SlideShare uma empresa Scribd logo
1 de 28
User-Centric Data Dissemination in Disruption Tolerant Networks Wei Gao and Guohong Cao INFOCOM 2011 05/26/2011 MDC Lab Meeting Yao-Jen Tang
Outline Introduction Problem and Models Approach Analysis Simulation Conclusion Outline MDC Lab Meeting 05/26/2011
05/26/2011 MDC Lab Meeting Introduction Introduction
Disruption Tolerant Network (DTN) Example 1: Connected Network 1 MDC Lab Meeting 05/26/2011
Disruption Tolerant Network (DTN) Example 2: Disruption Tolerant Network 2 MDC Lab Meeting 05/26/2011
Flooding Example of User-Centric Data Dissemination in DTN 05/26/2011 MDC Lab Meeting 3 1 9 3 6 4 2 7 5 10 8 11 2 3 5 6 1 8 4 9 10 11 12 12
05/26/2011 MDC Lab Meeting Problem  and Models Problem and Models
User-Centric Data Dissemination: Uncontrollable and Controllable Parts 05/26/2011 MDC Lab Meeting 4 Controllable 1 3 4 2 4 3 4 2 1 2 3 4
Maximize Cost-Effectiveness User-Centric Data Dissemination in DTN 05/26/2011 MDC Lab Meeting 5 1 9 3 6 4 2 7 5 10 8 11 2 3 5 6 1 8 4 9 10 11 12 12 =01+3=0   =04+2=0   =16+1=0.143   =17+2=0.111   =16=0.167   =01=0   =1+19+1=0.2   Cost-Effectiveness
Maximize Expected Cost-Effectiveness User-Centric Data Dissemination in DTN 05/26/2011 User Paper Interested in Paper = 0.3*0 + 0.2*0.5 + 0.3*0 + 0.1*0 + 0.1*0.5 = 0.15 MDC Lab Meeting 6 0.35 0.32 0.4 0.34 0.8 0.3 0.48 0.3 0.75 0.8 1 9 3 6 4 2 7 5 10 8 11 12 0.4 0.8 0.8 0.15 0.28 0.4 0.6 0.15 0.38 0.3 1.02 0.3*0.4+0.3*0.4+0.35*0.4 = 0.4 0.8 0.8 0.4 0.8 0.35 0.15 0.15 0.27 0.35 0.63 0.25 0.66 0.25 0.09 0.6 0.6 0.6
05/26/2011 MDC Lab Meeting Approach Approach
Relay Selection with Centrality 05/26/2011 MDC Lab Meeting 7 Expected Cost-Effectiveness 0.35 0.24 0.35 0.32 =1.4+0.752+1 =0.717   =0.381=0.38   =0.38+1.021+1 =0.7   0.4 0.06 0.4 0.34 0.8 0.3 0.48 0.8 0.3 0.54 0.3 0.78 0.3 0.75 0.717 0.8 1 9 3 6 4 2 7 5 10 8 11 2 3 12 12 0.4 0.8 0.8 0.15 0.28 0.4 0.6 0.15 0.24 0.15 0.38 0.3 1.02 0.3 0.74 0.717 0.38 0.7 0.7 0.4 0.8 0.8 0.4 0.8 0.35 0.15 0.15 0.27 0.15 0.41 0.35 0.63 0.25 0.39 0.25 0.66 0.25 0.09 0.6 0.6 0.6 0.4 0.35 0.06 =13=0.33   Cost-Effectiveness
Relay Selection with Multi-Hop Centrality 05/26/2011 MDC Lab Meeting 8 Expected Cost-Effectiveness 0.35 0.456 0.35 0.344 =0.4681=0.468   =0.468+1.10751+1 =0.788   =1.5755+0.8432+1 =0.806   =2.4185+0.833+1 =0.812   0.4 0.084 0.4 0.364 0.8 0.3 0.696 0.3 0.584 0.3 0.696 0.8 0.3 0.62 0.3 0.9 0.3 0.83 0.8 0.802 0.38+0.4×0.82×0.3+0.4×0.82×0.25   1 9 3 6 4 2 7 5 10 8 11 2 3 5 4 12 12 0.4 0.8 0.15 0.396 0.8 0.15 0.396 0.15 0.452 0.4 0.6 0.15 0.384 0.15 0.328 0.15 0.468 0.3 0.9395 0.3 1.1075 0.3 0.8275 0.468 0.802 0.788 0.788 0.806 0.4 0.8 0.8 0.4 0.8 0.35 0.122 0.35 0.285 0.15 0.334 0.15 0.474 0.35 0.753 0.25 0.126 0.25 0.843 0.25 0.633 0.25 0.168 0.6 0.6 0.6 0.4 0.35 0.114 0.806 =14=0.25   =24=0.5   Cost-Effectiveness
05/26/2011 MDC Lab Meeting Analysis Analysis
