected communication network
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
1 2Min-CSC Min-CTC GST
A graph G = (V E) with positive edge weight c for every edge e isin E
A specified vertex r
k subsets (or groups) of vertices G1Gk Gi sube V
Find a minimum total weight tree T contains at least one vertex in each Gi
Group Steiner Tree
Figure GST ProblemThis tree has minimum weight
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
1 2Min-CSC Min-CTC GST
Choose at least one sensor from each group
Coverage
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Consider communication network
Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
1 2Min-CSC Min-CTC GST
b3
b1
b2
b6
b5
b4
S1S2
S3S4
b7
Find a group Steiner tree in communication network
Min-Coverage amp Connectivity
b2 b6b3 b4b1 b5 b7
S1 S2
S1 S3
S1 S2 S3
S2 S3
S2 S4
S3 S4
Gi contains all sensors covering bi
S2 S3 S4
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
1 2Min-CSC Min-CTC GST
Garg Konjevod and Ravi [SODA 2000] showed with probability 1- ε an approximation solution of GROUP STEINER TREE on tree metric T is within a factor of Ο(log2 n log log n log k) from optimal
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
What Is Link Radius
3 example In this onintersecti sensing
nonempty having sensors twoof distancefor boundupper least theis radiuslink
r
r
Communication disk
Sensing disk
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Connected
Sensor Cover with Target
Area
Connected
Sensor Cover
with Target Points
Connect output of Min-TC into Min-CTC It can be done in Ο(r) - approximation
1
Algorithm 2
2Min-CSC Min-CTC Min-TC
Refer to my paper [INFOCOM 2013rsquo]
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
There exists a polynomial-time (1 + ε)- approximation for MIN-TC
Green is an opt (TC)
Orange is an approx (TC)
lt (1+ε) opt (TC) lt (1+ε) opt (CTC)
Step 2 Target Coverage
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Byrka et al [6] showed there exists a polynomial-time139-approximation of for Network Steiner Minimum Tree
Green is an opt (Network ST)
Red is an approx (TC)
Step 2 Network Steiner Tree
Let Sprime sube S be a (1 + ε)-approximation for MIN-TC Assign
weight one to every edge of G Interconnect sensors in Sprime to
compute a Steiner tree T as network Steiner minimum tree
All sensors on the tree form an approxfor min CTC
nodes approx for min CTC= edges +1 approx for Network STlt 139 opt (Network ST) +1lt 139 opt (CTC) + 1
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Step 2 Network Steiner Tree
Green is an opt (CTC)
Yellow is an approx (TC)
Each orange line has distance lt r
opt (Network ST)
lt opt (CTC) -1 + r = opt (CTC) O(r)
Note lt (1+ε) opt (CTC)
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Future Works
Ο(log3n log log n)
n is the number of sensors
1 Unknown Relationship
2 Constant-appro for Min-CSC
Ο(r)
r is the link radius
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
ldquoConstant-Approximations for Target Coverage Problem in Wireless Sensor Networksrdquo INFOCOM2012 (with Weili Wu et al)
ldquoApproximations for Minimum Connected Sensor Coverrdquo INFOCOM2013 (with Weili Wu et al)
ldquoPTAS for Routing-Cost Constrained Minimum Connected Dominating Sets helliprdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
ldquoAn Approximation Algorithm for Client Assignment helliprdquo INFOCOM2014 (with Weili Wu et al)
What I have donePublications on Optimization
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
CCF 0829993 Reliable Spatial-Temporal Coverage with Minimum Cost in Wireless Sensor Network Deployments
CNS 1018320 Undersea Sensor Networks for Intrusion Detection Foundations and Practice
CNS 0831579 Throughput Optimization in Wireless Mesh Sensor Networks
NSF SupportAbove work was supported under the following grants
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
ldquoThe small world networkis a type of mathematical graph in which most nodes are not neighbors of one another but most nodes can be reached from every other by a small number of hops or stepsrdquo
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Social Network A New Frontier
Most of social networks are small world networks with large size
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
The experimentbull Random people from Nebraska
were to send a letter (via intermediaries) to a stock broker in Boston
bull Could only send to someone with whom they know
Among the letters that found the target the average number
of steps was six
Milgram (1967)
Stanley Milgram (1933-1984)
Itrsquos a small world after all
Six Steps of Separation
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Family Friend
Family
Friend
Interviewer
Friend
Supervisor
Friend
Roommate
Friend
Six Steps of Separation
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Social Networks in Life
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Increasing Popularity
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
The trend effect that Kate Duchess of Cambridge has on others from cosmetic surgery for brides to sales of coral-colored jeansrdquo
ldquoKate Middleton Effect
Usage Example 1
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
