Performance Enhancement for IRS-Aided NOMA Networks with Underlay D2D Communications*

The co-channel interferences inherent in the non-orthogonal multiple access (NOMA) networks with underlay D2D communications drastically degrade the system performance. Intelligent reflecting surface (IRS) is able to mitigate the co-channel interferences through designing its reflected signals. In this work, an IRS-aided downlink NOMA system with underlay D2D communications is investigated. The expression of the total throughput of NOMA users and D2D terminal is analysed under the consideration of the co-channel interferences. An optimization problem is formulated to maximize the total throughput by joint optimizing the phase shifts of IRS, the transmitting power of all links as well as the decoding orders of NOMA users, under the minimum SINRs requirements. A novel dynamic programming approach based on Lagrange duality and genetic algorithm is developed to solve the problem. Finally, the simulation results demonstrate that the total throughput is significantly improved by utilizing IRS with a larger number of reflected elements.