A Practical Multi-Channel Access Control Protocol for Wireless Sensor Networks

1. A Practical Multi-Channel Access Control Protocol for Wireless Sensor Networks HieuKhac Le al etc. Presented By XinChe 10/26/09

2. Introduction

3. Related Works

4. Theoretical Analysis • Why do we need to minimize the inter-channel communication ? • Extract cost in channel switching • Retranmissions due to the deafness problem

5. Theoretical Analysis • Observations • New channel should only be allocated when needed. • Some nodes should be more likely to initiate channel switch than others • Global view • Nodes with limited view should act locally to minimize cross-channel communications • The best local action is to follow a node with a better view.

6. Theoretical Analysis • Problem Formulation: • We partition the nodes in the network into different sets, each assigned a separate home channel, such that • The communication within each set is limited to local capacity • Communication across sets is minimized

7. Theoretical Analysis • In a graph, • Each node is a communication device • Link cost is the amount of communication • The K-way Cut Problems • Minimize the K • Each cluster/set should satisfy the capacity constraint.

8. Theoretical Analysis • Solutions to The K-way Cut Problems • [4] O. Goldschmidt and D. S. Hochbaum. Polynomial algorithm for the K-cut problem, 1988 • For K = 3 , O(n4); For K = 4, O(n9) • [10] Y. Kamidoi, S. Wakabayashi, and N. Yoshida. Faster algorithms for finding a minimum K-way cut in a weighted graph, 1997 • More efficient, For K = 3 , O(n4); For K = 4, O(n9) All require a-priori knowledge of K and heavy weight!

9. Theoretical Analysis • Constraints • The K (channel number) is not fixed • Limited capacity for each sensor nodes

10. 3.1.1 The Algorithm

11. 3.1.1 The Algorithm • Channel Congestion Measure

12. 3.1.1 The Algorithm • Channel Switch Method

13. 3.1.1 The Algorithm • Summary The • The question reduces to • who should initiate the split ? • who should follow into the new cluster?

14. 3.1.1 The Algorithm • sinks are better positioned to make decisions on channel allocation ! • observe that in a wireless sensor network, nodes are usually not equal in contributing to network load. • they should act differently in terms of channel switching probability. • two extreme examples : • Data Source node • Sink node • Channel congestion typically occurs at sinks. • Hence, sinks have a more global view of traffic than sources.

15. 3.1.1 The Algorithm

16. 3.1.1 The Algorithm • Summary

17. 3.1.1 The Algorithm • Summary • Channel Expansion • sink-like nodes initiate such transitions with a higher probability and the senders of them followed. • Minimize cross cluster communicaiton • Channel Shrinking • When a channel is no longer congested, nodes on this channel invite those from the next (higher) channel in the ladder to switch to the underutilized frequency. • sink-like nodes initiate such transitions with a higher probability and the senders of them followed.

18. 3.2 The Self-Configuration Problem • considers the dynamics of channel expansion and channel shrinking • it is important that such transitions are stable. • Use feedback Control theory • the control signal is the probability for a node to switch channel • The use of probabilities takes the distributed nature of the control system into account. • Should prevents all nodes from switching at the same time, which would not improve the situation.

19. 3.2 The Self-Configuration Problem • 3.2.1 Channel Expansion if if

20. 3.2 The Self-Configuration Problem • 3.2.3 Choosing the Controller Gains

21. 3.2 The Self-Configuration Problem • 3.2.4 Channel Overflow

22. 4. Protocol Design • Component Structure

23. 4. Protocol Design • Time-trigger Activity • Channel States Update • Neighbors’ Home Channel Maintenance • Message Types

24. 4. Protocol Design

25. 4. Protocol Design • Functional Description

26. 4. Protocol Design • Functional Description

27. 6. Evaluation • 6.1 Experimental Test bed Evaluation

28. 6. Evaluation • 6.1 Experimental Test bed Evaluation

29. 6. Evaluation • 6.1 Experimental Test bed Evaluation

30. 6. Evaluation • 6.1 Experimental Test bed Evaluation

31. 6. Evaluation • 6.1.2 Effect of Utilizing Mutiple Channels

32. 6. Evaluation • 6.1.2 Effect of Utilizing Mutiple Channels

33. 6. Evaluation • 6.1.3 Network of Independent Sub-networks

34. 6. Evaluation • 6.1.3 Network of Independent Sub-networks

35. 6. Evaluation • 6.1.3 Network of Independent Sub-networks

36. 6. Evaluation • 6.1.4 Network of Lightly Connected Sub-networks

37. 6. Evaluation • 6.2 Simulation and Scaling

38. 6. Evaluation • 6.2 Simulation and Scaling

39. 7. Conclusion • Pro • A pratical design, implementation, and evaluaiton of a multi-channel MAC protocol for WSNs. • A distributed heuristic algorithm for channel partition to minimize cross-channel communication. • A feed-back control strategy to stable to channel partition and avoid congestion. • Light-weight MAC protocol on MicaZ motes. • Cons • Only throughput metrics, no delay, message loss • Only Used for converged cast.