Vehicular Network Applications

1. Vehicular Network Applications VoIP Web Email Cab scheduling Congestion detection Vehicle platooning Road hazard warning Collision alert Stoplight assistant

2. Congestion Detection Vehicles detect congestion when: # Vehicles > Threshold 1 Speed < Threshold 2 Relay congestion information Hop-by-hop message forwarding Other vehicles can choose alternate routes

3. Deceleration Warning Prevent pile-ups when a vehicle decelerates rapidly 2004, over 2,300 deaths from rear-end collisions2004, over 2,300 deaths from rear-end collisions

4. Wireless Technologies for Vehicular Networks Cellular networks High coverage, low bandwidth, expensive WiFi networks Moderate coverage, high bandwidth, free Combine all of them to achieve low cost, high bandwidth, and high coverage Sprint's newly launched Xohm service is now offering America's first WiMax network. Computerworld's Brian Nadel went to Baltimore to try it out, and he reports that Xohm delivered data smoothly to a car moving at highway speeds, played YouTube videos flawlessly, and on average, pushed through more than 3Mbit/sec., compared with 1.3 Mbit/sec. for the AT&T network Brian used as a comparison. But right now, coverage is only planned in a few US cities; if Sprint isn't able to ramp up its coverage quickly, it may lose its advantage." Sprint's newly launched Xohm service is now offering America's first WiMax network. Computerworld's Brian Nadel went to Baltimore to try it out, and he reports that Xohm delivered data smoothly to a car moving at highway speeds, played YouTube videos flawlessly, and on average, pushed through more than 3Mbit/sec., compared with 1.3 Mbit/sec. for the AT&T network Brian used as a comparison. But right now, coverage is only planned in a few US cities; if Sprint isn't able to ramp up its coverage quickly, it may lose its advantage."

6. Target Scenarios A car is within the range of multiple APs How common? Low data rate but low delay Alternatives?

7. Overview 7

8. Outline 8

9. VanLAN: Vehicular Testbed 9

10. Measurement study 10

11. Handoff policies studied Practical hard handoff Associate with one BS Current 802.11 Ideal hard handoff Use future knowledge Impractical 11

12. Handoff policies studied Practical hard handoff Associate with one BS Current 802.11 Ideal hard handoff Use future knowledge Impractical Ideal soft handoff Use all BSes in range Performance upper bound 12

13. Comparison of handoff policies 13

14. Outline 14

15. Design a practical soft handoff policy 15

16. Why are existing solutions inadequate? 16

17. 17

18. 18

19. 19

21. 21

22. 22

23. 23

24. Outline 24

25. Evaluation 25 Mention the voip and tcp were deployedMention the voip and tcp were deployed

26. ViFi reduces disruptions in our deployment 26

27. ViFi improves VoIP performance 27

28. ViFi improves performance of short TCP transfers 28

29. ViFi uses medium efficiently 29

30. Conclusions Improves performance of interactive applications for vehicular WiFi networks Interactive applications perform poorly in vehicular settings due to frequent disruptions ViFi, a diversity-based handoff protocol significantly reduces disruptions Experiments on VanLAN shows that ViFi significantly improves performance of VoIP and short TCP transfers 30

31. Comments Interesting problem domain Target low-bandwidth applications, for which cellular networks are sufficient Have multiple APs within range tuned into the same channel May not be common and lose spatial diversity Use the lowest data rate Common to have multiple or fewer than 1 relay(s) for each tx Relay is not compelling Uplink: sufficient to relay data to one AP Downlink: if best AP is selected, the need for relay is low If relay has to be used, MORE like opportunistic routing may be more efficient They dismissed opportunistic routing due to its potential large delay due to batch But their delay can be high since retx timeout is generally large in order to account for variable contention delay

32. Modulation Rate Adaptation in Vehicular Environments:Cross-Layer Implementation and Experimental Evaluation Joseph Camp Edward Knightly ACM MobiCom 2008

33. Background: Link Characteristics Time-varying link quality � Mobility of sender, receiver, or obstacles - Multiple paths existing Ideal modulation rate for channel condition Modulation rate with highest throughput for channel condition

34. Goal of Protocol Designer Use available information (loss, SNR, �) to track ideal modulation rate Many protocols have been invented ARF, RBAR, OAR, RRAA, CARA, ONOE, �

35. Problem Existing rate adaptation algorithms fail to track the ideal rate � Urban propagation environment � Even with non-mobile sender and receiver� Result = loss and under-utilization

36. Objective Understand the origins of the failure to track link variation Identify core mechanisms needed to succeed in urban channels

37. Methodology Unified Implementation Platform � Implement multiple algorithms on a common platform � First implementation of SNR-based protocols � Extract General Rate Adaptation Principles Evaluate rate selection accuracy packet-by-packet Compare against ideal rate found via exhaustive search Use repeatable controlled channels Accurately measured outdoor channels Design core mechanisms to track real-world link variation

38. Wireless Open-Access Research Platform (WARP) Limits of Off-the-shelf platforms � Programmability and observability WARP is clean-slate MAC and PHYneeded to implement: � CSMA/CA (802.11-like MAC) Cross-layer rate adaptation framework � Core mechanisms for rate selection protocols � Channel measurements � Evaluation of selected rate versus ideal rate

39. Rate Adaptation Schemes Studied Consecutive packet decision 10 success ? increase rate 2 failures ? decrease rate Historical decision Compute pkt loss rate using a window and select the rate that gives the highest throughput SNR based RTS/CTS/DATA/ACK, where CTS reports channel quality Equal air-time assuration Measure SNR per data packet Opportunistic better channel Send back-to-back pkts (without backoff) whenever the rate is above the base rate Is it a good idea?

