Research On Low Complexity Detection Algorithm In MIMO-SCMA System

With the development of communication technology,the demand for terminals' network accessing has also increased,but it is difficult for traditional communication systems to meet a large number of terminals' accessing requests.The emergence of the joint multiple-input multiple-output(MIMO)and sparse code multiple access(SCMA)technology can not only achieve higher spectrum efficiency,but also meet the demands of more terminals' accessing requests.For the traditional SCMA system detection algorithm,the message passing algorithm(MPA)can use the code sparsity of SCMA to reduce the excessive computation complexity in the detection process.However,the MIMO technology increases the modulation order and the number of users on each resource block,which greatly increases the decoding complexity and is not conducive for the hardware implementation.Based on this,the goal of this thesis is to effectively control the detection complexity of the receivers while guaranteeing the BER performance of MIMO-SCMA system.The main works of this thesis are as follows:1.The low-complexity and low-delay symbol detection algorithm in the uplink of MIMO-SCMA system under Rayleigh flat fading channel is studied.By improving the expectation propagation framework based on the spreading factor graph and introducing Orthogonal Triangle(QR)decomposition and clustering-based factor node(FN)processing,the convergence speed of the detection algorithm can be increased.Meanwhile,the message passing method between variable node(VN)and FN can also improve the efficiency of the algorithm iteration.Theoretic analyses and simulation results show that the complexity of improved algorithm is linear with the size of constellation and is irrelevant to the receiving antennas,and the convergence speed of improved algorithm is fast.2.The low-complexity and low-delay symbol detection algorithm in the downlink MIMO-SCMA system under Gaussian and Rayleigh flat fading channels is studied.A new index matrix is constructed to obtain a joint factor graph,and then virtual codebooks are constructed by using the joint factor graph information and channel state information.The transmitting codebooks are resumed by combining the MPA detection algorithm with factor graph and virtual codebook.Theoretic analysis and simulation results show that the improved joint MPA detection algorithm can significantly reduce complexity and has better detection performance compared with maximum likelihood(ML)and other similar detection algorithms.In addition,this thesis briefly analyzes the influence of the codebook constellation points' minimum Euclidean distance on detection signal.The analysis results show that the Euclidean distance of the constellation points can be increased by using different codebook models with different transmitting codebooks on the same antenna,which is beneficial to reduce the difficulty of signal detection.