PLC for In-Vehicle Network: A DRL-Based Algorithm of Diversity Combination of OFDM Subcarriers

Abstract

For low latency communication service of vehicles, it is critical to improve the delay performance of power line communication (PLC) for in-vehicle network, which can decrease the weight and cost of the vehicle. In order to minimize the total time slots used in a transmission task, an orthogonal frequency-division multiplexing (OFDM) subcarrier diversity combination algorithm of PLC based on the deep reinforcement learning (DRL) is proposed herein. The short packet communication theory is used to develop an optimal combination model with constraints on short packet reliability, transmitting power and the amount of data. The state, action, and reward function of double deep Q-learning network (DDQN) are defined, and diversity combination for OFDM subcarriers is performed using DDQN. An adaptive power allocation algorithm based on the thresholds of error rate and the data amount is used. Simulation results show that the proposed algorithm can effectively improve the delay performance of PLC under the constraints of power and data amount.