Reinforcement Learning-Based Adaptive Modulation for Vehicular Communication

Abstract

Vehicular communications allowed the vehicles connected to traffic lights, road signs, or building infrastructures. Adaptive modulation enabled the system to maintain the quality level as the vehicles moved through different environments. In this paper, QPSK, 8PSK, and 16-QAM modulation schemes are utilized for low, medium, and high noise environments with the AWGN channel model. Reinforcement learning is implemented in the V2I scheme with an epsilon-greedy algorithm to decide which modulation scheme should be used based on the environment condition. The simulation showed promising results with most selected modulation scheme in low noise environment is 16-QAM, in medium noise is 8PSK, and in high noise is QPSK.