1. A Better Choice for Sensor Sleeping Ou Yang and Wendi Heinzelman 1-28-2009

2. 2 Outline Motivation and Goal Methodology - Sleep at single layer individually - Sleep at multiple layers simultaneously Simulation Results - Performance of single layer sleeping - Comparisons of single layer sleeping - Comparisons of multi-layer sleeping Conclusions

3. 3 Motivation Wireless sensors are energy constrained Save energy to extend network lifetime Make sensors sleep when they are not used

4. 4 Motivation How to make sensors sleep? Application Layer Routing Layer MAC Layer Source node selection e.g. target tracking [1] Topology control e.g. GAF [2] Routing protocol e.g. [3] Duty-cycled MAC protocol e.g. SMAC [4]

5. 5 Goal Which is better? - Given application requirements - Given network scenario Is there a single layer providing most benefit? Multi-layer better than individual layer? Need cross-layer coordination? Application Layer Routing Layer MAC Layer Source node selection e.g. target tracking [1] Topology control e.g. GAF [2] Routing protocol e.g. [3] Duty-cycled MAC protocol e.g. SMAC [4]

6. 6 Methodology Focus on the routing and MAC protocol Application Layer Routing Layer MAC Layer Source node selection e.g. target tracking [1] Topology control e.g. GAF [2] Routing protocol e.g. [3] Duty-cycled MAC protocol e.g. SMAC [4]

7. 7 Methodology All combinations of sleeping schemes Non-sleepingSleeping Directed Diffusion [5] Sleeping Directed Diffusion IEEE 802.11 [6] SMAC [4] Routing Protocol MAC Protocol

8. 8 Non-sleeping Routing Protocol - Directed Diffusion [5] Data-centric - Data is named in attribute-value pairs Exploratory phase - Discover source nodes - Low data rate flooding Reinforcement phase - “Pull down” data from selected source node - High data rate unicast

9. 9 Directed Diffusion SinkSource INTEREST (periodic flooding) exploratory DATA (periodic flooding) positive REINFORCEMENT (unicast) high rate DATA (unicast)

10. 10 Directed Diffusion Interest Data Interest (low rate) Pos reinforcement (high rate) exploratory data (low rate) Data (high rate) Put them to sleep!

11. 11 Sleeping Routing Protocol - Sleeping Directed Diffusion Node INTEREST (periodic flooding) exploratory DATA (periodic flooding) positive REINFORCEMENT (unicast) high rate DATA (unicast) INTEREST timer gap DATA timer gap

12. 12 Sleeping MAC – SMAC [4] Duty-cycled MAC - Fixed awake time in a cycle Synchronization - Periodic SYNC packets Similar to IEEE 802.11 - CSMA - RTS/CTS

13. 13 Sleep at both layers Sleeping Directed Diffusion SMAC Routing Protocol MAC Protocol Coordination? How to coordinate?

14. 14 Sleep at both layers Problem - When sleeping at routing layer, no need to wake up at MAC layer Coordination - Routing protocol has higher priority to put sensor to sleep

15. 15 Sleep at both layers Problem - MAC updates topology according to periodic SYNC packets - MAC cannot differentiate routing layer sleeping and energy depletion Coordination - Add energy info into SYNC packets

16. 16 Performance: single layer sleeping QoS pause Lower data deliver ratio

17. 17 Comparisons: single layer sleeping Change node density Higher density Higher contention Lower throughput Lower delivery ratio Higher density Higher redundancy Higher throughput Lower delivery ratio due to more QoS pauses

18. 18 Comparisons: single layer sleeping Change network scale Larger network More hops to sink Lower throughput Lower delivery ratio Larger network Same redundancy Similar throughput Similar delivery ratio

19. 19 Comparisons: single layer sleeping Change number of source nodes More source nodes, fewer hops to sink Higher throughput Higher delivery ratio More source nodes, more redundancy Higher throughput Lower delivery ratio due to more QoS pauses

20. 20 Comparisons: single layer sleeping Change application data rate Higher data rate, higher contention Lower throughput/rate Lower delivery ratio Higher data rate, same redundancy Similar throughput/rate Similar delivery ratio

21. 21 Comparisons: multi-layer sleeping Change node density No coordination no gain Need coordination in lower contention scenarios Sacrifice delivery ratio

22. 22 Discussions Other routing protocols? - Sensors sleep when not routing data - e.g., Geographic routing Other MAC protocols? - Duty-cycled MAC - e.g., T-MAC, B-MAC

23. 23 Conclusion and Future Work MAC layer sleeping is better when - Low contention - Small network scale Routing layer sleeping is better when - Higher redundancy - Larger network scale Multi-layer sleeping is better when - Proper coordination - Low contention

24. 24 References [1] Zoghi, M.R., Kahaei, M.H.: Sensor Selection forTarget Tracking in WSN Using Modified INS Algorithm. In: 3rdInternatioanl Conference on Information and CommunicationTechnologies: From Theory to Applications, pp. 1-6. 2008 [2] Xu, Y, Heidemann, J., Estrin, D.: Geography-informed Energy Conservation for Ad HocRouting. In: 7th Annual International Conference on Mobile Computingand Networking, pp. 70-84. 2001 [3] Zheng, R., Kravets, R.: On-demand PowerManagement for Ad Hoc Networks. In: 22nd Annual Joint Conference ofthe IEEE Computer and Communications Societies, vol.1, pp. 481-491. 2003 [4] Ye, W., Heidemann, J., Estrin, D.: Medium Access Control with Coordinated Adaptive Sleeping for WirelessSensor Networks. IEEE/ACM Trans. on Networking, vol. 12,issue 3, pp. 493-506. 2004 [5] Intanagonwiwat, C. Govindan, R. Estrin, D. Heidemann,J.: Directed Diffusion for Wireless Sensor Networking. IEEE/ACM Trans. on Networking, vol. 11, issue 1, pp. 2-16. 2003 [6] LAN MAN Standards Committee of the IEEE Computer Society: Wireless LAN medium access control (MAC) and physical layer (PHY) specification, IEEE, New York, NY, USA, IEEE Std 802.11-1997 edition, 1997

25. 25 Thank you Q & A