Reinforcement Learning Based Reliable Route Selection for Internet of Vehicles

Abstract

Abstract Self-driving cars have been receiving much attention recently, and communication problems between vehicles have become an issue. Due to frequent topology changes, routing problems occur in communication between vehicles. This is an important issue in the VANET (Vehicular Ad-hoc Network) and several papers have been presented to address this issue. However, existing papers are routing protocols that can resolve issues after they occur or only under certain circumstances, such as urban. Therefore, it is necessary to select the optimal relay nodes according to the circumstances surrounding the agent to ensure optimal performance at all times. For this purpose, this paper proposes RLSR (Reinforcement Learning based Selective Route Selection) algorithm that selects relay nodes through reinforcement learning. The algorithm proposed in this paper can ensure reliability by selecting the best relay nodes in any situation.