An Effective Broadcast Scheme for Alert Message Propagation in Vehicular Ad Hoc Networks

1. An Effective Broadcast Scheme for Alert Message Propagationin Vehicular Ad Hoc Networks E. Fasolo, A. Zanella and M. Zorzi Department of Information Engineering University of Padova{fasoloel, zanella, zorzi}@dei.unipd.it Speaker Stefano Tomasin June, 14th 2006.

2. Main aims of this study • To develop and improve a broadcast protocol to deliver alert messages as soon as possible in a vehicular scenario [3] • Maximize the reliability • Minimize the delivery latency • To propose an analytical model in order • To evaluate the protocol performance • To optimize the protocol parameters • Compare the proposed solution with other broadcast protocols by means of simulations

3. Smart Broadcast Protocol (SBP) • Main Features • Position-based scheme • Using only position information • Running on the top of an IEEE 802.11-like system • Completely distributed • Absence of control traffic • System Model • Long and narrow rectangular area (street) • Nodes placed according to a Poisson distribution • Nodes known own position

4. The current source coverage area is split into n sub areas At each sub area is associated a time interval named contention windows according to SBP: Initial Assumptions AIM: Support the maximum advancement of the broadcast message towards the propagation direction Not considered area Sn … S1 Positive advancement towards the propagation direction Propagation direction

5. SBP: Relay Election • The source sends the an RTB (Request To Broadcast) message • Each node that receives correctly the RTB message • Determine the sector it belongs to • Schedules the retransmission of a CTB (Clear To Broadcast)message after abackoff time bselected in the contention window according to a uniform probability distribution function • Listen to the channel • Source send the message to the node which has transmitted the CTB message NEXT RELAY bn2 bj1 b11 Not considered area Sn … S1 bn1 b12 Propagation direction

6. SBP: Collision Resolution NEXT RELAY • If a node receives correctly another RTB message adjusts its backoff time according to the position of the new source • If a collision occurs ( two or more nodes select the same backoff time and send at the same time a CTB message) the procedure continue. b11 bn2 Not considered area Sn bj1 bn1 … b11 S1 A COLLISION OCCURS Propagation direction

7. Theoretical Analysis: Initial Assumptions • qh = the number of nodes that pick the same backoff value h in W • qh = 0  IDLE (I)  No node transmits • qh > 0  COLLISION (C) A collision occurs • qh = 1  BROADCAST (B)  A node wins the contention and transmits the broadcast message • nU= number of unsuccessful events before the completion of the procedure • TU = average duration of an unsuccessful countdown step

8. Theoretical Analysis: One-hop latency • The one-hop latency is defined as the mean time required before the broadcast message is successfully forwarded to the next relay node where the last terms accounts for the extra time spent to restart the procedure is negligible • Finally, we have the simplify equation for the one-hop latency where K = TC / TI

9. The one–hop message progress, δ, is defined as the additional distance covered by the message in a rebroadcast phase, on average We only need to determine the statistic of J: We evaluate the conditioned probability that s = h, given that s in W  Ps(h) And we use Ps(h) to evaluate Pj(r) and the the mean value of J Theoretical Analysis: One-hop message progress

10. Theoretical Analysis: Optimization • Consider the following cost function defined as the time required to get the success retransmission over the successful probability COST FUNCTION If we fix Ns Single solution in [1/K, 1]

11. Validation of the theoretical analysis • One hop latency at varying of node density for different cw (assuming Ns = 10) • Lines refer to the theoretical results • Marks refer to the simulation outcomes • The interpolation of the minimum values correspond to the line obtained with cwopt

12. Analytical Model versus Simulations • Average one–hop progress δ and propagation speed vversus the node density λ (assuming cw = cwopt) Good matching between analytical model and simulations High node densities assure the maximum progress

13. Protocol comparison SB is compared with MCDS-based, GeRaF and UMB From the figure we can observe that the propagation speed achieved by SB is almost constant when varying the node density SB may lead to a slightly lower advancement than the other schemes. This is due to the fact that SB balances both the message progress and the latency.

