Sparse Signature Matrix Optimization for NOMA under Imperfect CSI

Abstract

Non-orthogonal multiple access (NOMA) can support the rapid development of the Internet of Things (IoT) with its potential to support high spectral efficiency and massive connectivity. The low-density superposition modulation (LDSM) scheme is one of the NOMA schemes and uses the sparse signature matrix to reduce multiple access interferences (MAI). In order to improve the NOMA system performance in practice, this paper focuses on designing the sparse signature matrix with a large girth for LDSM under imperfect channel state information (CSI). Based on the orthogonal pilot and linear minimum mean square error (LMMSE) estimation, the LDSM optimized by bare-bone particle swarm optimization (BBPSO) algorithm has a larger girth and can gather more accurate information in the process of iterative decoding convergence. An extrinsic information transfer (EXIT) chart analysis is designed for the LDSM-OFDM system as a theoretical analysis tool. The simulation results show that the optimized LDSM outperforms the reference LDSM system, bringing about a 0.5 dB performance gain.