Optimizing Semi-Persistent Scheduling in Vehicular Communications: A Comparative Study of Reinforcement Learning Techniques

This paper presents the implementation and comparison of various reinforcement learning (RL) techniques for optimizing Semi-Persistent Scheduling (SPS) in Sidelink vehicular communication environments. The methods compared include Heuristic-based, Q-Learning, Monte Carlo, Deep Q-Network (DQN), and Actor-Critic. Metrics such as Packet Reception Ratio (PRR), Throughput, Average Delay, Average Reselections, and Average Conflicts are evaluated to determine the effectiveness of each technique.