Research On Multi-User Detection For Asynchronous Non-Orthogonal Multiple Access Systems

With the development of wireless communications and Internet technology,supporting the Internet of Thing has become one major task for future communications systems.Accordingly,the need for data traffic exhibits explosive growth,and users' demand for Quality-ofService also increases dramatically.These demands bring out a new technical challenge,i.e.,the massive connectivity from users and devices should be supported for ensuring the communication requirements of high reliability and low latency.The existing fourth generation(4G)system fails to solve this urgent technical challenge,which motivates the research on the fifth generation(5G)communication system.Non-orthogonal multiple access(NOMA)has been proposed as a promising technology in 5G communication system,which can be used to construct a new multiple access communication network with high reliability and low latency.NOMA has been adopted as a key technology in the 5G communication protocol.Compared with the traditional orthogonal multiple access(OMA)technology,NOMA supports more users and devices by sharing the same spectrum resources.Although the NOMA technology has developed rapidly in recent years,the existing works mainly consider the synchronous NOMA system,and whereas the asynchronous NOMA(A-NOMA)system still has many unsolved problems.This thesis studies the multi-user detection algorithms and parameter optimization for the A-NOMA system,and provides effective solutions for the multi-user detection in the A-NOMA system.The specific contents are presented as follows.Firstly,a low-complexity message passing(MP)detection algorithm is proposed for symbol asynchronous NOMA(SA-NOMA)system with equal power,whose detection accuracy is superior to the counterpart under ordinary NOMA systems.In the asynchronous transmission,the receiver employs a filter that matches the users' time offsets.Due to this filter,the interference between users becomes more complex,i.e.,users' information is interfered by other users' data from not only the current time slot,but also the adjacent time slots.To represent the relationship between the output of the matched filter and the symbols of users' in different time slots,a three-dimensional factor graph is constructed.Based on the representation of the factor graph,a cross-slot MP detection algorithm is proposed.The analytical results prove that asynchronous transmission can effectively mitigate multi-user interference,and the modulation waveform affects the MP detection-based performance.The simulation results show that the SA-NOMA system based on MP detection algorithm can achieve better bit error rate(BER)performance than the NOMA system.Secondly,the linear minimum mean square error(LMMSE)algorithm and partial message passing(P-MP)algorithm are proposed for the symbol-asynchronous NOMA system with imperfectly time offsets.When the variance of estimation error is smaller than 0.1,the SA-NOMA system exhibits better tolerance against interference and noise than the NOMA system.The LMMSE detection algorithm is a multi-user detection algorithm based on iteration,and it consists of the LMMSE detector and single-user decoder.The LMMSE detector treats the estimation error as noise,and the interference from adjacent time slots is treated as the interference brought by the memory channel.The LMMSE algorithm needs to calculate matrix inverse,so its complexity is relatively high.To further reduce the complexity of detection algorithm,a P-MP detection algorithm based on the directed factor graph is proposed,and we also analyze the effect of estimation error on the detection algorithm.The simulation results show that when the variance of estimation error is 0.1,the SA-NOMA system with imperfectly estimated time offsets can achieve the same or even better BER performance than the NOMA system.Thirdly,the extrinsic information transfer(EXIT)chart is proposed to analyze the statistical characteristics of time offsets on the system performance for the SA-NOMA system,and the time offsets are optimized to improve the tolerance of SA-NOMA system against multi-user interference.The EXIT chart can effectively track the flow of mutual information in the process of message passing.Therefore,the EXIT function of sum nodes under the MP detection algorithm is given for the SA-NOMA system.It is found that the EXIT function of sum node is related to the time offsets.Then we consider the effect of random time offsets on the EXIT function of sum nodes from the statistical perspective.For the two-user SA-NOMA system,the time offset is optimized by maximizing the EXIT function of sum node.The results show that the optimized time offset can effectively improve the BER performance of the SA-NOMA system.In addition,the detection threshold of the SANOMA system with repeat accumulate(RA)code serially concatenated with repetition code can approach the capacity.Finally,we propose the distance enumeration function for the two-user frame asynchronous NOMA(FA-NOMA)system,and provide the analytical result for the word error rate(WER)performance under finite code length,as well as the design of channel coding parameters.In the FA-NOMA system,the receiver receives a superposition information from two frame asynchronous users.The misplaced superposition operation is defined as misplaced superposition(MS)coding,and the distance between two MS codewords is defined as a misplaced Euclidian distance.The two-user codewords set for the FA-NOMA system is composed of two single-user code serially concatenated with a MS code.The misplaced distance enumerator for the FA-NOMA system is obtained based on the distance analysis method for serially concatenated codes.Then the WER union bound of FA-NOMA system with finite code length is obtained by average misplaced distance enumerator.The simulation result of WER for the two-user FA-NOMA system is shown to be consistent with the union bound.Then,the asymptotic misplaced distance enumerator is given,and we take the repetition codes and random codes as examples to explain the usage of the asymptotic misplaced distance enumerator.The analysis of asymptotic misplaced distance enumerator provides the design scheme of channel coding parameters for the FA-NOMA system.