Sift: A MAC Protocol for Event-Driven Wireless Sensor Networks

1. Sift: A MAC Protocol for Event-Driven Wireless Sensor Networks Kyle Jamieson, Hari Balakrishnan, Y.C. Tay LNCS Springer2006 Suho Yang (CS710: November 4, 2008)

2. Contents Introduction Motivation Our proposal: Sift Performance Evaluation Conclusion & Discussion

3. Introduction • Event-driven WSN • Report an event information when an phenomenon occurs • ↔ periodic traffic • Latency-sensitive applications

4. Introduction • Sift • Focus • Designing MAC protocol to handle event-driven traffic • Challenges • Low latency • Good throughput • Good fairness

5. Motivation: Problems of Traditional CSMA • Problems of traditional contention-based MAC inWSN • Spatial correlated contention • Not suitable for bursty traffic

6. Motivation: Problems of Traditional CSMA • Problems of traditional non-persistent CSMA • Timeslot: opportunity for a node to begin transmitting • Pick a timeslot chosen uniformly in [0, CW] • Listen up to chosen slot • Transmit if nobody else started transmitting • Wait more if somebody else started transmitting • When the channel is idle, • 1-persistent CSMA: transmit immediately with 100% probability • p-persistent CSMA : transmit immediately with p*100% probability • Non-persistent CSMA: transmit after waiting for a random amount of time and checks again Busy Medium

7. Slot choice (slot #4) Slot choice (slot #8) Slot choice (slot #4) Slot choice (slot #4) Motivation: Problems of Traditional CSMA • Problems of traditional non-persistent CSMA • Successful transmission case • Collision case Node A: Node B: Node A: Node B:

8. Motivation: Problems of Traditional CSMA • Problems of traditional non-persistent CSMA • High contention causes collisions in CSMA Due to uniform distribution! Unacceptable collision rate above ~15 transmitting sensors

9. Motivation: Problems of Traditional CSMA • Solution for this problem • Create more slots • Conventional approach • Called “binary exponential backoff” (BEB) Acknowledgement? Yes No Reduce CW Double CW and resend

10. Motivation: Problems of Traditional CSMA • Problem of BEB • Takes time for every node to increase CW • Especially if traffic is spatially-correlated and bursty • Waste backoff slots if collisions cause CW to increase • We are interested in the collision-free transmission of only the first R of N potential reports of some event sink sink

11. Our proposal: Sift • Sift is a MAC protocol for ... • Event-driven traffic • Low-latency requirements • Sift’s properties • Extremely simple • Offers up to 7-fold lower latency • Goal • Design a MAC protocol that minimizes the latency taken to send R of without collisions

12. Our proposal: Sift • Main Idea • Change the way we pick slots • Instead of uniform distribution • Use small and fixed-size contention window • No BEB • Not all sensing nodes need to report an event • It is enough for a subset of the event reports to reach the data sink • Out of N nodes, only the first R nodes report (the remainder are suppressed) • Changing the Distribution for picking transmission slot • Use an geometrically-increasing probability distribution • → Reduce the chance of collisions • → Reduce wastage of backoff slots

13. Our proposal: Sift • Sift’s slot selection distribution • Increasing exponential distribution

14. Our proposal: Sift • Why use this pdf? Nodes choosing each slot → A Bins represent backoff slots →

15. Our proposal: Sift • Optimal non-persistent CSMA performance • With knowledge of number of nodes (IEEE JSAC ’04)

16. Our proposal: Sift • Sift approaches Optimal • Sift needs no knowledge of the number of nodes Sift keeps success rate above this unacceptable range

17. BS Performance Evaluation • Simulation-based • NS-2 • Comparisons with • 802.11 (BEB), 802.11/copy (=copy overheard CW+countdown timer) • Experiment Setup (Event-driven traffic pattern) • Topology: Single-hop to one base station • N nodes sense and report an event • R (≤ N ) reports are required • If a node hears ≥ R reports then it suppresses its own event report E.g. N=4, R=3

18. Performance Evaluation R=16 R=1 • Simulation_1-1: Latency • Sift outperforms when N is large

19. Performance Evaluation • Simulation_1-2: Latency • Sift outperforms as R Increases

20. Performance Evaluation 64 nodes Eight nodes • Simulation_2: Fairness • Sift outperforms 802.11 in terms of fairness

21. Base Station Performance Evaluation • Simulation_3: Hidden terminal experiment setup • Separate 128 sensors into mutually-hidden clusters • Nodes in one cluster cannot hear nodes in another • All nodes send to the base station • Result: hidden terminal collisions at the base station

22. Performance Evaluation Simulation_3: Hidden terminal experiment setup

23. Conclusion • Sift is ideal for sensor networks, where... • it is often sufficient that any R of N sensors that observe an event report it • spatially-correlated contention occurs • sudden changes in the number of sensors that are trying to send data • Key idea • Use a geometrically-increasing probability distribution for picking a transmission slot within a fixed-size contention window • Sift is a latency-enhancing MAC for event-driven sensor networks

24. Discussion • The lack of mentions about ... • Energy consumption • No attention about ... • How to determine R? • Only manual setting • Too strong assumption • Single-hop communication to one base station

25. Thank you

26. Appendix A: Optimal Non-Persistent CSMA • Existing MAC protocols in WSN • Let s be a slot number, assume N≥ 2 sensors transmitting. Define: “Collision Minimizing CSMA and its Applications to Wireless Sensor Networks.” IEEE JSAC, 2004