A Reinforcement Learning-Based Routing Protocol for Clustered EV-VANET

Abstract

Vehicular Ad hoc Network for electric vehicles (EV-VANET) is an information and communication network composed of electric vehicles (EVs), charging stations and power grid. Due to the mobility and sparse distribution of electric vehicles, there exist many problems such as unstable link connection and routing void, so the reliability of communication between nodes cannot be guaranteed. In this paper, we propose a reinforcement learning-based routing protocol for clustered EV-V ANET (RLRC), where the network is divided into some clusters by employing the improved K-Harmonic Means (KHM) algorithm to improve the stability of cluster structure. RLRC uses reinforcement learning (RL) to calculate the Q-Value and evaluate the future reward of a decision, in which the available bandwidth and relative EVs movement are taken into account to improve the reliability and efficiency of the route. The simulation results show the effectiveness of the proposed protocol.