1. Augmenting Mobile 3G Using WiFi Sam Baek Ran Li Modified from University of Massachusetts Microsoft Research

2. Outline The necessity of augmenting 3G Basic idea of Wiffler Improvement of Wiffler and test results Questions 2

3. Demand for mobile access growing 3 Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2011–2016 global mobile data traffic will increase 18-fold between 2011 and 2016. All of this is understandable given the massive adoption of mobile devices such as smartphones. Mobile data traffic will grow at a compound annual growth rate (CAGR) of 78 percent from 2011 to 2016, reaching 10.8 exabytes per month by 2016.

4. How can we reduce 3G usage? 1. Behavioral 2. Economic 3. Technical 4 like ATT wants to educate users by imposing a limitation of 5GB per month Data Plan Using WiFi to reduce 3G traffic

5. Augmenting Mobile 3G using WiFi Offload data to WiFi when possible Easy to do when you are stationary Focus on vehicular mobility 5

6. Offloading 3G data to WiFi 6 Wiffler

7. Basic Information 1.What is the availability of 3G and WiFi networks as seen by a vehicular user? 2.What are the performance characteristics of these two networks? (throughput and loss rate) 7

8. 8 Measurement  Measurement: Joint study of 3G and WiFi connectivity  Across three cities: Amherst, Seattle, SFO  Testbed: Vehicles with 3G modom and WiFi (802.11b) radios  Amherst: 20 cars, Seattle: 1 car, SFO: 1 car  Software: Simultaneously probes 3G and WiFi  Availability, loss rate, throughput  Duration: 3000+ hours of data over 12+ days

9. 3G and WiFi access availability 9 Availability (%) 3G+WiFi combination is better than 3G

10. Special distribution of 3G/WiFi availability 10 Amherst

11. WiFi (802.11b) throughput is lower 11 Cumulative fraction WiFi 3G WiFi 3G Upstream Downstream 0.350.72 Throughput = Total data received per second 0.40.8

12. WiFi loss rate is higher 12 Cumulative fraction WiFi 3G 28% 8% Loss rate = Fraction of packets lost at 10 probes/sec

13. Summary In summary, the measurement study shows that A non-trivial amount of WiFi is available, but is limited around 10 percent. (3G:90%) Unlike stationary environments, WiFi throughput is much lower than 3G throughput. The WiFi loss rate performance is also poorer compared to 3G. 13

14. 14 Implications of measurement study  Wiffler : simply switch from 3G to WiFi  Drawbacks  Can offload only ~11% of the time  Can hurt applications because of WiFi’s higher loss rate and lower throughput. (VoIP)

15. 15 Key ideas in Wiffler Increase savings for delay- tolerant applications  Problem: Using WiFi only when available saves little 3G usage  Solution: Exploit delay- tolerance to wait to offload to WiFi when availability predicted Reduce damage for delay- sensitive applications  Problem: Using WiFi whenever available can hurt application quality  Solution: Fast switch to 3G when WiFi delays exceed threshold

16. Prediction-based offloading D = Delay-tolerance threshold (seconds) S = Data remaining to be sent (bytes) Each second, 1. If (WiFi available), send data on WiFi 2. Else if (W(D) < S), send data on 3G 3. Else wait for WiFi. 16 Predicted WiFi transfer size in next D seconds

17. 17 Negligible benefits with more sophisticated prediction, eg future location prediction + AP location database Predicting WiFi capacity  History-based prediction of # of APs using last few AP encounters  WiFi capacity = (expected #APs) x (capacity per AP)  Simple predictor yields low error both in Amherst and Seattle

18. 18 Fast switching to 3G  Problem:  WiFi losses bursty => high retransmission delay  Approach:  If no WiFi link-layer ACK within 50ms, switch to 3G  Else, continue sending on WiFi

19. Wiffler implementation 19 Wiffler proxy  Prediction-based offloading upstream + downstream  Fast switching only upstream  Implemented using signal-upon-ACK in driver

20. 20 Evaluation Roadmap  Prediction-based offloading  Deployment on 20 DieselNet buses in 150 sq. mi region around Amherst  Trace-driven evaluation using throughput data  Fast switching  Deployment on 1 car in Amherst town center  Trace-driven evaluation using measured loss/delay trace using VoIP-like probe traffic

21. Deployment results Data offloaded to WiFi Wiffler’s prediction-based offloading 30% WiFi when available10% 21 % time good voice quality Wiffler’s fast switching68% WiFi when available (no switching)42% File transfer size: 5MB; Delay tolerance: 60 secs; Inter-transfer gap: random with mean 100 secs VoIP-like traffic: 20-byte packet every 20 ms

22. 22 Trace-driven evaluation  Parameters varied  Workload, AP density, delay-tolerance, switching threshold  Strategies compared to prediction-based offloading:  WiFi when available  Adapted-Breadcrumbs: Future location prediction + AP location database  Oracle (Impractical): Perfect prediction w/ future knowledge

23. Wiffler increases data offloaded to WiFi 23 Workload: Web traces obtained from commuters Wiffler increases delay by 10 seconds over Oracle. 42% 14% Wiffler close to Oracle Sophisticated prediction yields negligible benefit WiFi when available yields little savings

24. Even more savings in urban centers 24

25. Fast switching improves quality of delay-sensitive applications 25 40% 58% 73% 30% data offloaded to WiFi with 40ms switching threshold

26. 26 Future work  Reduce energy to search for usable WiFi  Improve performance/usage by predicting user accesses to prefetch over WiFi  Incorporate evolving metrics of cost for 3G and WiFi usage

27. 27 Summary  Augmenting 3G with WiFi can reduce pressure on cellular spectrum  Measurement in 3 cities confirms WiFi availability and performance poorer, but potentially useful  Wiffler: Prediction-based offloading and fast switching to offload without hurting applications Questions?

29. Demand projected to outstrip capacity 29 http://www.nytimes.com

30. Error in predicting # of APs 30 Relative error N=1 N=4 N=8

31. Fast switching improves performance of demanding applications 31 % time with good voice quality Oracle Only 3G Wiffler No switching