1. A Hybrid Routing Approach for Opportunistic Networks Ling-Jyh Chen 1, Chen-Hung Yu 2, Tony Sun 3, Yung-Chih Chen 1, and Hao-hua Chu 2 1 Academia Sinica 2 National Taiwan University 3 University of California at Los Angeles

2. Opportunistic Networks A type of DTN  Intermittent Links  Very Large Delays  High Link Error Rates Network contacts appear arbitrarily w/o prior information; neither scheduled optimal routing nor mobile ferry based approaches can be applied. Delay Tolerant Networks (DTN) Opportunistic Networks Scheduled Networks Predictable Networks

3. Opportunistic Networks Potential Applications  interconnect mobile search and rescue nodes in disaster areas  allow message exchanges in developing areas  permit scientific monitoring of wilderness Examples  ZebraNet, DieselNet, CenWits, UWSN, …

4. Our Contribution We proposed an effective data forwarding scheme for opportunistic networks, called H-EC. H-EC combines the strength of erasure coding and the advantages of aggressive forwarding. aggressively robust We showed that H-EC performs aggressively for very small delay performance cases and remain robust for worst-case delay performance cases.

5. Outline Background and Overview  Related Work  Erasure coding H-EC details Evaluation Summary and Future Work

6. Related Work Data forwarding schemes for opportunistic networks  Flooding based Epidemic routing (Vahdat ‘00) Controlled flooding (Harras ‘05)  Coding based Erasure coding based data forwarding (Wang ‘05, Liao ‘06) Network coding based data forwarding (Widmer ‘05) Prediction based schemes  Probabilistic routing (Lindgren ‘04)  Mobility pattern based forwarding (Leguay ‘05)

7. Erasure Coding Provide better fault-tolerance by adding redundancy without the overhead of strict replication (e.g., Reed-Solomon, Gallager, Tornado, and IRA codes) Applications: P2P, overlay routing, WSN, data storage, etc. Our work is based on the generic erasure coding concept.

8. Erasure Coding ABCD A-1 A-2 A-3 A-4 B-1 B-2 B-3C-1 C-4 D-1 ABC DA-1 A-2 A-3 A-4 B-1 B-2 B-3 B-4 C-1 C-2 C-3 C-4 D-1 D-2 D-3 D-4 Lossy Channel (r,n)=(2,4)

9. Previous Approaches direct contact (DC) simple replication (srep) (k=2) EC: erasure coding based data forwarding (r=2, n=4)

10. Proposed Approach (I) A-EC A-EC: erasure coding + aggressive forwarding (r=2, n=4) Issues: Black-holes  unreliable (limited battery power and/or buffer size)  hardly moving closer towards the destination

11. Proposed Approach (II) H-EC H-EC: Hybrid of EC and A-EC  First copy is sent using EC  Second copy is sent using A-EC during the residual contact duration after sending the first EC block *Algorithms for scheduling the 2 nd copy of EC blocks: SF, FI, and BI

12. Simulation DTNSIM: a java based simulator Implementation  EC: erasure coding based data forwarding  R-EC: EC + simple replication  A-EC: erasure coding + aggressive forwarding  H-EC: EC + A-EC Scenarios:  34 nodes (including source and destination)  two-hop scenario  1200 bytes/msg, 12 msg/day, 160 days  Contact time & Inter-contact time: power-law w/ coefficient 0.6  EC: block size = 150 bytes, r = 2, n = 16  Black-holes: buffer size = 2 msg

13. Results I: General scenario

14. Results II: Black-hole scenario

15. Overhead Analysis r: replication factor of erasure coding n: the number of relays among which erasure code blocks are split k: the replication factor of srep algorithm Strategies to reduce overhead  Explicit ACK (or passive cure)  Adaptive Coding  Probabilistic Forwarding

16. Summary An effective data forwarding scheme is essential for opportunistic networks. H-EC combines the strength of erasure coding and the advantages of aggressive forwarding. H-EC is effective and robust, even when black-holes are present.

17. Future Work H-EC Evaluation  Using realistic mobility traces*  Analytical model H-EC Adaptation  Redundancy level  Probabilistic forwarding H-EC Applications  Scalable data transfer*  Testbed

18. Thank You! For more information… http://nrl.iis.sinica.edu.tw/ http://nrl.iis.sinica.edu.tw/DTN/