A Prediction-based Routing Algorithm for Vehicular Ad Hoc Networks

Some Inherent Vehicular Ad Hoc Networks (VANETs) characteristics, such as intermittent connectivity, highly dynamic topology, and hard delay constraints, make data communication a challenging task in these networks. Due to such peculiarities, in this work we study the impact of using vehicles predicted locations as a metric for data communication in VANETs. In our proposal, called LPRV (Localization Prediction-based Routing for VANETs), packet forwarding is performed by nodes with predicted future localization closer to the delivery destination, without the need for exchanging additional control message. The proposed algorithm also explores the knowledge of a digital map to limit the scope of message exchanges in the shortest path for vehicles between source and destination. We compared our proposal to both classic Flooding and SIFT (Simple Forwarding over Trajectory) algorithms. Our results clearly demonstrate the efficiency of the proposed solution in different scenarios, especially in terms of delivery rate, number of hops and delay, with a reduced number of message transmissions.