14. Conclusions • We designed a protocol for message dissemination which guarantee high reliability and low latency • The developed analytical model assures good matching with the simulation results • SB outperforms the other message dissemination mechanisms. • Future work • 1 • 2 • 3

15. References • [1] D. Cottingham, “Research Directions on Inter-vehicle Communication,” http://www.cl.cam.ac.uk/users/dnc25/references.html, Dec. 2004. • [2] M. Rudack, M. Meincke, K. Jobmann, and M. Lott, “On traffic dynamical aspects intervehicle communication (IVC),” in 57th IEEE Semiannual Vehicular Technology Conference (VTC03 Spring), Jeju, South Korea, Apr. 2003, http://portal.acm.org/citation.cfm?id=778434. • [3] Fasolo, E. and Furiato, R. and Zanella, A., “Smart Broadcast for inter–vheicular communications,” in Proc. of WPMC05, Sep. 2005. • [4] Zanella, A. and Pierobon, G. and Merlin, S., “On the limiting performance of broadcast algorithms over unidimensional ad-hoc radio networks,” in Proceedings of WPMC04, Abano Terme, Padova, Sep. 2004. • [5] Korkmaz, G. and Ekici, E. and O¨ zgu¨ner, F. and O¨ zgu¨ner, U¨ ., “Urban multi-hop broadcast protocol for inter–vehicle communication systems,” in Proc. of the first ACM workshop on Vehicular ad hoc networks , 2004. • [6] M. Zorzi and R. Rao, “Geographic Random Forwarding (GeRaF) for ad hoc and sensor networks: energy and latency performance,” IEEE Transaction on Mobile Computing, vol. 2, no. 4, Oct.–Dec. 2003. • [7] B. Williams and T. Camp, “Comparison of broadcasting techniques for mobile ad hoc networks,” in MOBIHOC, 2002. • [8] K.M. Alzoubi and P.J. Wan and O. Frieder, “New distributed algorithm for connected dominating set in wireless ad hoc networks,” in Proc. Of 35th Hawaii Int’l Conf. on System Sciences (HICSS-35), Jan. 2002. • [9] P.J. Wan and K. Alzoubi and O. Frieder, “Distributed construction of connected dominating set in wireless ad hoc networks,” in Proc. of IEEE INFOCOM’2002, June 2002. • [10] S. Giordano and I. Stojmenovic, Position based routing algorithms for ad hoc networks: a taxonomy. Kluwer, 2004, pp. 103–136. • [11] I. Stojmenovic, “Position-based routing in ad hoc networks,” IEEE Communications Magazine, vol. 40, no. 7, pp. 128–134, July 2002.

16. An Effective Broadcast Scheme for Alert Message Propagationin Vehicular Ad Hoc Networks E. Fasolo, A. Zanella and M. Zorzi Department of Information Engineering University of Padova{fasoloel, zanella, zorzi}@dei.unipd.it Speaker Stefano Tomasin June, 14th 2006.

21. Applications and services Emergency notification Cooperative driving assistance Car to car audio/video communications Internet access Traffic control Topical features No energy constraints High mobility Availability of timing and localization information Main Issues New paradigm (physical, MAC, routing layer solutions) New broadcast propagation mechanisms Efficient Reliable Low latency Inter-vehicular networks (IVNs)

22. The Broadcast Storm Problem • Flooding • High Data Redundancy • Collision Problem • MCDS-based algorithms • Minimize the retransmitting node number • Solve the collision problem • Not feasible in high dynamic networks

23. Broadcast Protocol Overview • Probabilistic Schemes • Not solve collision and redundancy problem • Neighbor-based Schemes • Require control traffic, depend on the network topology • Topology-based Schemes • More efficient but require a complete topology knowledge (not feasible for high dynamic networks) • Cluster-based Schemes • High cost to maintain clustering structure in mobile networks • Position-based Schemes • Flat, not require control traffic • Urban Multi-hop Protocol (UMBP)

24. Some theoretical observations • The time wasted during the re-broadcast procedure depends on • The collision probability • The probability that the furthest sub-areas are empty • Fixed Ns, for each node density, there is an optimum contention window size such that • The time wasted on re-broadcast procedure is minimized