Section: New Results

VANET

Participants : Enrico Natalizio, Thierry Delot.

Today, thanks to vehicular networks, drivers may receive useful information produced or relayed by neighboring sensors or vehicles (e.g., the location of an available parking space, of a traffic congestion, etc.). In [33] , we address the problem of providing assistance to the driver when no recent information has been received on his/her vehicle. Therefore, we present a cooperative scheme to aggregate, store and exchange these events in order to have an history of past events. This scheme is based on a dedicated spatio-temporal aggregation structure using Flajolet-Martin sketches and deployed on each vehicle. Contrary to existing approaches considering data aggregation in vehicular networks, our main goal here is not to save network bandwidth but rather to extract useful knowledge from previous observations. In this paper, we present our aggregation data structure, the associated exchange protocol and a set of experiments showing the effectiveness of our proposal.

In [36] , we present a novel vehicular communication protocol, which aims to reduce the effect of broadcast storm problem in VANETs (Vehcular AdHoc NETworks). When the traffic density is above a certain value (e.g., when vehicles are in congested traffic scenarios), one of the most serious problems is the increase of packet collisions and medium contentions among vehicles which attempt to communicate. Our proposed technique, namely Selective Reliable Broadcast protocol (SRB), is intended to limit the number of packet transmissions, by means of opportunistically selecting neighboring nodes, acting as relay nodes. As a result, the number of forwarder vehicles is strongly reduced, while network performance is preserved. SRB belongs to the class of broadcast protocols, and exploits the traditional vehicular partitioning behavior to select forwarders. Each cluster is automatically detected as a zone of interest, whenever a vehicle is approaching, and packets will be forwarded only to selected vehicles, opportunistically elected as cluster-heads. In respect of traditional broadcast approaches, the main strengths of SRB are the efficiency of detecting clusters and selecting forwarders in a fast way, in order to limit the broadcast storm problem. Simulation results have been carried out both in urban and highway scenarios, in order to validate the effectiveness of SRB, in terms of cluster detection and reduction of number of selected forwarders.