Intelligent Node Selection for Ad Hoc Networks-Assisted CBTC Systems Based on Proximal Policy Optimization

With the advancement of rail transit, communication-based train control (CBTC) system has attracted great attention. Due to the rapid movement of the train, traditional communication technologies cannot meet the requirements of safe handover between access points (AP) and trains. Meanwhile, the deployment of traditional networks requires the support of core network equipment and optical fibers, which is costly and complex. In this paper, flexible ad hoc network is applied as a novel technology in the train-to-train (T2T) communications. Wayside nodes are deployed to assist the situation where the distance between two trains exceeds the communication range of a single-hop distance. To ensure the throughput of the transmission link and minimize energy consumption, a multi-hop node selection strategy plays a vital role in ad hoc networks. Therefore, a proximal policy optimization (PPO)-based routing strategy is presented to deal with the node selection problem in different states. In the simulation, the results illustrate that the proposed node selection strategy is superior to other strategies in terms of optimizing energy and throughput.