1. Vehicular Grid Communications: the role of the Internet Infrastructure Wicon 2006 Boston, August 3, 2006 Presented by Mario Gerla UCLA CSD gerla@cs.ucla.edugerla@cs.ucla.edu www.cs.ucla.edu/NRL

2. Outline  Emerging urban vehicle applications  Routing in a highly mobile urban environment  A case for geo-routing  Extending Geo routing to the infrastructure  To use or not to use the infrastructure?  Load balancing  Conclusions

3. Urban Vehicle Grid Applications –Safe navigation –Content distribution (video, ads) –Vehicle as mobile sensor platform

4. Safe Driving Vehicle type: Cadillac XLR Curb weight: 3,547 lbs Speed: 65 mph Acceleration: - 5m/sec^2 Coefficient of friction:.65 Driver Attention: Yes Etc. Vehicle type: Cadillac XLR Curb weight: 3,547 lbs Speed: 45 mph Acceleration: - 20m/sec^2 Coefficient of friction:.65 Driver Attention: No Etc. Vehicle type: Cadillac XLR Curb weight: 3,547 lbs Speed: 75 mph Acceleration: + 20m/sec^2 Coefficient of friction:.65 Driver Attention: Yes Etc. Vehicle type: Cadillac XLR Curb weight: 3,547 lbs Speed: 75 mph Acceleration: + 10m/sec^2 Coefficient of friction:.65 Driver Attention: Yes Etc. Alert Status: None Alert Status: Passing Vehicle on left Alert Status: Inattentive Driver on Right Alert Status: None Alert Status: Slowing vehicle ahead Alert Status: Passing vehicle on left

5. Co-operative Download: Car Torrent Vehicle-Vehicle Communication Internet Exchanging Pieces of File Later

6. Digital Billboard: Ad Torrent Every Access Point (AP) disseminates Ads that are relevant to the locality Passing cars pick up the Ads An Ad can be: –simple text message –trailer of nearby movies, –virtual tour of hotels etc Business owners in the vicinity subscribe to this digital billboard service for a fee.

7. Vehicular Sensor Network (VSN) (UCLA) Infostation Car-Car multi-hop 1. Fixed Infrastructure 2. Processing and storage 1. On-board “black box” 2. Processing and storage Car to Infostation

8. Vehicular Sensor Applications Environment –Traffic congestion monitoring –Urban pollution monitoring Civic and Homeland security –Forensic accident or crime site investigations –Terrorist tracking after strike

9. Accident Scenario: storage and retrieval Designated Cars: –Continuously collect images - cars, license plates etc (store data locally) –Process the data and detect an event –Classify the event as Meta-data (Type, Option, Location, Vehicle ID) –Post it on “distributed index” (P2P network) Police search the index and retrieve data from “witness” cars Crash! Meta-data : Img, -. (10,10), V10 Meta-data : Img, Crash, (10,5), V12

10. How to set up the index and retrieve the data? “Epidemic diffusion” : –Mobile nodes periodically broadcast meta-data of events to their neighbors A mobile agent (the police) queries nodes and harvests {event + witness ID} Data dropped when stale and/or geographically irrelevant

11. Epidemic Diffusion + Harvesting

12. 1) “periodically” Relay (Broadcast) its Event to Neighbors 2) Listen and store other’s relayed events into one’s storage

13. Epidemic Diffusion + Harvesting Meta-Data Req 1.Agent (Police) harvests Meta-Data from its neighbors 2.Nodes return all the meta-data they have collected so far Meta-Data Rep

14. Routing in the Vehicle Grid Mostly “proximity routing”; some long range routing also present Proactive routing (eg OLSR) –Does not scale to hundreds of 1,000’s On Demand routing (eg AODV) –AODV type flood search too costly Enter geo-routing –Most scalable (no state needed in routers) –GPS available; local coordinates used in blind areas (tunnels, parking lots, urban canyons) –Geo Location Service: distributed implementation

15. GLS (Geo Location Service) Equivalent to DNS to find geo addresses Maps vehicle ID (driver, VIN, license plate, etc) to the (more or less) current location Distributed implementation For resilience, dual implementation: –In the urban Internet infrastructure –In the wireless Vehicle Grid (to survive Infrastructure collapse)

16. Infrastructure based GLS: Overlay Location Service (OLS) Vehicular ID hashed into overlay proxies (like Chord P2P overlay) Mapping: Vehicular ID location

17. Georouting through the infrastructure IPv6 addressing (xy coordinates in header extension) How to make the system resilient to failures/attacks? –If access points fail, use GLS implemented in grid

18. Grid vs Infrastructure routing The trade offs: grid short paths vs fast wires Baseline: Shortest path routing –Short connections should go grid –Packets to remote destinations on infrastructure Next step: Access Points and Overlay assist in the decision –Propagation of congestion info from Overlay to wireless using 3 hop beaconing (say) every second

19. Simulation Experiments Wired link Car AP

20. Traffic pattern Car to access point: – APT fraction of traffic ( APT = 25%, 50%, 75%) Car to Car: –1- APT Traffic fraction Source/destination pair distance –Say, 40% of pairs < 300 m away –30% < 1km –30% < 10 km Total # of sessions: 200 UDP sessions with variable offered rate

21. Four routing strategies Set #1: all C2C connections are grid routed Set #2: all C2C connections are routed to the nearest AP’s Set #3 - shortest geo-distance routing Set #4 - same as #3, but now use also the load info advertised by AP’s

24. Wireless grid vs infrastructure load split as a function of path length

25. Summary  Transparent Geo routing across Infrastructure  Efficient Grid/Infrastructure load balancing  Simulation (Qualnet)  Analytic, multicommodity flow optimization  The “role” of the Internet  Geo Location Service Support  Load balance  V-Grid Congestion control  Future work: Infrastructure assisted Authentication; security; DoS protection

26. The case for geo-address IP address exacts very high maintenance cost –vehicles switch rapidly across different radio media (WiFI, WiMAX, 3G, Satellite, etc), changing IP address each time –Conventional Mobile IP and DHCP do not scale well to velocity; must re-register each IP address change –May group vehicles in mobile IP subnets - problems with merge/split

27. Geo Addressing Geo address is less disruptive: –Independent of radio medium –More predictable (motion prediction) –GPS is getting cheaper; efficient interpolation in GPS-blind spots (tunnels, urban canyons)

28. 8. Experiments (cont) Each cell may be 2-3 hop deep Assume each car knows the geo locations of all AP’s and destinations Conventional GPSR used in the grid between source/destination and to AP’s The ad hoc grid net dense enough so that all nodes are reachable from each other (No “voids”)

29. Simulation Experiments –5x5 grid with 25 cells total –One Access Point per cell –16 cars per cell –Free space propagation –No urban street constraints, we can assume transmissions are totally free space, unconstrained

32. 8. Result evaluation Should find that it is better to discriminate based on shortest path (kind of obvious) If load info available, shortest path + load combination will give near optimal solutions Review delays on grid only and on Infrastructure (possibly, real time UPD traffic should go only on AP’s)

33. 8. Result evaluation (cont) 1.Future: review TCP performance; comment on capture effects, improvements needed etc (if TCP performance is really bad, we may just skip and do only UDP) 2.Discuss congestion control hooks on UDP provided by the info received from the Overlay 1.This is quite open handed, so we will decide if we can do it only after all else is done