Lower Bound on Expected Cost-Effectiveness at t Expected Cost-Effectiveness ≥   05/26/2011 MDC Lab Meeting 9 #𝑅𝑒𝑙𝑎𝑦×(0.15×0.6)#𝑅𝑒𝑙𝑎𝑦   0.3 0.48 0.3 0.75 0.8 1 3 6 4 2 7 5 0.8 0.8 0.4 0.6 0.3 1.02 0.15 0.38 0.4 0.8 0.4 0.8 0.35 0.15 0.35 0.63 0.25 0.66 0.6 0.6
Lower Bound on Probability of Increasing Cost-Effectiveness within t Cost-Effectiveness= 03=0   05/26/2011 MDC Lab Meeting 10 ≥𝐏𝐫𝑡≥𝑇𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡𝑒𝑟 ×𝐏𝐫𝑡<𝑇𝑁𝑒𝑥𝑡𝑅𝑒𝑙𝑎𝑦   0.35 0.24 = 13=0.33   0.3 0.48 0.8 1 3 6 4 2 7 5 12 12 0.3 0.75 0.3 0.78 0.8 0.8 0.8 0.4 0.6 0.3 1.02 0.15 0.38 0.4 0.8 0.4 0.8 0.35 0.15 0.35 0.63 0.25 0.66 0.6 0.6
Upper Bound on Maintaining Overhead with r-Hop Range 05/26/2011 MDC Lab Meeting 11 2 r 1 𝑘+𝑘𝑘−1+…+𝑘𝑘−1𝑟−1 =O(𝑐𝑟)   11 10 12 9 3 13 2 8 0 O(𝑟𝑐𝑟)   14 4 5 1 15 16 6 7
The Most Valuable Lemma Assumption: Each node maintains the entire network information. For any relay s with locally expected cost-effectiveness, when it contacts node i at time t: If i’s centrality < expected cost-effectiveness: selecting any i’s neighbor j as relay will decrease expected cost-effectiveness. Ifi’s centrality >= expected cost-effectiveness: there exists one i’s neighbor j, such that selecting j as relay will increase expected cost-effectiveness. 12 MDC Lab Meeting 05/26/2011
The Most Valuable Lemma 05/26/2011 MDC Lab Meeting 13 Expected Cost-Effectiveness =3.51=3.5   =3.5+3.2+3.33=3.33<3.5   1 3 0.6 1 0.9 1 4 5 2 3 5 3 1 4 2 1 3.2 1 3.3 1 3.5 1 1 0.6 1 0.9 Expected Cost-Effectiveness 1 3 =3+3.3+33=3.1>3   =31=3  
The Most Valuable Lemma 05/26/2011 MDC Lab Meeting 13 Expected Cost-Effectiveness =3.31=3.3   =3.3+3.2+3.53=3.33>3.3   1 3 1 3 Goal (Idea) What’s your  approach? 0.6 1 0.9 1 4 5 2 3 1 3 5 3 4 1 1 3.2 1 3.3 1 3.5 1 1 0.6 1 0.9 Expected Cost-Effectiveness 1 3 =3.2+3.3+33=3.17<3.2   =3.21=3.2  
05/26/2011 MDC Lab Meeting Simulation Simulation
Performance Evaluation Realistic DTN traces: MIT Reality and Infocom06 Schemes for comparison: Flooding Random Flooding ContentPlace SocialCast 14 MDC Lab Meeting 05/26/2011
Data Dissemination with Different Time Constraints 05/26/2011 MDC Lab Meeting 15
Data Dissemination with Different Buffer Constraints 05/26/2011 MDC Lab Meeting 16
Data Dissemination with Different Scope of Maintaining Network Information 05/26/2011 MDC Lab Meeting 17
05/26/2011 MDC Lab Meeting Conclusion Conclusion
Conclusion Solve the user-centric data dissemination problem in DTN using social contact pattern and greedily expected cost-effectiveness approach 18 MDC Lab Meeting 05/26/2011 Thanks for Your Attention! The slides are made and presented by Yao-Jen Tang (yjtang@cs.nthu.edu.tw)
My Next Presentation Topic List Data Dissemination:R. Masiero and G. Neglia, “Distributed Subgradient Methods for Delay Tolerant Networks”, INFOCOM, 2011. Data Caching:W. Gaoand G. Cao, “Supporting Cooperative Caching in Disruption Tolerant Networks”, ICDCS, 2011. Power Control:E. Altman et al., “Risk Sensitive Optimal Control Framework Applied to Delay Tolerant Networks”, INFOCOM, 2011.  Appendix MDC Lab Meeting 05/26/2011

