Research On Low Complexity Detection Algorithm For Uplink Massive MIMO System

The massive multiple input and multiple output(Massive MIMO)technology,which is one of the hot spots of 5G,has been receiving more and more attention in recent years.As a result of the large scaled antennas located in Massive MIMO system,it is an nondeterministic polynomial hard problem for the traditional detection algorithms to obtain the optimum bits error rate(BER)performance during low polynomial complexity time,which is not appropriate for the realistic detection.It is difficult to be solved,but is significant and urgent for realizing the communication operation in Massive MIMO system.Thus,the research on low complexity detection algorithm for uplink Massive MIMO system is carried out in this paper.The meta-heuristic algorithms,which belong to the kind of swarm intelligent technology,are of great potential to be exploited for signal detection in Massive MIMO system.Focusing on applying the living swarm intelligent methods to solve signal detection problem in Massive MIMO system,an artificial bee colony detection algorithm is proposed in this paper.The proposed algorithm detects uncoded signal in uplink Massive MIMO system,and is able to research the approximately optimum BER performance of maximum likelihood(ML)detection algorithm with a polynomial computational complexity.Besides,the proposed algorithm can also reach the theoretical optimum spectral efficiency with a lower average received signal to noise ratio(SNR).Due to poor BER performance,the exiting detection algorithms for Massive MIMO system are not applicable to detect high order modulation signals.To improve the detection performance of high order modulation signals in Massive MIMO system with limited computational complexity,a semidefinite further relaxation detection algorithm and a likelihood ascend search based semidefinite further relaxation detection algorithm are proposed for detecting high order quadrature amplitude modulation(QAM)signals in Massive MIMO system.With a low polynomial computational complexity,not only the optimum BER performance of the proposed algorithms can coincide well with that of the ML detection algorithm,but also the optimum spectral efficiency of the proposed algorithms converges to the theoretical one with a low average received SNR.The proposed algorithms are especially suitable for detecting high order modulation signals in large scale multiple antenna systems.In uncoded Massive MIMO system,when the number of operating transmitters is larger than the number of receiving antennas at the base station,the existing detection algorithms tend to fail since their detection performances are extremely poor.In this paper,this case is defined as overloaded scenario in uplink Massive MIMO system.The research on overloaded uplink Massive MIMO system is important and meaningful for busy-hour processing in 5G wireless communications.To solve detection problems in uncoded and overloaded uplink Massive MIMO system,a sparse overloaded detection model is constructed in this paper,and a compressed sensing based semidefinite relaxation detection algorithm and a compressed sensing based semidefinite relaxation for mixed Gibbs sampling detection algorithm are further proposed to detect combined MQAM signals.In overloaded scenarios,without knowing users' access state information at the base station,the proposed algorithms are able to approximately achieve the same BER performance as the cases of knowing users' access state information at the base station,when the average received SNR is in low to medium regime.Besides,under a limited polynomial computational complexity,the proposed algorithms can approximately have a perfect correct detection rate for the users' accessed sate,and obtain the theoretical spectral efficiency with a low average received SNR.Polar code,which is one of the advantage codes used in 5G,is the only kind of code that can be used to achieve Shannon boundary.Furthermore,its coding and decoding methods are simple and can be realized with a low computational complexity.However,the methods of applying polar codes in Massive MIMO system are seldom seen at present.To apply polar codes in Massive MIMO system and solve the corresponding coding,decoding,and detecting problems,a detection model is constructed for Polar coded uplink multiuser Massive MIMO system,and a successive cancellation based bitwise factor graph algorithm is proposed.This proposed algorithm can exhibit excellent BER performance and obtain high spectral efficiency with linear computational complexity,and is suitable for detecting as well as decoding Polar coded 4QAM signal in uplink Massive MIMO system.In addition,in this paper,a successive cancellation based semidefinite relaxation algorithm is further proposed.This proposed algorithm is suitable for detecting and decoding Polar coded 16 QAM and 64 QAM or other high order QAM signals in uplink Massive MIMO system.Moreover,it can obtain an excellent BER performance with limited polynomial computational complexity,and it is able to achieve the theoretical spectral efficiency with a low average received SNR.Note that,in this paper,it is pointed out that the Polar coded signals between users are non-orthogonal.For the overloaded cases,the latter proposed algorithm has good BER performance and high spectral efficiency when it is employed for detecting and decoding Polar coded signals,which gives another nice solution for busy-hour processing in 5G wireless communications.