University of Southern California NetPD Lab of Department of Electrical Engineering The USC Andrew and Erna Viterbi School of Engineering USC
Site Home

Project Title: METRANS: Efficient routing for safety applications in vehicular networks

Funding: METRANS Transportation Center

PI: Konstantinos Psounis
Duration: Oct. 2007 - Dec. 2008


Vehicular ad hoc networks have a number of real-life applications. Examples include increasing road safety by reducing the number of accidents as well as reducing their impact in case of non-avoidable accidents, improving local traffic flow and efficiency of road traffic, and offering comfort and business applications to driver and passengers. While there has been a significant effort to define applications, there are still some hard technical challenges that need to be resolved. Perhaps the hardest of them all is how to achieve communication in an environment where network nodes (vehicles) move so fast that the very concept of a wireless link between two nodes is meaningless for time scales larger than a few seconds, and where the density of the nodes can vary drastically in space and time, making the network intermittently connected.

To address this challenge, we use a new approach of routing that is tailored to the needs of vehicular ad hoc networks. We refer to it as "mobility-assisted routing". This routing approach departs drastically from the traditional view of networking: When a node (moving vehicle or a static roadside station) wants to send a message to one or more nodes (vehicles), it may transmit a number of copies of the message to one or more distinct relay nodes. Each relay will carry the message further, and may transmit it to a new, better relay or directly to a destination.

Flooding-based schemes are simplest examples of routing schemes belonging to these routing paradigm. However, flooding-based schemes create so much contention for the wireless channel that their performance, in practice, is quite bad. We propose a new routing approach that significantly reduces the overhead of flooding, while achieving good performance. The idea is to distribute only a bounded number of copies to a number of relay-vehicles, each of which can then deliver it to the destination or to a new, better relay-vehicle. The important questions we address include: (i) Which vehicles should receive each message? (ii) How many copies should the routing protocol use to ensure fast and efficient message delivery? (iii) How should these copies be distributed? (iv) How should each copy be routed towards the destination? (v) Are the resulting schemes scalable? (vi) Can these schemes take advantage of installing static roadside stations at intersections and at regular distances along highways?