Study Of SIC Reception Scheme In MIMO-NOMA System

With the rapid growth of mobile users’ demands for wireless services,Non-orthogonal Multiple Access(NOMA)scheme can be used to ensure the sustainability of mobile services,allowing multiple users to share the same time-frequency resources through superposition coding and successive interference cancellation(SIC).The combination of Multiple-Input Multiple-Output(MIMO)and NOMA is considered to be a multiple access technology that can increase system capacity and spectral efficiency.In MIMO-NOMA system,interference cancellation and signal detection are the key technologies to improve spectrum efficiency and guarantee system performance.However,there are some limitations in practical applications of SIC reception approach,such as the problem of error propagation and the sensitivity of SIC receiver to the accuracy of channel state information(CSI).To address these two issues,the SIC reception approach in downlink MIMO-NOMA system is investigated in the thesis from the perspectives of conventional communication signal processing and deep learning signal processing respectively.The main work is as follows:1.Aiming at the problem of error propagation in the SIC reception,two error propagation models are established firstly,and the influence of SIC error propagation on the performance of MIMO-NOMA system is analyzed by taking the Model 2 as a reference.And then,a SIC detection algorithm that proactively introduces the interference terms is proposed,adding residual signal terms generated by the previous user due to incomplete decoding at each update of the detection weight matrix and reconstructing the detection weight matrix,so that the impact of residual signals on subsequent user signal detection is further reduced.The proposed algorithm is suitable for both minimum mean square error(MMSE)and zero forcing(ZF)criterion.Finally,the system simulations verify that the proposed SIC detection algorithm can effectively improve the bit error ratio(BER)performance at the receiver when using the two criteria.2.In order to solve the problem that it is difficult to obtain accurate channel state information in conventional communication signal detection,a deep learning-based SIC reception method for MIMO-NOMA signal is proposed,which only requires a single deep neural network(DNN)signal detection model to complete signal estimation and detection.Firstly,in the training process of the DNN model,suitable label data is designed to facilitate the signal recovery problem as a classification problem in the output layer.By minimising the loss function,a robust DNN model is obtained,and when performing the actual signal detection,MIMO-NOMA signal is fed into the DNN detection model and the signals of different users can be recovered at the same time,so that there is no error propagation.Finally,it is verified by numerical simulations that the DL-SIC receiver based on deep learning achieves better performance than the conventional SIC receiver,and with strong robustness,it is able to decode the user signal correctly with limited channel state information.