1. Evaluation of ad hoc routing over a channel switching MAC protocol Ethan Phelps-Goodman Lillie Kittredge

2. Motivation Ad hoc wireless networking is cool –Distributed applications –Scalability Wireless technologies have many channels, but typically only one used Can we take advantage of multiple channels to increase bandwidth? Yes: Channel hopping!

3. A channel hopping protocol Bahl et. al. introduce Slotted Seeded Channel Hopping (SSCH) Nodes switch channels multiple times a second Distributed algorithm so nodes converge to channels with neighbors for whom they have traffic Demonstrates better throughput than 802.11

4. Our Project: SSCH and routing Worthwhile networks will be multihop –Routing is key We see potential conflict with SSCH Bahl et al. do some evaluation, but leave out important aspects We ask: How serious are the effects of SSCH on routing?

5. SSCH protocol Many times a second, change channel according to precomputed schedule Advertise schedule Adjust schedules to –Share channel with a node we have traffic for –Avoid congested channels Collect dynamic data on accuracy of assumptions – if a send fails, assume we were wrong about that neighbor’s schedule

6. Example A CD B

7. Ad hoc routing Dynamic Source Routing (DSR) is popular and well-studied –caches all info it overhears for later use in route building –To establish new flow, sender floods network with request for route

8. Broadcasting Route request system needs broadcast Since nodes are not all on same channel, broadcasts don’t get to everybody in range SSCH deals with this by rebroadcasting 6 times on successive channels Intuitively hacky; not quite semantically the same.

9. Potential problems with DSR over SSCH Node only hears ~1/13 th of incidental routing traffic Route request broadcasts reach fewer other nodes Less information to build new routes off of Have to wait to forward until on same channel as destination

10. Expected symptoms of those problems Longer delay in finding routes Longer routes Longer delay in traffic Less route optimization over time

11. Experimental setup We simulate SSCH on the open source network simulator ns –models wireless channels, channel-switching –Existing DSR and 802.11 implementations Wrote an SSCH module Compared SSCH and 802.11 in various ad hoc networks

12. Why use SSCH? Nodes all within range, pairwise flows Shows flows vs. throughput SSCH scales w/ # flows

13. Why use SSCH? Nodes all within range, pairwise flows Shows flows vs. delay Delay stays small

14. Ad hoc routing Multi-hop ad hoc network, 10 flows SSCH maintains much higher througput

15. Routes and delay Expect SSCH to cause increased delay and longer routes Simulation not reliable enough to draw conclusions

16. Conclusions SSCH greatly increases throughput But to be useful, it must interact well with routing protocols SSCH has a significant effect on performance of DSR But we can’t quantify how much yet.

17. Conclusions Channel hopping protocols need to be designed specifically to interact with ad hoc routing protocols. Possible fixes: –smarter rebroadcasting mechanism –smarter sharing of cache information –Smarter scheduling Share more information between layers

18. Future work More simulation Do other routing protocols interact better? Could we write our own protocol? Implementation on real hardware