Data-Driven Based Algorithm for Joint HetNets Power Allocation and NOMA User Pairing

Introducing non-orthogonal multiple access (NOMA) to heterogeneous networks (HetNets) improves its spectral efficiency (SE) for the next generation of mobile communication networks. Critical technical challenges on resource allocation in NOMA HetNets due to the inter-user interference of NOMA and the cross-tier interference that occurs in HetNets. The goal of this work is to enhance the NOMA HetNets spectrum efficiency by jointly optimizing both resource block (RB) allocation over NOMA users (i.e., user pairing (UP)) and power allocation (PA) over HetNets base stations (BSs). The formulated problem considers users’ quality of services (QoS) and successive interference cancellation (SIC) constraint, which contribute to a non-convex optimization problem. A linear model-free based technique, data-driven (DD), is applied as an optimization tool to efficiently solve the original NP-hard problem for a near-optimum solution. Our results show that HetNets and NOMA aided with DD for joint UP and PA satisfy the high data rate required and improve the system performance compared to other conventional schemes.