1. On Optimizing the Backoff Interval for Random Access Scheme Zygmunt J. Hass and Jing Deng IEEE Transactions on Communications, Dec 2003

2. Outline Introduction Introduction Related Work Related Work SBA (Sensing Backoff Algorithm) SBA (Sensing Backoff Algorithm) Performance Evaluation Performance Evaluation Conclusions Conclusions

3. Introduction Two fundamental characteristics of a good backoff algorithm Two fundamental characteristics of a good backoff algorithm High channel throughput High channel throughput Low delay Low delay Fairness among competing nodes should also be considered Fairness among competing nodes should also be considered

4. Related Work(1) Binary exponential backoff (BEB) Binary exponential backoff (BEB)

5. Related Work(2) AB CW=B min =7 AB CW=B min =15 AB CW=B min =7CW=B min =255 ………

6. Related Work(3) Multiplicative increase linear decrease (MILD), introduced in MACAW Multiplicative increase linear decrease (MILD), introduced in MACAW

7. Related Work(4) MILD MILD Maintains a backoff interval for each stream instead of each node Maintains a backoff interval for each stream instead of each node Improves the fairness performance Improves the fairness performance

8. Related Work(5) MILD still has 2 disadvantages MILD still has 2 disadvantages Backoff interval stored into the transmitted packets increases the overhead and the probability of packet collision Backoff interval stored into the transmitted packets increases the overhead and the probability of packet collision Backoff intervals migrate to the whole network, the channel throughput in these areas will be degraded Backoff intervals migrate to the whole network, the channel throughput in these areas will be degraded

9. SBA(1) α( > 1) θ( < 1) β( steps) γ( transmission time of a packet

10. SBA(2) Assumptions Assumptions N nodes within the transmission range of each other ( N <= 100 ) N nodes within the transmission range of each other ( N <= 100 ) Packet collisions are the only source of packet error Packet collisions are the only source of packet error Transmission can be overheard by other nodes Transmission can be overheard by other nodes Packets should be transmitted in turn Packets should be transmitted in turn All nodes have the same CW min and CW max All nodes have the same CW min and CW max

11. SBA(3) T f : duration of faild period T f : duration of faild period I : idle period I : idle period T s : success busy period T s : success busy period U : utilization period U : utilization period

12. SBA(4) Within, the mean interarrival time at each node is B/2, where B is a fixed backoff interval Within, the mean interarrival time at each node is B/2, where B is a fixed backoff interval Prob{ A node starts transmission in } = 2 / B * Prob{ A node starts transmission in } = 2 / B * P s = Prob{ None of the other nodes transmit in γ} = P s = Prob{ None of the other nodes transmit in γ} =

13. SBA(5) By approximating the arrivals of all nodes by a Poisson arrival process, the total arrival rate is N * 2 / B, so the average idle time is By approximating the arrivals of all nodes by a Poisson arrival process, the total arrival rate is N * 2 / B, so the average idle time is

14. SBA(6) The average failed period can be expressed as The average failed period can be expressed as

15. SBA(7) Applying the above 3 equations into channel model, the channel throughput as a function of N and B can be obtains Applying the above 3 equations into channel model, the channel throughput as a function of N and B can be obtains

16. SBA(8) To find the optimum B(B opt ), must be solved numerically To find the optimum B(B opt ), must be solved numerically B opt (N) = 4Nγ B opt (N) = 4Nγ

17. SBA(9)

18. SBA(10)

19. SBA(11)

20. SBA(12)

21. SBA(13) Based on the above calculation of the optimum backoff intervals, we can find the optimum values of (α, β, and γ) Based on the above calculation of the optimum backoff intervals, we can find the optimum values of (α, β, and γ) (α, β, γ) = (1.2, 0.8, 0.93) (α, β, γ) = (1.2, 0.8, 0.93)

22. Performance Evaluation(1)

23. Performance Evaluation(2)

24. Performance Evaluation(3)

25. Performance Evaluation(4)

26. Performance Evaluation(5)

27. Conclusions SBA operates close to the optimum, maximizing the network throughput with fair access from active nodes, without the precise knowledge of the number of active nodes SBA operates close to the optimum, maximizing the network throughput with fair access from active nodes, without the precise knowledge of the number of active nodes SBA can also be extended to other types of MAC schemes SBA can also be extended to other types of MAC schemes

28. Thank you