Research On Low-complexity Detection For SCMA

With the widespread utilization of LTE in commercial system,the research work of the 5th generation communication technology has become quite active in Industry-Academia-Research circle.The main target of the 5G technology is to improve the data rate in a large scale.In order to achieve this target,higher spectrum should be developed and spectrum efficiency should be improved.Therefore,5G technology group has promoted non-orthogonal multiple access(NOMA)to enhance spectrum efficiency.Sparse code multiple access(SCMA)is a classic NOMA technology,which can realize vast user accessing.The principle is to spread user's signals on sparse spreading sequence and then to superpose several user signals on the shared resource blocks.The research work about SCMA currently hosts many immature aspects.At present,most of research work focus on codebook design,contention access function,blind detection method,power allocation and so on.This thesis works on low complexity detection receiver design.The traditional receiver of SCMA takes advantage of message passing algorithm(MPA),which is an iterative non-linear near-optimal receiver.Compared with maximum likelihood(ML)algorithm,it decreases complexity evidently.But it still has a large complexity due to iterations and large amounts of exponential operations.So MPA is difficult to apply to Internet of Things(IoT).This thesis has proposed two kinds of low complexity receivers for SCMA system.One uses serial interference cancellation(SIC)strategy.The other is improved quasi-ML receiver based on QR decomposition and tree search algorithms.To match with SIC receiver,this thesis improves the spreading scheme.For improved spreading method,a complex signal is divided into real part and imaginary part.Two parts are located in different chips.SIC receiver divides the received signals into real part and imaginary part to solve user signals step by step.Single SIC receiver has a coarse demodulation result.So this thesis has proposed jointed SIC-MPA receiver.The SIC module provides an initial value and the MPA module further refines input demodulation result.Quasi-ML receiver based on QR decomposition and tree search receiver has been used in MIMO multi-user detection.Conventional quasi-ML receiver is not available for SCMA system,because the number of received signals is larger than users in NMIMO,but less in SCMA.Therefore,this thesis improves tree search method.Compared with MPA receiver,newly proposed receivers have lower complexity and small performance loss.In order to guarantee the correctness of the algorithm,this thesis uses the rigorous theoretical derivation,also excutes simulations through MATLAB.In addition,this thesis compares newly proposed receivers with MPA receiver in terms of complexity and performance.