IEEE 802.11 Protocol: Design and Performance Evaluation of An Adaptive Backoff Mechanism JSAC, vol.18, No.9, Sept. 2000 Authors: F. Cali, M. Conti and E. Gregori April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Chunyu Hu, University of Illinois at Urbana-Champaign Outline Overview of 802.11 backoff algorithm Cali’s model Optimality condition The adaptive backoff algorithm Performance evaluation Conclusion April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Overview of 802.11 Backoff Algorithm The time after a busy period is slotted A node transmits if its backoff timer counts down to 0 Freeze the backoff timer if the medium becomes busy Resume the backoff time if the medium becomes idle for DIFS time April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Chunyu Hu, University of Illinois at Urbana-Champaign Cali’s Model Independent access assumption: Every node transmits at the beginning of idle slots with probability p The access processes are regenerative: idle period, virtual transmission time The protocol capacity (i.e. total throughput): Avg. msg length virtual transmission time April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Cali’s Model (Continued) Derive E[tv] (renewal process) Derive them based on the independent access assumption Nc, r.v. ~ Negative Binomial (1, PColl/PSuss) Idle_p, r.v. ~ Negative Binomial (1, PIdle) April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
The Optimality Condition Choosing p (=pmin) can maximize max pmin approximately satisfies the following condition: Left: the average channel time spent in collision Right: the average channel time spent in idle April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Chunyu Hu, University of Illinois at Urbana-Champaign Properties The pmin value from the optimality condition guarantees that E[NC] < 1 Throughput obtained > 0 Proof of the above and derivation of the condition, see [1] F. Cali, M. Conti and E. Gregori, “Dynamic Tuning of the IEEE 802.11 Protocol to Achieve a Theoretical Throughput Limit,” IEEE Trans. on Networking, Vol.8, No.6, Dec. 2000 [2] R.G. Gallager, “A Perspective on Multiaccess Channels”, IEEE Trans. Information Theory, vol.31, pp124-142, 1985 April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
An Illustration of pmin Estimate [1] April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Protocol Capacity Comparision Study the protocol capacity of theoretical bound, IEEE 802.11 and the proposed dynamic IEEE 802.11. April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
From the Optimality Condition to the Adaptive Backoff Algorithm Given the probability p, can estimate M from measured E[Idle_p]: M known, can compute p to satisfy the optimality condition from measured E[Coll]: April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Chunyu Hu, University of Illinois at Urbana-Champaign Outline Overview of 802.11 backoff algorithm Cali’s model Optimality condition The adaptive backoff algorithm Performance evaluation Conclusion April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
The Adaptive Backoff Algorithm – Measurements Measured variables (E[Idle_p] and E[Coll]) and estimated variables (M, p) are updated using the moving averaging window, e.g., -- smoothing factor April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
The Adaptive Backoff Algorithm – Algorithm Performed by each node at the end of every transmission interval April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Chunyu Hu, University of Illinois at Urbana-Champaign Outline Overview of 802.11 backoff algorithm Cali’s model Optimality condition The adaptive backoff algorithm Performance evaluation Conclusion April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Performance Evaluation (1) The estimates of variables under different α, M = 10. April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Performance Evaluation (2) The estimates of variables under different α, M = 20. April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Performance Evaluation (3) Study the convergence rate. April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Performance Evaluation (4) Study the convergence rate when the start state is wrong. M = 10. April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign
Chunyu Hu, University of Illinois at Urbana-Champaign Conclusion Model 802.11 MAC and study the protocol capacity Estimate pmin with an optimality condition Design an adaptive backoff algorithm April 05, 2006 Chunyu Hu, University of Illinois at Urbana-Champaign