1. 1 Enabling High-Bandwidth Vehicular Content Distribution Upendra Shevade, Yi-Chao Chen, Lili Qiu, Yin Zhang, Vinoth Chandar, Mi Kyung Han, Han Hee Song and Yousuk Seung The University of Texas at Austin

2. People want to communicate while on the move – Average one way commute (2005): US: 24.3min, World: 40min – Passengers want to watch videos, listen to songs, etc. Why not just use 3G? – 3G expensive: $30-$60/month 5GB/month -> 2Kbps! 40% 3G capable devices have no 3G plan iPod Touch sales ~ iPhone sales – Bandwidth and backhaul limitations Limited video quality (96-128kbps, < 10min long) Carriers interested in WiFi offloading – Arms race between Increase in cellular bandwidth Higher resolution screens and videos Motivation 2

3. 3 Opportunistic WiFi connectivity Internet Compelling usage scenario Taxis and buses provide value-added services to passengers Previous work: low-bandwidth applications We focus on delivering high-bandwidth content e.g. video streaming Devices in vehicles contact roadside APs Gas stations and local shops deploy APs Passengers watch videos, download files

4. 4 Challenges in Vehicular WiFi Vehicles move at high speed – WiFi contacts are short, intermittent and difficult to predict – 70% contacts less than 10sec [Cartel] Sparse AP coverage – Dense coverage over large area expensive Internet access links to APs are bottleneck – DSL throughput between 768Kbps to 6Mbps – WiFi capacity much higher 11b: 22.2Mbps, 11n (2.4GHz): 39.7Mbps, 11n (5GHz): 56.1Mbps – Naïve solution does not work well Download from Internet during contact Insufficient b/w if data fetched during contact

5. 5 Key Ideas Wireless b/w (e.g. 56Mbps) >> Internet access b/w(~1Mbps) – Predict which APs a car will visit in the near future – Pre-fetch content at AP before vehicle arrives – Vehicle can download at wireless capacity Use vehicles to carry data between APs – Replicate content to those APs not connected to the Internet – APs as content caches for other vehicles Use mesh network as backhaul alternative – APs are often close together to form mesh networks – Content can be replicated over mesh in addition to Internet links – Mesh network can act as content cache

6. 6 Synergy among connections High b/w, short-lived High b/w, high delay Low b/w, persistent High b/w, low coverage VCD High b/w, persistent

7. 7 Contributions New techniques for replication optimization – Goal: Fully utilize wireless bandwidth during contacts – Optimized wireline replication to Internet-connected APs – Replication using vehicular relays to unconnected APs – Use mesh n/w for replication and caching New algorithm for mobility prediction – Predict set of APs that will be visited by vehicle Critical for success of replication techniques – Algorithm: voting among K nearest trajectories

8. 8 VCD Architecture Controller Content Source Internet Download and upload data Upload GPS location updates, video demands, what videos a device has Download and upload data Upload GPS location updates, video demands, what videos a device has

9. 9 Wireline Replication Controller collects vehicle demands for interval (i+1) and what content is present at vehicles and APs Predicts set of APs visited by vehicle in interval (i+1) Computes what content should be replicated to which APs Content servers replicate content to APs At start of interval i Vehicle downloads content from APs During interval (i+1) During interval i

10. 10 Optimized Wireline Replication Interval length, Content present at cars and APs, car demand, AP-to-visit Content to transfer to APs and content to download to cars Total content downloaded to cars weighted by interest, while minimizing the amount of content replicated to APs Total download from AP to car bound by wireless capacity Per-file download to car bound by file size, what car already has Per-file download to car cannot exceed what AP already has and what is replicated to it from the Internet Per-file replication to AP bound by file size and what AP already has Total replication to AP does not exceed Internet access link capacity For each interval i, compute replication strategy maximizing user satisfaction for interval (i+1)

11. 11 Vehicular Replication Vehicles act as data relays between APs Simple strategy: Epidemic Dissemination – Vehicle uploads content to AP Based on expected future demand at AP AP computes future demand, car notifies what it has AP requests content from the car – Vehicle downloads content from AP First of it’s own interest In remaining time, download content randomly

