Reinforcement learning-based channel access mechanism for multi-base station slotted Aloha with cooperative reception

Abstract

With the increasingly dense deployment of base stations in the internet of things (IoT), the importance of interference management becomes ever more pronounced. In IoT environments, devices often employ random access, connecting to channels in a distributed manner. In scenarios involving massive numbers of devices, severe interference may arise between nodes, leading to significant degradation in the throughput performance of the network. To address interference control issues in networks with random access, a multi-base station slotted Aloha network based on cooperative reception was considered, the reinforcement learning techniques was leveraged to design adaptive transmission algorithms that effectively managed interference, optimized network throughput performance, and enhanced network fairness. Firstly, an adaptive transmission algorithm were devised based on Q-learning, which was verified to maintain high network throughput performance under varying traffic conditions through simulation. Secondly, to improve network fairness, the penalty function method was employed to refine the adaptive transmission algorithm. Simulations confirm that the fairness-optimized algorithm significantly enhances network fairness while preserving satisfactory network throughput performance.

Keywords

• reinforcement learning;internet of things;random access;multi-base station network;slotted Aloha