Grouped Intersection-based Routing using Reinforcement Learning for Urban VANETs

Under the rapid upgrowth of internet of vehicles (IoV), routing in vehicular ad-hoc networks (VANETs) has gained a great amount of attention in the past few years in academic and industry communities. Due to the complexity of urban territory and the scale of vehicular mobility, infrastructure resources are widely used in VANETs to improve network performance. We propose a grouped intersection-assisted routing protocol in VANETs using the Q-learning algorithm for an urban environment. The simulation results show our method can dramatically decrease the communication complexity of the learning procedure and improve the convergence speed compared to the conventional Q-learning algorithm without grouping.