1. 1 TBD: Trajectory-Based Data Forwarding for Light-Traffic Vehicular Networks IEEE ICDCS’09, Montreal, Quebec, Canada Jaehoon Jeong, Shuo Gu, Yu Gu, Tian He and David Du Computer Science and Engineering University of Minnesota {jjeong,sguo,yugu,tianhe,du}@cs.umn.edu June 23rd, 2009

2. Motivation 2  The vehicular networking is getting a hot research topic.  Internet Access, Driving Safety, Data Dissemination, etc.  The environments for the vehicular networks  Every vehicle has a DSRC device for wireless communication.  Every vehicle has a GPS-based navigation system for driving information.  The Internet Access Points (APs) are sparsely deployed in road networks.  The objective in this paper  The vehicles can deliver their packets to APs through the multi-hop forwarding with the help of other vehicles.

3. Problem Definition 3 Next hop? Carrier-1’s Moving Trajectory Carrier-2’s Moving Trajectory Light-Traffic Road Network Forwarding Path Delay-1 Delay-2 Delay-1 < Delay-2

4. Problem Definition 4 Road Network with Unbalanced Traffic Density Light Traffic Path Heavy Traffic Path Next hop? Forwarding Path Delay-1 Delay-2 Delay-1 > Delay-2

5. Contribution and Challenges 5  Contribution  Data forwarding based on Vehicle Trajectory With vehicle trajectory, TBD outperforms the existing scheme (VADD, Infocom’06) only using vehicular traffic density.  Challenges  A more accurate link delay model than VADD’s Mathematical model for the link delay  End-to-End delay model based on vehicle trajectory E2E delay modeling based on (i) vehicular traffic density and (ii) individual vehicle trajectory

6. Link Delay Model 6  Objective  To compute the expected link delay over a one-way road segment.  Road Segment for Link Delay Model  Given the vehicle arrival rate and the vehicle speed, How to compute the Forwarding Distance ?

7. Forwarding Distance for Vehicle Arrivals 7 Forwarding Distance over Time Forwarding Distance Vehicle arrives at time.

8. Forwarding Distance for Vehicle Arrivals 8 Forwarding Distance

9. Link Delay Model Comparison between VADD and TBD 9  VADD Link Delay Model  Given the vehicle arrival rate and the vehicle speed, the forwarding distance is the sum of the network components.  This model is inaccurate since it misses the following fact Only the first network component can be used for data forwarding.  Performance Comparison  For Average Forwarding Distance, TBD is much closer to the Simulation result than VADD.

10. E2E Delay Model 10  Objective  To compute the expected end-to-end delay from a Vehicle to an Internet Access Point (AP).  Road Network Graph for Data Forwarding Given a vehicle’s trajectory, how to compute the E2E delay? Since node1 and node2 have different trajectories, their E2E delays are different.

11.  :Expected Delivery Delay (EDD) at Intersection 1  Where a packet is sent towards Intersection 2. 11 Expected Delivery Delay at Intersection (VADD Model) Link Delay for Road Segment (1,2)

12. Expected Delivery Delay at Intersection (VADD Model) 12  Average Forwarding Probability ( )  The probability that a packet at intersection i can be delivered towards neighboring intersection j.  We consider all the possible moving directions of the current packet carrier at intersection i.  How to compute ? Packet Delivery Direction Moving Direction-3 Moving Direction-1 Moving Direction-2

13.  Limitation of EDD at Intersection  The vehicle trajectory is not used to compute the EDD.  Node1 and Node2 have the same EDD regardless of their different trajectories.  Thus, we cannot determine which node is a better next carrier.  How to involve the vehicle trajectory into EDD computation?  The main idea is to divide the delivery process recursively into two steps: 1. The packet forwarding process at the current carrier. 2. The packet carry process by the current carrier. 13 Expected Delivery Delay at Intersection (VADD Model) They are very close to each other.

14. Expected Delivery Delay for Vehicle Trajectory (TBD Model) 14  Vehicle Trajectory:  Case 1: The packet is forwarded at intersection 1.  Case 2: The packet is carried to intersection 2 and is forwarded at intersection 2.  Case 3: The packet is carried to intersection 3 and is forwarded at intersection 3.

15. 15 Expected Delivery Delay for Vehicle Trajectory

16. Forwarding Protocol 16  TBD Forwarding Rule  Within a connected component, packets are forwarded to the vehicle with a minimum EDD.

17. Performance Evaluation 17  Evaluation Setting  Performance Metric: Average Delivery Delay  Parameters: (i) Vehicular traffic density, (ii) Vehicle speed, and (iii) Vehicle speed deviation.  Simulation Environments  36-intersection road network (4.2 miles X 3.7 miles)  Vehicle mobility model: Random-Way Point  Vehicle speed distribution: N(40,5) MPH  Communication range: 200 meters  Time-To-Live (TTL): infinite (i.e., no timeout)

18. Average Delivery Delay Comparison between TBD and VADD 18 TBD outperforms VADD under the light traffic, such 20~50 vehicles. As the traffic density increases, two schemes are converged.

19. Conclusion 19  In this talk, the data forwarding scheme called TBD is introduced based on the vehicle trajectory:  Data Forwarding from Vehicle to AP.  Also, the link delay model is introduced for TBD data forwarding scheme:  This link delay model can be used for other VANET routing or forwarding schemes.  As future work, the multiple-hop Internet access will be investigated in the vehicular networks:  Vehicle trajectory will be used for the data forwarding for the Internet access.

20. 20 Target Point Future Work: Reverse Data Forwarding for Internet Access

21. Challenge in Reverse Data Forwarding 21  As packet destination, the vehicle is moving, not static.  The packet from AP needs to be delivered to the vehicle, considering the rendezvous point along the vehicle trajectory. Target Point Target Missing! The reason of the target missing?  Inaccurate estimation of the vehicle arrival How to provide this reverse forwarding?