12. Nearby APs can be organized into mesh networks – Replicate content to APs using mesh in addition to Internet link – Fetch missing content from other mesh nodes rather than Internet Changes to linear program – Constraint C3: – Two new constraints: – Objective function: Add Mesh Networks of APs Per-file download to car cannot exceed what AP already has and what is replicated to it from the Internet and from the mesh AP cannot replicate more content over mesh than it has Interference constraint: Total active time of all mesh nodes cannot exceed 100%, assuming all nodes interfere with each other Prefer a replication which uses less mesh traffic among ones supporting equal traffic demands 12

13. 13 Mobility Prediction Predict which APs a car will meet in next interval Challenges: – Vehicles move at high speeds – GPS location updates from vehicles Low frequency Irregular updates – Road and traffic conditions highly dynamic Previous work: 1 st and 2 nd order Markov models – Do not perform well on our dataset

14. 14 Voting among K Nearest Trajectories Exploit history to predict contact: Vehicle’s near history Past trajectories from other vehicles Find K trajectories that most closely match the vehicle’s recent history Obtain future path for K trajectories Report all APs visited by at least T of K trajectories Find K trajectories that most closely match the vehicle’s recent history Obtain future path for K trajectories Report all APs visited by at least T of K trajectories

15. 802.11b testbed – 14 APs deployed inside 8 campus buildings – APs are 20-60ft from the road – 802.11b radios with fixed rate of 11Mbps – 3 APs form a mesh network – Smartphone clients HP iPAQ and HTC Tilt Stream H.264 videos at 64Kbps 802.11n testbed – 4 APs deployed outdoors – APs are 1-5ft from the road – All 4 APs form mesh – Laptops used as clients 15 Testbed Deployment 1 1 2 2 3 3 4 4 5 5 6 6 7,8,9,10, 11,12 13, 14

16. 16 Evaluation Methodology Trace-driven simulation – San Francisco Yellow Cabs 30 day trace of 500 Cabs Contacts with APs: 1120 gas stations, 1620 coffee shops – Seattle city buses Several week-long traces of 1200 buses Contacts with APs: 618 gas stations, 738 coffee shops Two campus testbeds: Prototype implementation – Over 100 hours of road tests Emulab – Run unmodified Controller and AP code – Virtual cars emulating trace mobility Evaluate our system at scale Effect of modifying various parameters Evaluate our system at scale Effect of modifying various parameters Evaluate under real world wireless effects Validate the simulator Show system is efficient and light-weight Validate the simulator Show system is efficient and light-weight

17. #Correctly predicted APs #Total predicted APs Setup: Gas stations as APs, radio range = 200m, prediction interval 3min 1200 Seattle city buses Mobility Prediction Results Voting among K nearest trajectories performs best for our dataset #Correctly predicted APs #Total APs actually visited ( 2 ) (1/precision+1/recall) 17 Bus mobility is more predictable 500 San Francisco Yellow Cabs

18. APs: Gas stations, 100m range Results – Simulation (1/2) Setup: 50 cars, Zipf-like demands, 50% APs not connected to Internet 18 APs: Coffee Shops, 100m range Internet is the bottleneck Benefit from wireline replication Vehicular relay helps! Wireline+relay 5.2X baseline 6.3X better than baseline VCD achieves higher throughput by combining wireline, relay and mesh replication Mesh adds 3-13%

19. Results – Simulation (2/2) Setup: 50 cars, Zipf-like demands, 50% APs not connected to Internet 19 APs: Coffee Shops, 100m range Mesh benefits 14- 20% Benefits increase with higher range and dense AP deployment APs: Coffee shops, 200m range Low Medium High Video quality over 3G

20. 802.11b testbed: 8 APs, 3 connected by mesh 802.11n testbed: 4 APs, all connected by mesh Results - Testbed 20 Download (kB)Play time (sec) No replication292973662 Wireline719308991 Wireline + Mesh794409930 Full replication9249311562 Download (kB)Play time (sec) No replication168572107 Wireline12317515387 Wireline + Mesh13082716353 Full replication13647917060 2.7X 7.8X

21. Summary VCD: high-bandwidth content distribution – Synergy: Internet, wireless, relay, mesh connectivity – Predict AP-vehicle contacts – Wireline replication: LP formulation – Vehicular and mesh replication Deploy on two campus testbeds – Smartphone and laptop clients 21