Multi-User Code Design And Detection For Non-Orthogonal Multiple Access

Recently,as Internet,Internet of things,and intelligent wireless devices have become more mature and popular,the communication demands have rapidly dvolved form original aim-plex voice services to multi-category and high-quality multimedia services,shile promoting the rapid development of conesponding communication technologies.Considering that the orthogonal multiple accesstechnology in 4G wire lesssystmcannot support the low-latency,high-reliability,and massive-access communication demands,which impels the emergence of a new aceess technology,namely Non-Orthogonal Multiple Access(NOMA).NOMA enables the massive access by allowing multiple users to sheme the same spectrum resouces.Companed with the orthogonal multiple access that allocates different spectrum resources to each user,NOMA can achieve higher spectral efficiency and lower lateney.To make full use of space resources,NOMA and multiple-input multiple-output(MIMO)are combined,termed as MIMO-MOMA,which can achieve higher transmission rate and more reliable performance.Therefore,MIMO-MOMA is one of the key technologies in the fuiure 5G wireless system.Althougy MIMO-MOMA can satisfy the huge communication demands,there are still many unsolved problems.This paper deeply studies the design of multi-user coding scheme and signal detection over MIMO-NOMA,which provides effective solutions for the iimplemen-tation of MIMO-NOMA.The specific contents are given as follows.Firstly,this paper designs a simple Pautially-Repeated Spatially-CoupledLow-Density Parity-Check(PR-SCLDPD)coding scheme for NOMA systems.The PR-SCLDPC code is con-structed bg selecting a pont of bits of an SC-LDPC code to be nepeated.Since the decoding increment is only a log-like lihood ratio(LLR)combination,compared with the original SC-LDPC code,the decoding complexity scarcely increases.It is shown that this LLR com-bination is effective to overeome the multi-userinterference.By optimizing the ratio ofcoded bits for repetition,the decoding threshold of the PR-SCL DPC code can be close to the capacity of NOMA system.Experimental results that the PR-SCLDPC codes can achieve better performances than the graph-extended and parallelly connected SC-LDPC codes over the NOMA systems.Secondly,many existing works of MIMO-MOMA systems focus on the oretical analysis or the design of one part(either transmitter or neceiver).To the best of our knowledge,thereate no works on ooded MIMO-MOMA systems simultaneously taking low-complexity im-plementation and performance optimization into consideration.Therefore,this paper pro?poses a practical coded MIMO-NOMA system with capacity-approaching performance as well as low implementation complexity.Specifically,the employed receiver consists of Lin-ear Minimum Mean-Square Error(LMMSE)detector and message-passing decoders,which can decompose the optimal recovery of signal vector into many distributed low-complexity calculations.Then,an asymptotic analysis is proposed to trace the statistical characteristic of the employed receiver.By exploiting this asymptotic analysis,a kind of sparse-graph based multi-user channel codes is designed,such that the code can match with the LMMSE detector.As a result,the proposed MIMO-NOMA system can achieve a decoding threshold within 0.2 dB from the theoretical capacity.Simulation results validate the reliability and robustness of the proposed MIMO-NOMA system in practical settings,including different channel loads,finite code lengths,channel fading conditions,and imperfect channel esti-mations,i.e.,the gaps between bit error ratio(BER)performances at 10-5 of the proposed system and capacities are within 2 dB.In contrast to precoded MIMO-NOMA,the pro-posed system does not need instantaneous channel feedback.Compared to MIMO-NOMA with single-user coding,the proposed multi-user coded system achieves considerable per-formance gain.Coupled with its low complexity,the proposed system can be an attractive solution for the MIMO-NOMA uplink in the 5G communications.To address the issue of poor robustness of multiuser coding schemes in existing MIMO-NOMA systems,a simplified design method for multiuser codes is proposed.Generally,multi-user codes require to be redesigned when the number of users,transmit antennas,or receive antennas change.Therefore,the previous design methods of multi-user codes are too complicated to be applied to massive-access MIMO-NOMA systems.To solve this problem,at first,a simple Multi-User Detector(MUD)is proposed to detect the signal for each user by regarding the superimposed signal from the other users as interference.Then,based on extrinsic information transfer analysis for the MUD,this paper proposes three criteria to simplify the code design,which copes with the changes of user number and antenna configuration.In addition,when there are many users in the systems,each user requires to employ a low-rate multi-user code to overcome severe multi-user interference.Based on the proposed criteria for simplified code design,this paper proposes a low-rate Multi-User Irregular Repeat Accumulate(MU-IRA)code.Experimental results confirm the validity of proposed three criteria,i.e.,the gaps between BER performances at 10-5 of the well-designed MU-RA codes and the capacities are within 2 dB.plementation and performance optimization into consideration.Therefore,this paper pro-poses a practical coded MIMO-NOMA system with capacity-approaching performance as well as low implementation complexity.Specifically,the employed receiver consists of Lin-ear Minimum Mean-Square Error(LMMSE)detector and message-passing decoders,which can decompose the optimal recovery of signal vector into many distributed low-complexity calculations.Then,an asymptotic analysis is proposed to trace the statistical characteristic of the employed receiver.By exploiting this asymptotic analysis,a kind of sparse-graph based multi-user channel codes is designed,such that the code can match with the LMMSE detector.As a result,the proposed MIMO-NOMA system can achieve a decoding threshold within 0.2 dB from the theoretical capacity.Simulation results validate the reliability and robustness of the proposed MIMO-NOMA system in practical settings,including different channel loads,finite code lengths,channel fading conditions,and imperfect channel esti-mations,i.e.,the gaps between bit error ratio(BER)performances at 10-5 of the proposed system and capacities are within 2 dB.In contrast to precoded MIMO-NOMA,the pro-posed system does not need instantaneous channel feedback.Compared to MIMO-NOMA with single-user coding,the proposed multi-user coded system achieves considerable per-formance gain.Coupled with its low complexity,the proposed system can be an attractive solution for the MIMO-NOMA uplink in the 5G communications.Finally,in existing MIMO-NOMA systems,the base station is assumed to achieve activa-tion states of the users accurately through the complicated grant procedure between the base station and the users.However,when there are a large number of potential active users in the system and the activation state of each user may change at any time,the frequen-t interaction authorization process between the base station and the users will result in a huge latency for the communication system.Therefore,in order to satisfy the low-latency and high-throughput demands of massive-access communication systems,this paper studies a random-access distributed MIMO-NOMA system based on grant-free protocol.Then,a low-complexity message passing algorithm is designed,which detects both the activation s-tates and signals of all users.Based on the corresponding factor graph of the channel matrix,we design message update functions of sum and variable nodes to obtain a message passing detection algorithm,where the updated messages are associated with the activation states and transmitted signals of users.Experimental results verify that proposed message pass-ing detection algorithm achieves the linear complexity with user number and near-optimal performances under different sparsified channels.