SHORT: Self-Healing and Optimizing Routing Techniques for Mobile Ad Hoc Networks

1. SHORT: Self-Healing and Optimizing Routing Techniques for Mobile Ad Hoc Networks Chao Gui and Prasant Mohapatra Computer Science Department, University of California Davis ACM MobiHoc, vol. 2, 2003, pp. 279-290 Presenter: Sheng-Shih Wang October 30, 2003

2. Outline • Introduction • Path Aware (PA)-SHORT • Energy Aware (EA)-SHORT • Simulation • Conclusion Sheng-Shih Wang

3. Introduction • Routing Protocol Categories • Table-driven vs. on-demand • Proposed Routing Protocol • Self-Healing and Optimizing Routing Technique (SHORT) • Reduce energy consumption • Less overhead • Path Aware (PA)-SHORT • Reduce the number of hops • Energy Aware (EA)-SHORT • Conserve power Sheng-Shih Wang

4. Mobility (but retain connectivity) C H B I D F E A G J • J, E, and H is in the transmission ranges of A, J, and F, respectively • Desired path (after mobility) --- 5 hops H I F E A J Path Aware (PA)-SHORT --- Problem Description • Initial path (after route discovery process) --- 8 hops H C A E D G B F I Sheng-Shih Wang

5. Path Aware (PA)-SHORT --- Basic Short-Cut Path Formations Hop (2,1) short-cut Hop (3,2) short-cut • (n, k)  n routing hops can be reduced to k hops Sheng-Shih Wang

6. Path Aware (PA)-SHORT --- Short-cut Occurrence Analysis • Experimental model (proportion of packets encountering short-cuts) (n,2) short-cut is much more frequent than (n,1) short-cut Sheng-Shih Wang

7. Path Aware (PA)-SHORT --- Properties • Work with any underlying routing protocol • Shorten the route according to the up-to-date topology • Different routing protocol uses different SHORT • Hop comparison array • In packet header • <SA, DA, HC, NA> • SA  Source Address • DA  Destination Address • HC  Hop-Count • NA  Neighbor’s (Sender) Address • (current HC – stored HC) > 2  a short-cut path exists Sheng-Shih Wang

8. C H B I D F E A G <A,I,0,A> J Path Aware (PA)-SHORT --- Example Sheng-Shih Wang

9. <A,I,0,A> C H B I D F E A G <A,I,0,A> J <A,I,0,A> Path Aware (PA)-SHORT --- Example (cont.) 0 0 Sheng-Shih Wang

10. <A,I,1,B> <A,I,0,A> C H B I D F E A G <A,I,0,A> J <A,I,1,B> <A,I,0,A> current HC(1) – stored HC(0) = 1  nothing is updated Path Aware (PA)-SHORT --- Example (cont.) 1 1 Sheng-Shih Wang

11. <A,I,1,B> <A,I,0,A> C <A,I,2,C> H B I D F E A G <A,I,0,A> J <A,I,0,A> Path Aware (PA)-SHORT --- Example (cont.) 2 2 Sheng-Shih Wang

12. <A,I,1,B> <A,I,0,A> C <A,I,2,C> H B I D F <A,I,3,D> E A G <A,I,0,A> J <A,I,0,A> Path Aware (PA)-SHORT --- Example (cont.) 3 3 Sheng-Shih Wang

13. <A,I,1,B> <A,I,0,A> C <A,I,2,C> H <A,I,4,E> B I D F <A,I,3,D> 4 E A G <A,I,0,A> J <A,I,0,A> <A,I,4,E> • current HC(4) – stored HC(0) = 4  UPDATE !!! • node J sends message to A to update its routing table • node J modifies its own routing table Path Aware (PA)-SHORT --- Example (cont.) 4 4 Sheng-Shih Wang

14. <A,I,1,B> <A,I,5,F> <A,I,0,A> C <A,I,2,C> H <A,I,4,E> 5 B I D F <A,I,3,D> 5 5 E A G <A,I,0,A> J <A,I,5,F> <A,I,4,E> Path Aware (PA)-SHORT --- Example (cont.) Sheng-Shih Wang

15. <A,I,1,B> <A,I,5,F> <A,I,0,A> C <A,I,2,C> H <A,I,4,E> B I D F <A,I,3,D> 6 E 6 A G <A,I,0,A> J <A,I,5,F> <A,I,4,E> Path Aware (PA)-SHORT --- Example (cont.) Sheng-Shih Wang

16. current HC(7) – stored HC(4) = 3  UPDATE !!! • node F sends message to E to update its routing table <A,I,1,B> <A,I,5,F> <A,I,0,A> <A,I,4,E> C <A,I,2,C> 7 H 7 <A,I,7,H> B I D F 7 <A,I,3,D> <A,I,6,F> E A G <A,I,0,A> J <A,I,5,F> <A,I,4,E> Path Aware (PA)-SHORT --- Example (cont.) Sheng-Shih Wang

17. Path Aware (PA)-SHORT --- Example (cont.) • Before Self-Healing C H B I D F E A G J • After Self-Healing H I F E A J Sheng-Shih Wang

18. Energy Aware (EA)-SHORT --- Concept • Goal • Balancethe energy consumption at all nodes • Idea • Divertthe traffic to other nodes Sheng-Shih Wang

