Opportunistic Routing Based on Daily Routines
0 likes441 views
The document discusses a proposal for opportunistic routing based on daily routines, termed 'dlife,' which leverages time-evolving social ties between nodes. It evaluates various routing methods, including dlife and dlifecomm, through simulations showing significant performance improvements over traditional methods, particularly in dynamic and heterogeneous scenarios. The conclusions emphasize the importance of understanding user behavior for effective routing decisions and outline future work in developing an information-centric model.
Opportunistic Routing Based on Daily Routines
- 1. Opportunistic Routing Based on Daily Routines Waldir Moreira, Paulo Mendes, and Susana Sargento waldir.junior@ulusofona.pt June 25th, 2012 6th IEEE WoWMoM Workshop on Autonomic and Opportunistic Communications (AOC 2012) San Francisco, USA
- 2. Agenda • Introduction • Motivation • Our Proposal: dLife • Evaluation • Results • Conclusions and Future Work 2
- 3. Introduction • Powerful devices • Spontaneous networks • Opportunistic contacts - Intermittent connectivity 3
- 4. Motivation • Many routing solutions - epidemic, encounter history, social aspects ... • Instability of the created proximity graphs • Dynamism of users’ behavior • Daily life routines 4
- 5. Our Proposal: dLife • To capture the dynamics of the network represented by time-evolving social ties between pair of nodes • Two utility functions - Time-Evolving Contact Duration (TECD) - TECD Importance (TECDi) 5
- 6. Our Proposal: dLife 6
- 7. Our Proposal: dLife (1) A w(B,x) B If Mx Buffer(B) and w(B,x) > w(A,x) (2) Mx A B Otherwise I(B) (3) A B If I(B) > I(A) Mx (4) A B 7
- 8. Our Proposal: dLifeComm (1) A w(B,x) B If Mx Buffer(B) and B.sameComm(x) and w(B,x) > w(A,x) (2) Mx A B Otherwise I(B) (3) A B If I(B) > I(A) Mx (4) A B 8
- 9. Evaluation Parameters Values Simulator Opportunistic Network Environment (ONE) Routing Proposals Bubble Rap, dLife and dLifeComm Scenarios Heterogeneous Mobility Trace Cambridge (CRAWDAD) Simulation Time 1036800 sec 1000000 sec # of Nodes 150 (people/vehicles) 36 (people) Mobility Models Working Day, Bus, Shortest Path Map Based Human Node Interface Wi-Fi (Rate: 11 Mbps / Range: 100 m) Bluetooth Node Buffer 2 MB Message TTL 1, 2, 4 days, 1 and 3 weeks Message Size 1 – 100 kB Generated Messages 6000 K-Clique, k 5 (Bubble Rap and dLifeComm) K-Clique, familiarThreshold 700 sec (Bubble Rap and dLifeComm) Daily Samples 24 (dLife and dLifeComm) 9
- 10. Results Heterogenous scenario Cambridge traces - dLife up to 39.5% - dLife up to 31.5% - dLifeComm up to 31.2% - dLifeComm up to 31.3% - Bubble Rap (Global centrality) - Network dynamics (daily routines) - Few nodes (~17%) high centrality - Local centrality 10
- 11. Results Heterogenous scenario Cambridge traces - dLife up to 78% less - dLife up to 55% less - dLifeComm up to 68% less - dLifeComm up to 50.5% less - High social strength/importance - Variable patterns of contacts - Bubble rap further replicates - Forwarders not often available 11
- 12. Results Heterogenous scenario Cambridge traces - dLife up to 48.3% less - dLife up to 83.7% less - dLifeComm up to 46.1% less - dLifeComm up to 84.7% less - Smarter forwarding decisions - Smaller, well connected groups - Bubble Rap (weak ties to destin.) - Bubble Rap (Centrality not real) 12
- 13. Conclusions and Future Work • Dynamism of users’ social daily behavior => wiser forwarding decisions • Centrality presented higher impact => does not capture reality • Next steps Internet-Draft Information-Centric DTNRG Meeting version of dLife Vancouver, July 2012 13
- 14. Acknowledgements To FCT for financial support via PhD grant (SFRH/BD/62761/2009) and UCR project (PTDC/EEA-TEL/103637/2008) 14
- 15. Opportunistic Routing Based on Daily Routines Waldir Moreira, Paulo Mendes, and Susana Sargento waldir.junior@ulusofona.pt June 25th, 2012 6th IEEE WoWMoM Workshop on Autonomic and Opportunistic Communications (AOC 2012) San Francisco, USA