A Novel User Pairing in Downlink Non-Orthogonal Multiple Access

In this paper, we study the user pairing in downlink non-orthogonal multiple access (NOMA) networks. Users are divided into multiple 2-user clusters. In each cluster, the two users share the same transmit source, like time slot and spectrum, and the base station (BS) distributes the total transmit power to them by following certain rules. The performance of NOMA depends greatly on how we group them and how the power is allocated. In this paper, we propose two user pairing algorithms based on neighbor search methods, specifically the hill climbing and simulated annealing. Different from most existing methods, they don't have to work jointly with a certain power allocation scheme. Moreover, simulation results show that they significantly reduce the complexity compared with exhaustive search and can achieve a near-optimal solution.