Deep Reinforcement Learning Based Massive Access Management for Ultra-Reliable Low-Latency Communications

With the rapid deployment of the Internet of Things (IoT), fifth-generation (5G) and beyond 5G networks are required to support massive access of a huge number of devices over limited radio spectrum radio. In wireless networks, different devices have various quality-of-service (QoS) requirements, ranging from ultra-reliable low latency communications (URLLC) to high transmission data rates. In this context, we present a joint energy-efficient subchannel assignment and power control approach to manage massive access requests while maximizing network energy efficiency (EE) and guaranteeing different QoS requirements. The latency constraint is transformed into a data rate constraint which makes the optimization problem tractable before modelling it as a multi-agent reinforcement learning problem. A distributed cooperative massive access approach based on deep reinforcement learning (DRL) is proposed to address the problem while meeting both reliability and latency constraints on URLLC services in massive access scenario. In addition, transfer learning and cooperative learning mechanisms are employed to enable communication links to work cooperatively in a distributed manner, which enhances the network performance and access success probability. Simulation results clearly show that the proposed distributed cooperative learning approach outperforms other existing approaches in terms of meeting EE and improving the transmission success probability in massive access scenario.