Reinforcement Learning and Device‐to‐Device Communication for Low EMF Exposure

International mobile union (IMU) present new verticals for IMT‐2030 paradigm. The Beyond‐5G (B5G) is expected to support IMT‐2030 communication paradigm. The 3rd generation partnership project (3GPP) introduced narrowband internet‐of‐thing (NB‐IoT), a low power wide are (LPWA) technology to support ultra‐massive machine type communication (umMTC) with enhanced coverage. Enhanced ultra‐reliable and low latency communication (ERLLC) and massive channel access are the prime requirements of future generation communication. Device‐to‐device (D2D) communication, also known as proximity services (ProSe), is standardized for 5G and B5G network. D2D support network to transmit the uplink data in multi‐hop manner. In this study, we present machine learning (ML) enabled low power narrowband D2D communication relay selection procedure to augment reliability in terms of packet delivery ratio with less number of repetitions and minimize energy, which ultimately minimizes the electromagnetic field (EMF) exposure. The relay selection procedure is formulated as Markov decision process (MDP) and solved using reinforcement learning based $Q-Learning$ (QL) algorithm.