Predictable Performance Optimization for Wireless Networks

1. Predictable Performance Optimization for Wireless Networks Lili Qiu University of Texas at Austin lili@cs.utexas.edu Joint work with Yi Li, Yin Zhang, Ratul Mahajan, and Eric Rozner ACM SIGCOMM 2008 August 21, 2008

2. Motivation • Wireless networks are becoming ubiquitous • Managing wireless networks is hard • Our goal: develop systematic techniques to optimize wireless performance Wireline Wireless

3. Unpredictability of wireless networks 50% 100% bad-good D R S 100% 50% good-bad D S S R S Need predictable wireless performance optimization.

4. Model-driven optimization framework Network model Network measurement Traffic demands Optimization prescribedflow rates Routing Performance objectives: - Maximize fairness, total throughput, …

5. Existing models are insufficient • Asymptotic performance bounds [GP00,LB+01,GT01,GV02] • Cannot model any specific networks • Conflict graph based model [JPPQ03] • Assumes perfect scheduling and overestimates 802.11 performance • Requires an exponential number of constraints • 802.11 DCF models [Bianchi00,KA+05,GLC06,GSK05 QZWH+07,KDG07] • Not general: restricted topologies or traffic demands • Cannot be easily incorporated into optimization procedure Need a better 802.11 network model for optimization.

6. Our network model • Provide a compact characterization of feasible solution space to facilitate optimization • Simple: O(N) constraints for N links • Flexible and accurate • Handle asymmetric link loss rate • Handle asymmetric interference • Handle hidden terminals • Handle heterogeneous, multihop traffic demands Throughput constraints Loss rate constraints Sending rate constraints Network model Network measurement

7. Throughput constraints • Divide time into variable-length slot (VLS) • 3 types of slots: idle slot, transmission slot, deferral slot Success probability Probability of starting tx in a slot Expected payload transmission time Expected duration of a variable-length slot

8. Loss rate constraints • Inherent and collision loss are independent • Inherent loss • Based on one-sender broadcast measurement • Collision loss • Synchronous loss • Two senders can carrier sense each other • Occur when two transmissions start at the same time • Asynchronous loss • At least one sender cannot carrier sense the other • Occur when two transmissions overlap

9. Sending rate feasibility constraints • 802.11 unicast • Random backoff interval uniformly chosen [0,CW] • CW doubles after a failed transmission until CWmax, and restores to CWmin after a successful transmission or when max retry count is reached • CW(pi): the expected contention window size under packet loss rate pi [Bianchi00] • Sending rate feasibility constraints SIFS ACKTransmission DIFS Random Backoff Data Transmission

10. Extensions to the basic model • RTS/CTS • Add RTS and CTS delay to VLS duration • Add RTS and CTS related loss to loss rate constraints • Multihop traffic demands • Link load routing matrix  e2e demand • Routing matrix gives the fraction of each e2e demand that traverses each link • TCP traffic • Update the routing matrix: where reflects the size & frequency of TCP ACKs

11. Model-driven optimization framework Network model Network measurement Traffic demands Optimization prescribedflow rates Routing Performance objectives: - Maximize fairness, total throughput, …

12. Flow throughput feasibility testing • Test if given flow throughput are achievable • Challenge: strong interdependency • Our approach: iterative procedure Input: throughput Output:feasible/infeasible Initializeτ= 0 and p = pinherent no Estimate τ from throughput and p Check feasibility constraints yes Converged? Estimate p from throughput andτ

13. Fair rate allocation Initialization: add all demands to unsatSet Scale up all demands in unsatSet until some demand is saturated or scale1 yes if (scale 1) no Move saturated demands from unsatSet to X yes if (unsatSet≠) no Output X

14. Total throughput maximization • Formulate a non-linear optimization problem (NLP) • Solve NLP using iterative linear programming Maximize total throughput Link load is bounded bythroughput constraints Sending rate is feasible E2e throughput is bounded by demand

15. Evaluation methodology • Model validation • How to quantify over-prediction error? • Verify if prescribed rates are achievable • How to quantify under-prediction error? • Scale up all prescribed rates by a common factor • Performance optimization • Fairness maximization: Jain’s fairness index • Total throughput maximization • This talk: testbed results only • 19 mesh nodes at UTCS building; up to 7 hops • Extensive simulation results are in the paper

16. Optimization schemes • Our rate optimization • No rate optimization (current practice) • Conflict graph based optimization • Plug conflict graph model to our framework • Conflict graph assumes perfect scheduling [JPPQ03] • Represent each wireless link with a vertex • Draw an edge between the vertices if the corresponding links interfere • Derive clique constraints – all links in a clique in the CG cannot be active together

17. Baseline: conflict graph model TCP UDP y=x y=x y=0.8x y=0.8x CG model significantly over-estimates sending rates.

18. Model validation: UDP traffic y=x y=0.8x 1) Most estimated rates are achievable within 20%.2) Rates scaled up by just 10% become unachievable.

19. Model validation: TCP traffic y=x y=0.8x Our model is accurate for TCP traffic.

20. Maximizing fairness UDP TCP Fairness index is close to 1 under our scheme, while it degrades quickly in other schemes.

21. Maximizing total throughput TCP UDP Our scheme significantly increases total throughput.

22. Conclusions • Main contributions • Predictable wireless performance optimization • A simple yet accurate wireless network model • Effective model-driven optimization algorithms • Demonstrate their effectiveness through testbed experiments and simulation • Future work • Handle dynamic traffic and topologies • Use passive measurement to seed our model

23. Thank you!

24. Impact on different routing schemes UDP TCP Our scheme helps all routing schemes considered.

25. TCP Pathologies under no rate control D1 S1 R D2 S2 TCP cannot set the rate that maximizes throughput.