Low Complexity Decoding Algorithm Of Uplink In SCMA System

With the rapid development of the Internet of Things and the Internet,the application field of 5G has further expanded.Massive devices and machine-to-machine(M2M)communication will be one of the important indicators when designing the system.To address the need for massive machine connections in the fifth generation of mobile communication networks,Huawei has proposed a new air interface core technology-Sparse Code Multiple Access(SCMA).SCMA improves the system's overload rate by using low-density spread spectrum in multiple access,so that under the same frequency resources,it can carry more user signals.The receiver is an important part of the SCMA system.MPA is used for multi-user detection.Compared with MAP,MPA has similar performance and greatly reduced complexity.However,MPA still suffers from poor convergence,large redundant operations during the iteration process,etc,resulting in a huge overall computational complexity,which is not suitable for multi-user detection at the receiving end of the SCMA system.If SCMA becomes a new air interface technology,the multi-user detection at the receiving end must meet the requirements of low latency,high reliability,and low complexity.This thesis proposes three multi-user detection schemes from different aspects:(1)Set different Euclidean distance thresholds at different SNR according to the different SNR,calculate the Euclidean distances between all superposed constellation points and the received signals,and compare the results with the thresholds.The superposed constellation points whose Euclidean distance is less than the threshold value are used in the MPA iterative operation at the receiving end.By discarding unnecessary superposed constellation points during the MPA iteration process,redundant operations during the iteration process are reduced,and the purpose of reducing the computational complexity is achieved.(2)Group users on the same resource into one group,classify user groups based on channel gain,and prioritize user groups on resources with large channel gain.When a user appears in another group,the user directly reads the detection result of the superior.This solution avoids the same user repeatedly detecting on different resources,thereby reducing the computational complexity.(3)The power is sorted according to the channel coefficients of users on different resources,the larger the transmission power is allocated on the channel with the larger power coefficient,and the transmission power is reduced on the resources with smaller channel coefficients.Choose to detect user codewords on resources with large channel coefficients.By improving the detection advantage on resources with large channel coefficients,iterative convergence during detection is accelerated,thereby reducing the overall detection complexity.