Latency-Aware Server Deployment in Internet of Vehicles Based on Multi-Agent Reinforcement Learning

The performance of the servers directly affects the efficiency of the whole intelligent Internet of Vehicles (lo V) system. A reasonable server deployment strategy can effectively reduce network latency and improve the efficiency of IoV.Therefore, it is necessary to study the problem of server deployment in Io V.In this paper, we formulate a latency-aware server deployment problem based on Multi-Agent Reinforcement Learning (MARL) and utilize an Actor-Critic network based on Long Short-Term Memory (LSTM) encoding network to solve it. Besides, four deployment strategies (DQN, DRQN, Random and No Movement) are used as the baseline methods for performance comparison with our proposed method on two real-world datasets. The simulation results show our solution has excellent performance in reducing network latency under different scenarios.