Mais conteúdo relacionado

Mais procurados

copy for Gary Chin.
copy for Gary Chin.copy for Gary Chin.
copy for Gary Chin.
Teng Xiaolu
 

Mais procurados (12)

Feature Reduction Techniques
Feature Reduction TechniquesFeature Reduction Techniques
Feature Reduction Techniques
 
Decision trees
Decision treesDecision trees
Decision trees
 
Dimensionality reduction
Dimensionality reductionDimensionality reduction
Dimensionality reduction
 
Introduction to Modeling
Introduction to ModelingIntroduction to Modeling
Introduction to Modeling
 
61_Empirical
61_Empirical61_Empirical
61_Empirical
 
copy for Gary Chin.
copy for Gary Chin.copy for Gary Chin.
copy for Gary Chin.
 
2-IJCSE-00536
2-IJCSE-005362-IJCSE-00536
2-IJCSE-00536
 
Estimating project development effort using clustered regression approach
Estimating project development effort using clustered regression approachEstimating project development effort using clustered regression approach
Estimating project development effort using clustered regression approach
 
ESTIMATING PROJECT DEVELOPMENT EFFORT USING CLUSTERED REGRESSION APPROACH
ESTIMATING PROJECT DEVELOPMENT EFFORT USING CLUSTERED REGRESSION APPROACHESTIMATING PROJECT DEVELOPMENT EFFORT USING CLUSTERED REGRESSION APPROACH
ESTIMATING PROJECT DEVELOPMENT EFFORT USING CLUSTERED REGRESSION APPROACH
 
Machine learning the next revolution or just another hype
Machine learning   the next revolution or just another hypeMachine learning   the next revolution or just another hype
Machine learning the next revolution or just another hype
 
T 8-gurjinder
T 8-gurjinderT 8-gurjinder
T 8-gurjinder
 
Social Learning in Networks: Extraction Deterministic Rules
Social Learning in Networks: Extraction Deterministic RulesSocial Learning in Networks: Extraction Deterministic Rules
Social Learning in Networks: Extraction Deterministic Rules
 

Destaque (9)

Flu Vaccines...What Good Are They?
Flu Vaccines...What Good Are They?Flu Vaccines...What Good Are They?
Flu Vaccines...What Good Are They?
 
