Successive Gaussian Approximation Based BP Detection for New Radio Multiple Access

Abstract

In this paper, we present a multi-user detection technique using successive Gaussian approximation based belief propagation (SGA-BP) detection with linear pre-filtering for the new radio multiple access (NR-MA) uplink communications. The NR-MA system represents the codebook-based, the sequence-based schemes, or the interleaver/scrambler-based non-orthogonal multiple access (NOMA) schemes. The SGA-BP detection approximates the interference-plus-noise term as a Gaussian random variable and performs iterative SGA-BP, where messages are updated successively during each iteration. By doing so, messages can be delivered efficiently and accurately with low computational complexity. For validation of the approach, two of the NR-MA schemes, i.e. the codebook based NOMA and the sequence based NOMA, are studied to evaluate the performance of SGA-BP detection. Simulation results show that SGA-BP detection achieves better diversity gain and error floor in BER performance than the GA-BP detection or the MMSE-SIC detection, especially when the number of antennas grows. We also evaluate the cases of high overloading of users, and verify the potential of SGA-BP detection with linear pre- filtering.