Bio-inspired approach for improving performance of information routing in Vehicular Ad Hoc Networks

Sammanfattning: As a key technology for Intelligent Transportation Systems (ITS), vehicle-2-vehicle (V2V) communication enables the interconnection of vehicles to limit emissions, decrease congestion, increase traffic safety and much more. Vehicular Ad-hoc Networks (VANETs), a subset of Mobile Ad-hoc Networks (MANETs) is a peerto-peer network among vehicles that is utilized to route information through different forwarding algorithms. Utilizing VANETs it is possible to share information out of communication range without building expensive infrastructure. The underlying technology is standardized by Institute of Electrical and Electronics Engineers (IEEE) in the US and by the European Telecommunications Standards Institute (ETSI) in Europe. One of the current problems in VANETs is the poor performance of traditional MANET routing algorithms when applied to VANETs due to its high speeds, rapid variation in link connectivity and extremely varied density of vehicular nodes in the network. Therefore there exists a need for a better routing algorithm for VANETs which can guarantee a sufficient end-to-end delay and delivery performance. This thesis seeks to evaluate a bio-inspired approach for improving performance of information routing in VANETs. Specifically, a swarm-based routing algorithm based on Particle Swarm Optimization (PSO) is used to optimize the Next Forwarding Vehicle (NFV) selection in the routing process. The PSO consists of a fitness function that includes position and mobility information of neighbouring vehicles. The fitness function is optimized with a Genetic Algorithm that determines the optimal weights of the different components. The weights are determined from three scenarios on which they are tested and compared to Greedy Forwarding (GF) and Contention-Based Forwarding (CBF). The results show that PSO performs better than GF in terms of both end-to-end delay and packet delivery ratio (PDR). CBF achieves better PDR than PSO but at the cost of larger end-to-end delay. The PSO shows potential as an information routing algorithm but the high variation in its results indicates that refinements are necessary