Research On Key Technologies Of Low-Latency Uplink Non-orthogonal Multiple Access

The fifth generation(5G)mobile networks have been expected to improve significantly in several key performance indicators,in which low latency is urgently required by numerous emerging services in the Internet of Things(IoT),such as,telesurgery,augmented reality,tactile Internet,and vehicle-to-vehicle.The non-orthogonal multiple access(NOMA)technology,which provides high spectrum efficiency,low latency,and high access density,has been widely regarded as one promising technology to enable 5G.NOMA can allow multiple users to transmit signals in the same frequency and time radio resource simultaneously,and can distinguish the signals of users at the receiver with advanced multi-user detection(MUD).At present,there are several candidate NOMA techniques proposed from academia and industry,such as,power-domain NOMA(PD-NOMA),pattern division multiple access(PDMA),and sparse code multiple access(SCMA).To ensure low latency communications with reliable,NOMA can be improved with multiple-input and multiple-output(MIMO),advanced modulation and coding,and full-duplex(FD).Generally speaking,the non-orthogonal superposition of signals from access users and grant-free scheduling can reduce the access latency;MIMO and FD can shorten the transmission latency by increasing spectrum efficiency;advanced modulation and coding as well as low-complexity MUD can decrease the processing latency.This thesis studies the design and enhancement of NOMA to guarantee low latency in the uplink.To adapt to the characteristics of IoT services,the impact of massive accessed users,a few transmit antennas,shadow fading,and imperfect channel state information(CSI)on the uplink NOMA is studied.Meanwhile,the effectiveness of the proposed schemes is analyzed and evaluated by utilizing the novel finite blocklength(FBL)information theory,thereby complying with the small packet size of IoT services.The main contributions of this thesis are summarized as follows.1.The random interleaver enhanced PDMA(RIePDMA),a novel multi-carrier NOMA,is proposed by bringing the random interleaver into the uplink PDMA.With the proposed iterative detection and decoding based on belief propagation and interference cancellation(BP-IDD-IC),the overload and reliability can be further improved.Simulation results show that RIePDMA and BP-IDD-IC can achieve higher reliability without markedly increasing complexity.At the same time,self-adaptive BP-IDD-IC can shorten the processing latency with a fewer number of iterations,especially in the high signal-to-noise ratio(SNR)region.2.The rate splitting algorithms and successive interference cancellation(SIC)detection in multi-user MIMO(MU-MIMO)NOMA are proposed,to minimize the maximum transmission latency of users.First,this thesis derives the achievable data rate of the minimum mean square error SIC(MMSE-SIC)and maximal-ratio combining SIC(MRC-SIC)detection,and designs two rate splitting algorithms matching with MMSE-SIC and MRC-SIC,respectively.Further,to reduce the complexity and latency of detection,group SIC and two-layer MRC-SIC detection are supported to sharply reduce the number of SIC operations.Numerical results validate that the rate splitting MU-MIMO NOMA can efficiently shorten the transmission latency and processing latency.3.The SCMA enhanced FD(FD-SCMA)scheme is first proposed,to support the simultaneous uplink and downlink short-packet transmissions.Utilizing the FBL information theory,this thesis derives the error probability of FD-SCMA under a given transmission latency with imperfect self-interference suppression of FD.FD-SCMA is theoretically proved to be able to ensure lower transmission latency than existing SCMA and FD schemes in the time-invariant flat-fading channel.Besides,in the time-invariant frequency-selective fading channel,this thesis derives the upper bounds for error probability in FD-SCMA,and validates the low transmission latency of FD-SCMA through the theoretical calculation and Monte Carlo simulation.4.The low latency transmission of the emerging massive MU-MIMO NOMA with perfect and imperfect CSI is studied.Assuming the users are uniformly and randomly deployed and facing log-normal shadow fading,this thesis derives the probability density function of effective SNRs,and uses the FBL information theory to calculate the error probability under a given transmission latency.Further,this thesis can optimize the length of pilots to minimize the error probability with the golden section search method.Numerical results verify that the massive MU-MIMO NOMA can support low latency transmissions of massive accessed users,even though the users are encountering moderate shadow fading.Overall,the advanced MUD enabled NOMA can be enhanced by the interleaver,MIMO,and FD technologies,and can remarkably lower latency from three aspects,which are access latency,transmission latency,and processing latency.Therefore,low latency short-packet transmissions can be guaranteed efficiently.