Can mobility predictions be compatible with cooperative active safety for VANET?

Cooperative Active Safety applications for VANET require an up-to-date knowledge of a vehicle's immediate surrounding (awareness) obtained by all vehicles broadcasting their status information (position, speed). Periodically transmitted, it leads to wireless congestion, while adapting the rate of its transmission to some predicted motions impacts the accuracy of this knowledge and the reliability of the safety application. In this paper, we investigate whether a trade-off can be found to this critical and challenging issue. As initial answer, we propose an enhanced particle filter prediction model based on bio-inspired glow-worms clustering capabilities. It is designed to react quicker to sudden traffic changes typically found in traffic safety situations (such as sudden breaks or sudden direction changes) to provide a high surrounding knowledge precision with less required transmissions. Simulations conducted with calibrated urban traffic in Bologna showed that we can ensure on the one hand a suitable adaptation of the channel load, and on the other hand, a high precision of awareness prediction and a capacity to detect traffic changes adapted to VANET traffic safety applications.