New Color Web Package Show
New Color Web Package ShowNew Color Web Package Show
New Color Web Package Show
 
Social Media...preparing for the new revolution
Social Media...preparing for the new revolutionSocial Media...preparing for the new revolution
Social Media...preparing for the new revolution
 
The total package drew davidsen
The total package drew davidsenThe total package drew davidsen
The total package drew davidsen
 
Net Neutrality and The FCC
Net Neutrality and The FCCNet Neutrality and The FCC
Net Neutrality and The FCC
 
Preston Williams Associations
Preston Williams AssociationsPreston Williams Associations
Preston Williams Associations
 
ODOT's Social Media Presence
ODOT's Social Media PresenceODOT's Social Media Presence
ODOT's Social Media Presence
 
Twitter
TwitterTwitter
Twitter
 
julia.lin_plastic arts
julia.lin_plastic artsjulia.lin_plastic arts
julia.lin_plastic arts
 

Semelhante a User centric data dissemination in disruption tolerant networkas

Semelhante a User centric data dissemination in disruption tolerant networkas (20)

Comments on Simulations Project Parts I & II Marking Contingencies.pdf
Comments on Simulations Project Parts I & II Marking Contingencies.pdfComments on Simulations Project Parts I & II Marking Contingencies.pdf
Comments on Simulations Project Parts I & II Marking Contingencies.pdf
 
An Elitist Simulated Annealing Algorithm for Solving Multi Objective Optimiza...
An Elitist Simulated Annealing Algorithm for Solving Multi Objective Optimiza...An Elitist Simulated Annealing Algorithm for Solving Multi Objective Optimiza...
An Elitist Simulated Annealing Algorithm for Solving Multi Objective Optimiza...
 
Dimension reduction(jiten01)
Dimension reduction(jiten01)Dimension reduction(jiten01)
Dimension reduction(jiten01)
 
Comments on Simulations Project Parts I through III.pdf
Comments on Simulations Project Parts I through III.pdfComments on Simulations Project Parts I through III.pdf
Comments on Simulations Project Parts I through III.pdf
 
Overfitting.pptx
Overfitting.pptxOverfitting.pptx
Overfitting.pptx
 
Comments on Simulations Project.pdf
Comments on Simulations Project.pdfComments on Simulations Project.pdf
Comments on Simulations Project.pdf
 
IRJET- Supervised Learning Classification Algorithms Comparison
IRJET- Supervised Learning Classification Algorithms ComparisonIRJET- Supervised Learning Classification Algorithms Comparison
IRJET- Supervised Learning Classification Algorithms Comparison
 
IRJET- Supervised Learning Classification Algorithms Comparison
IRJET- Supervised Learning Classification Algorithms ComparisonIRJET- Supervised Learning Classification Algorithms Comparison
IRJET- Supervised Learning Classification Algorithms Comparison
 
Comments on Simulations Analytics.pdf
Comments on Simulations Analytics.pdfComments on Simulations Analytics.pdf
Comments on Simulations Analytics.pdf
 
Big-data analytics: challenges and opportunities
Big-data analytics: challenges and opportunitiesBig-data analytics: challenges and opportunities
Big-data analytics: challenges and opportunities
 
Caravan insurance data mining prediction models
Caravan insurance data mining prediction modelsCaravan insurance data mining prediction models
Caravan insurance data mining prediction models
 
Caravan insurance data mining prediction models
Caravan insurance data mining prediction modelsCaravan insurance data mining prediction models
Caravan insurance data mining prediction models
 
Customer Segmentation with R - Deep Dive into flexclust
Customer Segmentation with R - Deep Dive into flexclustCustomer Segmentation with R - Deep Dive into flexclust
Customer Segmentation with R - Deep Dive into flexclust
 
600.412.Lecture06
600.412.Lecture06600.412.Lecture06
600.412.Lecture06
 
Simulations Project Marking Contingencies.pdf
Simulations Project Marking Contingencies.pdfSimulations Project Marking Contingencies.pdf
Simulations Project Marking Contingencies.pdf
 
Managing the Evolution of Information Systems with Intensional Views and Rela...
Managing the Evolution of Information Systems with Intensional Views and Rela...Managing the Evolution of Information Systems with Intensional Views and Rela...
Managing the Evolution of Information Systems with Intensional Views and Rela...
 
