Sparsely-structured Multiuser Detection for Large Massively Concurrent NOMA Systems

摘要: We study a new detection scheme for large-scale massively concurrent non-orthogonal multiple access (MC-NOMA) systems. The joint mutual user (MUD) is gradually transformed given its maximum likelihood (ML) formulation into sparsely-structured mixed-quadratic convex program. utilize to this extent an l 0 -norm of the ML followed by 1-bit hard quantization leading sparse signal representation which further relaxed via soft probabilistic quantization, approximation and fractional programming (FPG) generic yet tractable multiuser problem. resulting method solved efficiently proximal gradient descent has furthermore ability output decisions that can be subsequently integrated with modern coding schemes highly performant multi-user detectors. Numerical simulations demonstrate achievable performance proposed scheme, whereas reduced runtime complexity outlines practicality algorithm.