Novel user clustering schemes for downlink NOMA system

Abstract

In this paper, we propose novel user clustering schemes for downlink non-orthogonal multiple access (NOMA) system, where an N-antenna base station (BS) selects and serves 2N users from K single antenna user equipments (UEs) (K ≥ 2N and K is even). In particular, we propose a signal difference and alignment (SDA) framework to achieve fair and spectral efficient user clustering for NOMA with flexible performance-complexity tradeoff. To be more specific, the BS sorts UEs according to their large-scale fading (LSF) gains and selects N primary UEs with the highest LSF gains to guide the downlink zero-forcing beamforming (ZFBF) vector. Then the BS selects N complementary UEs from the second half of the sorted UE set based on the principle of signal space alignment (SSA). Unlike conventional user scheduling, SDA framework uses the ordered set to create signal strength difference for NOMA while employing opportunistic SSA method to select complementary UEs with minimal inter-cluster interference leakage. Moreover, the SDA framework can enable distributed implementation with graceful performance at lower system overhead. Numerical results verify that the proposed schemes improve the performance in downlink NOMA system.