IRJET- A Detailed Study on Classification Techniques for Data Mining
IRJET- A Detailed Study on Classification Techniques for Data MiningIRJET- A Detailed Study on Classification Techniques for Data Mining
IRJET- A Detailed Study on Classification Techniques for Data Mining
 
introduction-to-decision-trees.pdf
introduction-to-decision-trees.pdfintroduction-to-decision-trees.pdf
introduction-to-decision-trees.pdf
 
Twdatasci cjlin-big data analytics - challenges and opportunities
Twdatasci cjlin-big data analytics - challenges and opportunitiesTwdatasci cjlin-big data analytics - challenges and opportunities
Twdatasci cjlin-big data analytics - challenges and opportunities
 
Process coordinator in NUMA environment
Process coordinator in NUMA environmentProcess coordinator in NUMA environment
Process coordinator in NUMA environment
 

Último

Último (20)

Secure Zero Touch enabled Edge compute with Dell NativeEdge via FDO _ Brad at...
Secure Zero Touch enabled Edge compute with Dell NativeEdge via FDO _ Brad at...Secure Zero Touch enabled Edge compute with Dell NativeEdge via FDO _ Brad at...
Secure Zero Touch enabled Edge compute with Dell NativeEdge via FDO _ Brad at...
 
Speed Wins: From Kafka to APIs in Minutes
Speed Wins: From Kafka to APIs in MinutesSpeed Wins: From Kafka to APIs in Minutes
Speed Wins: From Kafka to APIs in Minutes
 
How we scaled to 80K users by doing nothing!.pdf
How we scaled to 80K users by doing nothing!.pdfHow we scaled to 80K users by doing nothing!.pdf
How we scaled to 80K users by doing nothing!.pdf
 
THE BEST IPTV in GERMANY for 2024: IPTVreel
THE BEST IPTV in  GERMANY for 2024: IPTVreelTHE BEST IPTV in  GERMANY for 2024: IPTVreel
THE BEST IPTV in GERMANY for 2024: IPTVreel
 
PLAI - Acceleration Program for Generative A.I. Startups
PLAI - Acceleration Program for Generative A.I. StartupsPLAI - Acceleration Program for Generative A.I. Startups
PLAI - Acceleration Program for Generative A.I. Startups
 
Measures in SQL (a talk at SF Distributed Systems meetup, 2024-05-22)
Measures in SQL (a talk at SF Distributed Systems meetup, 2024-05-22)Measures in SQL (a talk at SF Distributed Systems meetup, 2024-05-22)
Measures in SQL (a talk at SF Distributed Systems meetup, 2024-05-22)
 
Powerful Start- the Key to Project Success, Barbara Laskowska
Powerful Start- the Key to Project Success, Barbara LaskowskaPowerful Start- the Key to Project Success, Barbara Laskowska
Powerful Start- the Key to Project Success, Barbara Laskowska
 
Custom Approval Process: A New Perspective, Pavel Hrbacek & Anindya Halder
Custom Approval Process: A New Perspective, Pavel Hrbacek & Anindya HalderCustom Approval Process: A New Perspective, Pavel Hrbacek & Anindya Halder
Custom Approval Process: A New Perspective, Pavel Hrbacek & Anindya Halder
 
WebAssembly is Key to Better LLM Performance
WebAssembly is Key to Better LLM PerformanceWebAssembly is Key to Better LLM Performance
WebAssembly is Key to Better LLM Performance
 
Linux Foundation Edge _ Overview of FDO Software Components _ Randy at Intel.pdf
Linux Foundation Edge _ Overview of FDO Software Components _ Randy at Intel.pdfLinux Foundation Edge _ Overview of FDO Software Components _ Randy at Intel.pdf
Linux Foundation Edge _ Overview of FDO Software Components _ Randy at Intel.pdf
 
The Metaverse: Are We There Yet?
The  Metaverse:    Are   We  There  Yet?The  Metaverse:    Are   We  There  Yet?
The Metaverse: Are We There Yet?
 
