Deep Reinforcement Learning-based Multi-Channel Access for Industrial Wireless Networks with Dynamic Multi-User Priority

To address the high concurrent access of massive industrial devices with different QoS requirements for industrial wireless networks, a Deep Reinforcement Learning-based Dynamic Priority Multi-Channel Access (DRL-DPMCA) algorithm is proposed. According to the time-sensitivity of data, industrial devices are assigned with different priorities, based on which their channel access probabilities are adjusted. The Markov decision process is utilized to model the above problem. Because of the explosion of state space, DRL is used to map from states to actions. The long-term cumulative reward is maximized to obtain an effective policy. A compound reward with access reward and priority reward is designed. An experience replay with experience-weight is proposed. GRU, dueling architecture and step-by-step epsilon-greedy method are employed. Experiments show that, comparing with slotted-Aloha and DQN, DRL-DPMCA converges quickly, and guarantees the highest channel access probability and the minimum queuing delay for high-priority industrial devices in the context of minimum access conflicts.