Research On Multiuser Detection Of Non-orthogonal Multiple Access On SCMA System

With the development of communication technology,the 5G era of interconnection of everything has arrived.5G defines three main application scenarios,but it can no longer meet the diverse application requirements of Io T.In actual requirements,it is often the combination of several features in the three scenarios.Therefore,Huawei proposes the concept of 5.5G and expands three new scenarios.An application scenario is added between e MBB and m MTC,named UCBC,to compensate for the insufficient uplink broadband transmission capability of m MTC.Compared with mMTC,UCBC provides more flexible configuration requirements.In addition,UCBC supports diversified communication requirements.For example,typical intelligent manufacturing communication scenarios,workshop communication scenarios and mass sensor access scenarios all require mass terminal access,but have different application requirements for reliability and end-to-end delay.This paper mainly studies the two scenarios,namely,the mass terminal access scenario of delay insensitive and high reliability or the scenario of delay sensitive and low reliability requirements.The access of massive devices and terminals leads to the demand for massive data communication,which cannot be met by traditional OMA technology.NOMA technology is back in the spotlight as it offers higher spectrum utilization than OMA.SCMA technology is a NOMA technology based on multi-dimensional modulation and sparse code spread spectrum,which can not only improve spectrum efficiency,but also provide greater performance advantages compared with other NOMA technologies.The advantages of SCMA in bit error rate performance are mainly derived from the forming gain brought by multi-dimensional constellation mapping and the MPA algorithm which provides near-optimal multi-user detection performance.Therefore,the bit error rate performance of the system can be improved through the design of the parent constellation of the codebook and the study of the multi-user detection algorithm of the receiver,which are the two main research contents of this paper to achieve high reliability communication.However,for communication with low delay,the complexity of MPA algorithm based on iterative structure is still too high even if SCMA code word has sparsity.One of the research contents of this paper is to reduce the computational complexity of multi-user detection algorithm and shorten the processing time of multi-user detection.Aiming at the mass terminal communication scenario requiring high reliability,this paper proposes a design method of SCMA high-dimensional parent constellation diagram with low BER based on lattice theory.In order to maximize the forming gain of high-dimensional constellation boundary and improve the bit error rate performance of SCMA system,the high-dimensional parent constellation diagram is designed by using the characteristic of repeated grid pattern of lattices.Taking the geometry of the parent constellation as a constraint,The problem of maximizing the high-dimensional parent constellation CFM can be transformed into a series of combinatorial optimization problems.A simplified solution method is proposed.The simulation results show that the proposed design method is not only suitable for constellation diagrams with lower dimension and fewer constellation points,but also can design high-dimensional parent constellation diagrams with larger dimension,more constellation points and lower bit error rate.At the receiving end,the SCMA iterative multi-user receiver is improved to improve the multi-user detection performance,reduce the error rate of SCMA system,and further improve the system performance.In this paper,a highly reliable SCMA iterative multi-user receiver with channel metrics is proposed,which can improve the performance of the iterative receiver by correcting the overestimated external information during iteration.JS divergence is introduced to measure the similarity of probability distribution among transmitted information,and EXIT graph is used to compare the change of mutual information,so as to analyze the influence of introduced channel metric information on the performance of SCMA iterative multi-user receiver.An iterative multi-user detection algorithm based on JS divergence value as termination condition is proposed.The user information that has reached the set threshold is fixed as output information and no longer participates in the iterative process.The computational complexity of multi-user detection algorithm of SCMA iterative receiver is further reduced without affecting the performance.In order to meet the requirement of low delay communication in UCBC scenario,this paper studies the low-complexity multi-user detection algorithm of SCMA system.Deep learning technology is used to complete low-complexity multi-user detection.On the premise of ensuring the performance,the computational complexity of SCMA multi-user detection algorithm is reduced so as to reduce the signal processing delay.The SCMA signal detection algorithm based on deep learning proposed in this paper can automatically adjust the layers and nodes of the neural network structure according to the number of users and frequency resources,and adjust the weight parameters of the edge between nodes of the neural network through off-line training.SCMA signal detection is realized without iteration,which greatly reduces the computational complexity of signal detection algorithm compared with MPA algorithm.Moreover,since there is no iterative process,girth is no longer a factor that degrades signal detection performance.