Dynamic Multi-User Detection Scheme Based on CVA-SSAOMP Algorithm in Uplink Grant-Free NOMA

• 1、Key Laboratory of Computational Intelligence and Signal Processing Ministry of Education, Anhui University, Hefei 230039

Abstract：In the uplink grant-free non-orthogonal multiple access (NOMA) scenario, since active user at the sender has a structured sparsity transmission characteristic, the compressive sensing recovery algorithm is initially applied to the joint detection of active user and the transmitted data. However, the existing compressive sensing recovery algorithms with unknown sparsity often require noise power or signal-to-noise ratio (SNR) as the priori conditions, which greatly reduces the algorithm adaptability in multi-user detection. Therefore, an algorithm based on cross validation aided structured sparsity adaptive orthogonal matching pursuit (CVA-SSAOMP) is proposed to realize multi-user detection in dynamic change communication scenario of channel state information (CSI). The proposed algorithm transforms the structured sparsity model into a block sparsity model, and without the priori conditions above, the cross validation method in the field of statistics and machine learning is used to adaptively estimate the sparsity of active user through the residual update of cross validation. The simulation results show that, compared with the traditional orthogonal matching pursuit (OMP) algorithm, subspace pursuit (SP) algorithm and cross validation aided block sparsity adaptive subspace pursuit (CVA-BSASP) algorithm, the proposed algorithm can effectively improve the accurate estimation of the sparsity of active user and the performance of system bit error ratio (BER), and has the advantage of low-complexity.

Keywords： non-orthogonal multiple access multi-user detection cross validation structured sparsity orthogonal matching pursuit