19. Energy Aware (EA)-SHORT --- Basic Cases 1 B S A C D P • Data packets are successively forwarded by A, B, and C, node P can overhear the same packet three times (from A, B, and C) • If node P sees the current energy level of node B, and find out that the difference of energy level as node B and itself is significant enough • Node P will do the redirection Sheng-Shih Wang

20. Energy Aware (EA)-SHORT --- Example 1 3 2 2 A C B 4 P • Overhear table kept in each node has 3 fields: • source-destination pair • sequencenumber • overhearlist: <HC, lvl, NC> Sheng-Shih Wang

21. [S-D, 1, <0, 3, A>] Energy Aware (EA)-SHORT --- Example 1 (cont.) 3 2 2 A C B 4 P Sheng-Shih Wang

22. Energy Aware (EA)-SHORT --- Example 1 (cont.) 3 2 2 A C B 4 P [S-D, 1, <0, 3, A>] [S-D, 1, <1, 2, B>] Sheng-Shih Wang

23. Energy Aware (EA)-SHORT --- Example 1 (cont.) 3 2 2 A C B 4 P [S-D, 1, <0, 3, A>] [S-D, 1, <1, 2, B>] (HC(B) == HC(A) + 1) AND (lvl(P)  MAX(lvl(A), lvl(B) + 2)  candidate B Sheng-Shih Wang

24. Energy Aware (EA)-SHORT --- Example 1 (cont.) 3 2 2 A C B 4 P [S-D, 1, <0, 3, A>] [S-D, 1, <1, 2, B>] [S-D, 1, <2, 2, C>] Sheng-Shih Wang

25. Energy Aware (EA)-SHORT --- Example 1 (cont.) 3 2 2 A C B 4 P [S-D, 1, <0, 3, A>] [S-D, 1, <1, 2, B>] [S-D, 1, <2, 2, C>] • HC(C) == HC(B) + 1 == HC(A) + 2 • lvl(P)  MAX(lvl(A), lvl(C) • lvl(P)  MAX(lvl(B)) + 2 activate this new subpath Sheng-Shih Wang

26. Energy Aware (EA)-SHORT --- Example 1 (cont.) 3 2 2 A C B 4 P Sheng-Shih Wang

27. Energy Aware (EA)-SHORT --- Basic Cases 2 S A B C D P Q • Data packet travels along node A, B, and C, node P knows that it is a neighbor of both node B and its up-stream on-route neighbor (i.e., node A) • Node Q finds that node B needs to be circumvented and Q is adjacent with the down-stream on-route neighbor of node B (i.e., node C) • Node Q broadcasts a message to find the up-stream of node B • Node P gets the message, and replies with an acknowledgement Sheng-Shih Wang

28. 2 2 Case 2 S A B T C D E 4 4 4 2 2 S A B T Case 1 2 2 S A B T C D E 4 4 4 C D E 4 4 4 Energy Aware (EA)-SHORT --- Example Sheng-Shih Wang

29. Simulation Environment • ns-2 simulator • Radio model is based on Lucent Technologies WaveLAN product • Transmission rate: 2 Mbps • Transmission range: 250 m • MAC layer with DCF Sheng-Shih Wang

30. PA-SHORT Simulation --- Simulation Setup • Mobility model • 100 mobile nodes • Transmission range: 2000m  6000m • Pause time (mobility rate): 45, 90, 180, 270, 540 and 720 simulation seconds • Traffic model • 16 simultaneous flows with source-destination pairs spreading randomly • Packet size: 256 bytes Sheng-Shih Wang

31. PA-SHORT Simulation --- Path Optimality path optimality the ratio of the number of hops a packet took to reach its destination over the shortest hop distance between the source-destination pair at the time the packet is sent • AODV does not guarantee shortest path • DSR is better than AODV because it enjoys (n,1) short-cut Sheng-Shih Wang

32. PA-SHORT Simulation --- Delivery Rate • The loss rate is proportional to the route length Sheng-Shih Wang

33. PA-SHORT Simulation --- Routing Overhead routing overhead the total number of routing packets (short-cut information messages are included) divided by the total number of delivered packets Sheng-Shih Wang

34. PA-SHORT Simulation --- Number of Route Requests number of route requests re-broadcasts packets are not counted Sheng-Shih Wang

35. EA-SHORT Simulation --- Simulation Environment 5 src nodes 5 src nodes 5 dest nodes 5 dest nodes Sheng-Shih Wang

36. EA-SHORT Simulation --- Simulation Setup • Non-uniform traffic • 120 nodes • 100 nodes areforwarding nodes • 20 nodes are traffic nodes • 10 nodes are traffic source nodes • 10 nodes are sink nodes • Transmission range: 1200m  1200m • Maximum moving speed: 1, 10, and 20 m/s • Pause time: 0, 200, and infinite seconds • Packet size: 256 bytes Sheng-Shih Wang

37. EA-SHORT Simulation --- Network Capacity Time-Line • pause time  200 secs • moving speed  10 m/s • 6 packets per second per flow Sheng-Shih Wang

38. Conclusion • SHORT • Improve routing optimality dynamically • Redirect the path to the current more optimal one • Higher delivery rate • Longer network lifetime • Future work • QoS requirement • Multicast tree/mesh constructions Sheng-Shih Wang