Intersection-based Traffic-Aware Routing with Fuzzy Q-learning for Urban VANETs

In the urban vehicular ad hoc networks (VANETs), in addition to the frequent change of network topology, the movements of vehicles conform to the road distribution and the real-time traffic condition, and tall buildings may obstruct signal propagation and cause the rapid fading of signals. Due to the specific characteristics, it is a particularly tricky task to design an efficient routing protocol for urban VANETs. In this paper, by combining the advantages of traffic aware routing and fuzzy Q-learning algorithm, we propose an intersection-based traffic aware routing protocol with fuzzy Q-learning (ITAR-FQ) for urban VANETs. In ITAR-FQ, an optimal routing strategy is proposed based on collected traffic results and Q-learning algorithm, which consists of different strategies for forwarding data packets within roads and intersections, respectively. Moreover, fuzzy logic algorithm is employed to enhance the accuracy of Q-learning with a joint consideration of multiple metrics including Euclidean distance, link lifetime, link quality and available bandwidth. The simulation results show that the proposed protocol significantly improve the performance in terms of average packet delivery ratio without sacrificing too much average packet delivery delay.