Unsupervised Learning Based Resource Allocation for Low EMF NOMA Systems

In the 5G era, the densification of wireless infrastructure to fulfill ever‐increasing quality of service (QoS) needs, as well as the ever‐increasing number of wireless devices, may result in increased levels of electromagnetic field (EMF) exposure in the environment. The potential long‐term health impacts of EMF radiation are currently being debated and deserve consideration. As a result, we propose in this chapter a novel EMF‐aware resource allocation strategy based on power domain non‐orthogonal multiple access (PD‐NOMA) and machine learning (ML) technologies for lowering EMF exposure in cellular system uplinks. We employ the K‐means strategy (an unsupervised ML approach) to construct clusters of users to be allocated together, and then strategically organize and assign them on the subcarriers depending on their related channel attributes. Finding the optimal number of clusters in the PD‐NOMA environment is a critical challenge, and we utilized the elbow approach in conjunction with the $F$‐test method in this chapter to successfully manage the maximum number of users to be given at the same time per subcarrier. We have also derived an EMF‐aware power allocation by formulating and solving a convex optimization problem. Based on the simulation findings, our suggested ML‐based solution successfully minimizes EMF exposure when compared with current techniques.