| In wireless communication,channel fading interferes with the transmission of signals and prevents the receiver from obtaining accurate channel state information(CSI).With imperfect CSI,the classical nearest neighbor decoding rule(NNDR)is unable to effectively cope with the nonlinear relationship between the transmitting and receiving signals.To improve the information processing quality at the receiver,the recently proposed generalized nearest neighbor decoding rule(GNNDR)generalizes the form of NNDR by processing the channel output and scaling the channel input before searching for the nearest neighbor codeword.GNNDR can take full advantage of the nonlinear information between channel input signals and channel output signals in different nonideal scenarios,making it an effective receiver scheme for high capacity,low latency and high reliability data transmission.For real-time services,the slow fading channel model can capture the non-ergodic nature of delay-limited information transmission and thus becomes the fundamental model for the system design.For high-capacity demand services,the massive multipleinput multiple-output(MIMO)technique is a key technology to improve the spectral efficiency of the system through the deployment of low-power,high-spatial resolution antennas.Based on GNNDR design,this thesis optimizes the probability distribution of the achievable rate under slow fading channels,thus reducing the outage probability of the system,and improves the achievable rate of the downlink in massive MIMO system.The main work of this thesis is as follows:Under the single-input multiple-output slow fading channel,this thesis vectorizes scalar functions of GNNDR,then uses generalized mutual information(GMI)to evaluate the achievable rate with stochastic characteristics and establishes a function optimization model for minimizing the outage probability.Under the certain GNNDR function form,this thesis adjusts the probability distribution of GMI with finite number of receiving antennas by linearly shrinking the coefficients of CSI estimation expression under minimum mean square error(MMSE)estimation,which greatly reduces the outage probability of the system.When the number of receiving antennas grows without bound,this thesis demonstrates that no shrinkage coefficients are required at the receiver side,and the outage probability can be minimized using the MMSE estimator.Afterwards,this thesis extends the linear shrinkage idea directly to MIMO slow fading channels.Numerical experiments show that the proposed shrinkage design can achieve a signal-to-noise gain of more than 1 dB compared to the MMSE estimator in terms of reducing the system outage probability.In time division duplex(TDD)downlink achievable rate analysis of the massive MIMO system,existing statistical CSI processing and beam training methods are unable to fully utilize the nonlinear information of the system.Therefore,this thesis proposes to utilize the combination of GNNDR and successive interference cancellation(SIC)at the user terminal to extract information from joint observations.In the general block fading channel and channel aging cases,the achievable rates of the proposed scheme are analyzed in this thesis without downlink pilot and downlink orthogonal pilot-assisted transmission modes,respectively.Numerical experiments show that the proposed scheme is able to achieve up to 10%total information rate gain in the downlink orthogonal pilot-assisted transmission mode in block fading channels compared to the comparison methods. |