40. Rate Adaptation Accuracy Ideal rate found via exhaustive search of channel condition Consider case where at least one modulation rate succeeds Rate Selection Accuracy Categories Over-selection (loss) Accurate (achieving optimal rate) Under-selection (under-utilization)

41. Experimental Design Repeatable channels � Mean channel quality � Channel fading/coherence time � Multipath effect and interference Accurately measure urban channels � Residential and downtown scenarios � Measure coherence time � Static and vehicular Topologies Competing links (in paper) � Indoor, controlled environment � Urban environment

42. Impact of Coherence Time Issue: Increase fading of the channel to evaluate if rate adaptation can track Similar performance with long coherence of channel SNR: high overhead penalty (contrasts result of protocol designer) Opportunistic: overcomes RTS/CTS overhead penalty Dissimilar performance at short coherence of channel

43. Opposite Rate Choice Inaccuracies Issue: Packet-by-packet accuracy to reveal why throughput is low Average vs. consecutive mechanisms � Consecutive low due to underselection SNR: extremely low throughput � Due to overselection (loss) Per-packet analysis needed to show poor rate adaptation behavior

44. SNR-based Coherence Time Sensitivity Issue: SNR rate selection is per-packet (should track fading), why inaccurate? Fast to slow channel fading Accurate at long coherence Overselect at <1ms Overselection caused by coherence time sensitivity of SNR-rate relationship

45. Joint Consideration of SNR andCoherence Time Consider different SNR thresholds according to coherence time Ideal rate = f(SNR, CT)

46. Joint Consideration of SNR andCoherence Time Consider different SNR thresholds according to coherence time Ideal rate = f(SNR, CT) Retrain SNR-based decision (for the same protocol) Joint consideration of SNR and coherence time provides large gains

47. Scenarios and Channel Measurements Residential Urban (TFA) Single-family residential, dense foliage Coherence Time: 100 ms on average Driven to 15 ms with mobility of scatterers (in static topology) Downtown Houston Both sides of street lined with tall buildings (strong multipath) Coherence Time: 80 ms on average Driven to 300 usec with mobility of scatterers (in static topology)

48. Outdoor Static Topologies Issue: Evaluate rate adaptation accuracy in outdoor scenarios Consecutive and average: inaccurate in outdoor settings Downtown (strong multipath) Force loss-based to underselect SNR: over and underselect with low coherence time

49. Static Sender to Mobile Receiver (Urban) Issue: Evaluate rate adaptation ability to track with mobility SNR protocols are able to plateau for >4 sec Per-packet decision Loss-based protocols only able to spike to suboptimal rate choices Loss sensitivity prevents protocol from tracking Loss-based protocols unable to track with mobility

50. Heterogeneous Competing Links Lack of loss distinction Causes underselection Collision/fading differentiation able to overcome with equal links Large imbalances for slight differences in competing links Residential Urban Scenario Competing links with vehicular mobility

51. Heterogeneous Competing Links 51

52. Summary Implementation of multiple and previously unimplemented rate adaptation mechanisms and found via per-packet inspection Loss-based core mechanisms underselect with Fast-fading, interference, competing links (even with collision/fading differentiation), and mobile environments SNR-based mechanisms overselect with fast-fading but have Large gains from considering SNR and coherence time jointly Robust to interference, competing links, and mobility Despite 4-way handshake, SNR-based protocols outperform loss-based protocols

53. Comments RTS/CTS is expensive How to reduce its cost? How to model f(SNR, coherence time) Pure measurements as done in the paper does not scale Coherence time is continuous and infeasible to retrain with all possible values How about different packet sizes How to estimate coherent time in commodity hardware?

54. 54

55. 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 cellular networks? 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 Goal: Enable high bandwidth applications (e.g., video) in vehicular networks via WiFi Motivation 55

56. 56 Opportunistic WiFi connectivity

57. Synergy among connections

58. 58 Contributions New techniques for replication optimization Goal: Fully utilize wireless bandwidth during contact Optimized wireline replication to Internet-connected APs Replication using vehicular relays to unconnected APs Use mesh 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

59. Trace-driven simulation and emulation San Francisco cabs, Seattle buses, Shanghai cabs Two testbeds on UT campus 802.11b: 14 APs deployed inside 8 campus buildings, 20-60ft from the road 802.11n: 4 APs outdoor, 1-5ft from the road Smartphone and laptop clients HP iPAQ and HTC Tilt Stream H.264 videos at 64Kbps 59 Evaluation Methodology

60. Summary: Vehicular Content Distribution KNT: A new mobility prediction algorithm Based on voting among K nearest trajectories 25-94% more accurate than 1st and 2nd order Markov models A series of novel replication schemes Optimized wireline replication and mesh replication Opportunistic vehicular relay based replication Extensive evaluation: simulation + testbed + emulation Simulation using San Francisco taxi and Seattle bus traces 3-6x of no replication, 2-4x of wireline or vehicular alone Full-fledged prototype deployed on two real testbeds 14-node 802.11b testbed and 4-node 802.11n testbed 4.2-7.8x gain over no replication Emulab emulation with real AP/controller and emulated vehicles Show system works at scale and is efficient Validate our trace-driven simulator