According to Newsweek The Kate Effect may be worth pound1 billion to the UK fashion industry
Tony DiMasso L K Bennettrsquos US president stated in 2012 when she does wear something it always seems to go on a waiting list
Hike in Sales of Special Products
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
bull Influential Person
bull Kate is one of the persons that have many friends in this social network
How to Find Kate
For more kates itrsquos not as easy as you might think
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Challenge an overall consideration of influence
bullFor example Positive Influence Influence Maximization
Influence Minimization
Find More Kate
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
bullGiven k
bullFind k seeds (Kates) to maximize the number of influenced persons
Influence Maximization
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Influence Maximization
51
Influence Maximization
of influenced nodes is 6
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Influence Maximization
52
of influenced nodes is 6 of influenced nodes is 16
Influence Maximization
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
ldquoBetter Approximations for Influence Maximization in Online Social Networksrdquo Journal of Combinatorial Optimization 2013 (with Weili Wu et al)
Ongoing ResearchInitial Result
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Search Cheap Ticket
Usage Example 2
There are about 28537 commercial flights in the sky in the US on any given day
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
bull It is a shortest path problem in a big data network
How to find cheap ticket
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Cheap ticket-Graph
AA123
AA456
AA789Dallas Chicago
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Cheap ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
Each city has a set of startpoints and a set of endpointsThey are connected into a bipartite graph based on certain rules
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Cheap Ticket-GraphDallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
transition
thedo cost to a ngrepresenti weight a have also
may city each at graph bipartite theof edgeEach
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Cheap Ticket-Graph
Dallas
8am
9am
1pm 1pm
9am
3pm 3pm
8am
pathshortest node-to-Node
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
If searching area is larger then searching needs
more time but ticket price may be cheaper
Hard to do it in real-time
Better software is needed
Time VS Price
Challenge
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Ongoing ResearchInitial Result
ldquoSocial Network Path Analysis Based on HBaserdquo CSoNet 2013 (with Weili Wu et al)
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Outline
I Data Collection in Sensor System
II Data Analysis On Social Networks Kate Middleton Effect Search cheap ticket
III Final Remarks
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
SSS (Sensor and Sensing Systems) sensor networks with application in industrial engineering
Big Data Program Critical Techniques and Technologies for Advancing Big Data Science amp Engineering
NSF Grant PossibilitiesIn SSS Program amp Big Data Program
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Research Experiences for Undergraduates (REU) program supports active research participation by undergraduate students in any area funded by NSF
REU 0851848 Verification and Validation for Software Safety (co-PI Weili Wu)
NSF Grant PossibilitiesIn REU Program
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
THANK YOU
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
in nodes fromn informatio of
number half aleast at receives nodesother ofeach
such that (Kates) nodes seed of subset aidentify
want to wenetwokion communicat ain Suppose
D
D
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Theorem
problemset dominating
influence positive for theion approximat-)1(
time-polynomial a exists therenetworks socialIn
O
Mathematical Model (II)
Given a graph find a positive influence dominating set with minimum cardinality
Positive Influence Dominating Set Problem
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
Mathematical Model (I)
in nodes 2)deg(least at by
dominated is nodeevery ifset dominating
influence positive a called is eFurthermor
oadjacent tor in either is
nodeevery ifset dominating a is subset nodeA
)(graph aConsider
Dv
DVu
D
DD
D
EVG
Dominating set
Positive influence dominating set
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-
- Slide 1
- Slide 2
- Slide 3
- Slide 4
- Slide 5
- Slide 6
- Slide 7
- Slide 8
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Slide 14
- Slide 15
- Slide 16
- Slide 17
- Slide 18
- Slide 19
- Slide 20
- Slide 21
- Slide 22
- Slide 23
- Slide 24
- Slide 25
- Slide 26
- Slide 27
- Slide 28
- Slide 29
- Slide 30
- Slide 31
- Slide 32
- Slide 33
- Slide 34
- Slide 35
- Slide 36
- Slide 39
- Slide 40
- Slide 41
- Slide 42
- Slide 43
- Slide 44
- Slide 45
- Slide 46
- Slide 47
- Slide 48
- Slide 49
- Slide 50
- Slide 51
- Slide 52
- Slide 54
- Slide 55
- Slide 56
- Slide 57
- Slide 58
- Slide 59
- Slide 60
- Slide 62
- Slide 65
- Slide 66
- Slide 68
- Slide 69
-