Research on Data Distribution for VANET Based on Deep Reinforcement Learning

Abstract

In this paper, we propose a vehicular ad hoc network (VANET) data distribution scheme based on artificial intelligence software-defined network (SDN) controller. Our aim is to improve the throughput of VANET by selecting edge cluster head nodes and gateway cluster head nodes based on deep reinforcement learning. In the entire VANET network, SDN determines the edge cluster head through neural episodic control (NEC). The gateway cluster head is determined by Q-learning. Then the gateway cluster head nodes act as the role that download the data from road side units (RSU) and then transmit the data to vehicles which need these data. Numerical results have shown the proposed scheme can provide greater throughput in VANET compared with the existing scheme.