Behind the Scenes From the Manager's Chair: Decoding the Secrets of Successfu...
Behind the Scenes From the Manager's Chair: Decoding the Secrets of Successfu...Behind the Scenes From the Manager's Chair: Decoding the Secrets of Successfu...
Behind the Scenes From the Manager's Chair: Decoding the Secrets of Successfu...
 
SOQL 201 for Admins & Developers: Slice & Dice Your Org’s Data With Aggregate...
SOQL 201 for Admins & Developers: Slice & Dice Your Org’s Data With Aggregate...SOQL 201 for Admins & Developers: Slice & Dice Your Org’s Data With Aggregate...
SOQL 201 for Admins & Developers: Slice & Dice Your Org’s Data With Aggregate...
 
The UX of Automation by AJ King, Senior UX Researcher, Ocado
The UX of Automation by AJ King, Senior UX Researcher, OcadoThe UX of Automation by AJ King, Senior UX Researcher, Ocado
The UX of Automation by AJ King, Senior UX Researcher, Ocado
 
Extensible Python: Robustness through Addition - PyCon 2024
Extensible Python: Robustness through Addition - PyCon 2024Extensible Python: Robustness through Addition - PyCon 2024
Extensible Python: Robustness through Addition - PyCon 2024
 
Connecting the Dots in Product Design at KAYAK
Connecting the Dots in Product Design at KAYAKConnecting the Dots in Product Design at KAYAK
Connecting the Dots in Product Design at KAYAK
 
Strategic AI Integration in Engineering Teams
Strategic AI Integration in Engineering TeamsStrategic AI Integration in Engineering Teams
Strategic AI Integration in Engineering Teams
 
Choosing the Right FDO Deployment Model for Your Application _ Geoffrey at In...
Choosing the Right FDO Deployment Model for Your Application _ Geoffrey at In...Choosing the Right FDO Deployment Model for Your Application _ Geoffrey at In...
Choosing the Right FDO Deployment Model for Your Application _ Geoffrey at In...
 
ECS 2024 Teams Premium - Pretty Secure
ECS 2024   Teams Premium - Pretty SecureECS 2024   Teams Premium - Pretty Secure
ECS 2024 Teams Premium - Pretty Secure
 
Syngulon - Selection technology May 2024.pdf
Syngulon - Selection technology May 2024.pdfSyngulon - Selection technology May 2024.pdf
Syngulon - Selection technology May 2024.pdf
 

