Multi-agent deep reinforcement learning based resource management in SWIPT enabled cellular networks with H2H/M2M co-existence

Machine-to-Machine (M2M) communication is crucial in developing Internet of Things (IoT). As it well known that cellular networks have been considered as the primary infrastructure for M2M communications, there are several key issues to be addressed order deploy communications over networks. Notably, rapid growth traffic dramatically increases energy consumption, degrades performance existing Human-to-Human (H2H) traffic. Sustainable operation technology and resource management efficacious ways solving these issues. In this paper, we investigate a problem with H2H/M2M coexistence. First, considering energy-constrained nature machine type devices (MTCDs), propose novel network model enabled by simultaneous wireless information power transfer (SWIPT), which empowers MTCDs ability simultaneously perform harvesting (EH) decoding. Given diverse characteristics IoT devices, subdivide into critical tolerable types, further formulating an efficiency (EE) maximization under divers Quality-of-Service (QoS) constraints. Then, develop multi-agent deep reinforcement learning (DRL) based scheme solve problem. It provides optimal spectrum, transmit splitting (PS) ratio allocation policies, along efficient training designed behaviour-tracking state space common reward function. Finally, verify reasonable mechanism, multiple agents successfully work cooperatively distributed way, resulting outperforms other intelligence approaches terms convergence speed meeting EE QoS requirements.