User centric data dissemination in disruption tolerant networkas

  • 1. User-Centric Data Dissemination in Disruption Tolerant Networks Wei Gao and Guohong Cao INFOCOM 2011 05/26/2011 MDC Lab Meeting Yao-Jen Tang
  • 2. Outline Introduction Problem and Models Approach Analysis Simulation Conclusion Outline MDC Lab Meeting 05/26/2011
  • 3. 05/26/2011 MDC Lab Meeting Introduction Introduction
  • 4. Disruption Tolerant Network (DTN) Example 1: Connected Network 1 MDC Lab Meeting 05/26/2011
  • 5. Disruption Tolerant Network (DTN) Example 2: Disruption Tolerant Network 2 MDC Lab Meeting 05/26/2011
  • 6. Flooding Example of User-Centric Data Dissemination in DTN 05/26/2011 MDC Lab Meeting 3 1 9 3 6 4 2 7 5 10 8 11 2 3 5 6 1 8 4 9 10 11 12 12
  • 7. 05/26/2011 MDC Lab Meeting Problem and Models Problem and Models
  • 8. User-Centric Data Dissemination: Uncontrollable and Controllable Parts 05/26/2011 MDC Lab Meeting 4 Controllable 1 3 4 2 4 3 4 2 1 2 3 4
  • 9. Maximize Cost-Effectiveness User-Centric Data Dissemination in DTN 05/26/2011 MDC Lab Meeting 5 1 9 3 6 4 2 7 5 10 8 11 2 3 5 6 1 8 4 9 10 11 12 12 =01+3=0   =04+2=0   =16+1=0.143   =17+2=0.111   =16=0.167   =01=0   =1+19+1=0.2   Cost-Effectiveness
  • 10. Maximize Expected Cost-Effectiveness User-Centric Data Dissemination in DTN 05/26/2011 User Paper Interested in Paper = 0.3*0 + 0.2*0.5 + 0.3*0 + 0.1*0 + 0.1*0.5 = 0.15 MDC Lab Meeting 6 0.35 0.32 0.4 0.34 0.8 0.3 0.48 0.3 0.75 0.8 1 9 3 6 4 2 7 5 10 8 11 12 0.4 0.8 0.8 0.15 0.28 0.4 0.6 0.15 0.38 0.3 1.02 0.3*0.4+0.3*0.4+0.35*0.4 = 0.4 0.8 0.8 0.4 0.8 0.35 0.15 0.15 0.27 0.35 0.63 0.25 0.66 0.25 0.09 0.6 0.6 0.6
  • 11. 05/26/2011 MDC Lab Meeting Approach Approach
  • 12. Relay Selection with Centrality 05/26/2011 MDC Lab Meeting 7 Expected Cost-Effectiveness 0.35 0.24 0.35 0.32 =1.4+0.752+1 =0.717   =0.381=0.38   =0.38+1.021+1 =0.7   0.4 0.06 0.4 0.34 0.8 0.3 0.48 0.8 0.3 0.54 0.3 0.78 0.3 0.75 0.717 0.8 1 9 3 6 4 2 7 5 10 8 11 2 3 12 12 0.4 0.8 0.8 0.15 0.28 0.4 0.6 0.15 0.24 0.15 0.38 0.3 1.02 0.3 0.74 0.717 0.38 0.7 0.7 0.4 0.8 0.8 0.4 0.8 0.35 0.15 0.15 0.27 0.15 0.41 0.35 0.63 0.25 0.39 0.25 0.66 0.25 0.09 0.6 0.6 0.6 0.4 0.35 0.06 =13=0.33   Cost-Effectiveness
  • 13. Relay Selection with Multi-Hop Centrality 05/26/2011 MDC Lab Meeting 8 Expected Cost-Effectiveness 0.35 0.456 0.35 0.344 =0.4681=0.468   =0.468+1.10751+1 =0.788   =1.5755+0.8432+1 =0.806   =2.4185+0.833+1 =0.812   0.4 0.084 0.4 0.364 0.8 0.3 0.696 0.3 0.584 0.3 0.696 0.8 0.3 0.62 0.3 0.9 0.3 0.83 0.8 0.802 0.38+0.4×0.82×0.3+0.4×0.82×0.25   1 9 3 6 4 2 7 5 10 8 11 2 3 5 4 12 12 0.4 0.8 0.15 0.396 0.8 0.15 0.396 0.15 0.452 0.4 0.6 0.15 0.384 0.15 0.328 0.15 0.468 0.3 0.9395 0.3 1.1075 0.3 0.8275 0.468 0.802 0.788 0.788 0.806 0.4 0.8 0.8 0.4 0.8 0.35 0.122 0.35 0.285 0.15 0.334 0.15 0.474 0.35 0.753 0.25 0.126 0.25 0.843 0.25 0.633 0.25 0.168 0.6 0.6 0.6 0.4 0.35 0.114 0.806 =14=0.25   =24=0.5   Cost-Effectiveness
  • 14. 05/26/2011 MDC Lab Meeting Analysis Analysis
  • 15. Lower Bound on Expected Cost-Effectiveness at t Expected Cost-Effectiveness ≥   05/26/2011 MDC Lab Meeting 9 #𝑅𝑒𝑙𝑎𝑦×(0.15×0.6)#𝑅𝑒𝑙𝑎𝑦   0.3 0.48 0.3 0.75 0.8 1 3 6 4 2 7 5 0.8 0.8 0.4 0.6 0.3 1.02 0.15 0.38 0.4 0.8 0.4 0.8 0.35 0.15 0.35 0.63 0.25 0.66 0.6 0.6
  • 16. Lower Bound on Probability of Increasing Cost-Effectiveness within t Cost-Effectiveness= 03=0   05/26/2011 MDC Lab Meeting 10 ≥𝐏𝐫𝑡≥𝑇𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡𝑒𝑟 ×𝐏𝐫𝑡<𝑇𝑁𝑒𝑥𝑡𝑅𝑒𝑙𝑎𝑦   0.35 0.24 = 13=0.33   0.3 0.48 0.8 1 3 6 4 2 7 5 12 12 0.3 0.75 0.3 0.78 0.8 0.8 0.8 0.4 0.6 0.3 1.02 0.15 0.38 0.4 0.8 0.4 0.8 0.35 0.15 0.35 0.63 0.25 0.66 0.6 0.6
  • 17. Upper Bound on Maintaining Overhead with r-Hop Range 05/26/2011 MDC Lab Meeting 11 2 r 1 𝑘+𝑘𝑘−1+…+𝑘𝑘−1𝑟−1 =O(𝑐𝑟)   11 10 12 9 3 13 2 8 0 O(𝑟𝑐𝑟)   14 4 5 1 15 16 6 7
  • 18. The Most Valuable Lemma Assumption: Each node maintains the entire network information. For any relay s with locally expected cost-effectiveness, when it contacts node i at time t: If i’s centrality < expected cost-effectiveness: selecting any i’s neighbor j as relay will decrease expected cost-effectiveness. Ifi’s centrality >= expected cost-effectiveness: there exists one i’s neighbor j, such that selecting j as relay will increase expected cost-effectiveness. 12 MDC Lab Meeting 05/26/2011
  • 19. The Most Valuable Lemma 05/26/2011 MDC Lab Meeting 13 Expected Cost-Effectiveness =3.51=3.5   =3.5+3.2+3.33=3.33<3.5   1 3 0.6 1 0.9 1 4 5 2 3 5 3 1 4 2 1 3.2 1 3.3 1 3.5 1 1 0.6 1 0.9 Expected Cost-Effectiveness 1 3 =3+3.3+33=3.1>3   =31=3  
  • 20. The Most Valuable Lemma 05/26/2011 MDC Lab Meeting 13 Expected Cost-Effectiveness =3.31=3.3   =3.3+3.2+3.53=3.33>3.3   1 3 1 3 Goal (Idea) What’s your approach? 0.6 1 0.9 1 4 5 2 3 1 3 5 3 4 1 1 3.2 1 3.3 1 3.5 1 1 0.6 1 0.9 Expected Cost-Effectiveness 1 3 =3.2+3.3+33=3.17<3.2   =3.21=3.2  
  • 21. 05/26/2011 MDC Lab Meeting Simulation Simulation
  • 22. Performance Evaluation Realistic DTN traces: MIT Reality and Infocom06 Schemes for comparison: Flooding Random Flooding ContentPlace SocialCast 14 MDC Lab Meeting 05/26/2011
  • 23. Data Dissemination with Different Time Constraints 05/26/2011 MDC Lab Meeting 15
  • 24. Data Dissemination with Different Buffer Constraints 05/26/2011 MDC Lab Meeting 16
  • 25. Data Dissemination with Different Scope of Maintaining Network Information 05/26/2011 MDC Lab Meeting 17
  • 26. 05/26/2011 MDC Lab Meeting Conclusion Conclusion
  • 27. Conclusion Solve the user-centric data dissemination problem in DTN using social contact pattern and greedily expected cost-effectiveness approach 18 MDC Lab Meeting 05/26/2011 Thanks for Your Attention! The slides are made and presented by Yao-Jen Tang (yjtang@cs.nthu.edu.tw)
  • 28. My Next Presentation Topic List Data Dissemination:R. Masiero and G. Neglia, “Distributed Subgradient Methods for Delay Tolerant Networks”, INFOCOM, 2011. Data Caching:W. Gaoand G. Cao, “Supporting Cooperative Caching in Disruption Tolerant Networks”, ICDCS, 2011. Power Control:E. Altman et al., “Risk Sensitive Optimal Control Framework Applied to Delay Tolerant Networks”, INFOCOM, 2011. Appendix MDC Lab Meeting 05/26/2011