Efficient MIMO Detection Algorithm For 5G Wireless Communication

Massive multiple-input multiple-output(MIMO)has been widely researched by plenty of researchers as one of the most significant technologies of the 5th generation mobile communication technology(5G).To obtain better performance both on link quality and date rate,there are usually dozens or even hundreds of antennas at both transmitter and receiver.The great number of transmitting and receiving antennas not only bring advantages,but also ask for higher requirements for complexity.Hence,it is very important to study low-complexity and high-performance MIMO detection algorithms.Additionally,to improve the channel capacity of MIMO systems,some researcher proposed the iterative detection and decoding(IDD)receiver based on the original separate detection and decoding(SDD)receiver.This paper has done research on MIMO detection problem and efficient IDD receiver in the context of massive MIMO and IDD receiver.Main contents and contributions are as follows.First,the system model and IDD receiver are introduced,and polar codes are also presented because we choose polar codes as channel encoding method.In the above model,this paper has analysed low-complexity belief propagation(BP)detection algorithm and summarized predecessors’ methods of lowing the complexity of BP detection.Guassian approximate inference aided BP(GAI-BP)detection as the research object,since GAI-BP detection could adapt well for high-order high-dimension MIMO systems.Afterwards,we optimize the message updating schedule of GAI-BP detection and propose layered BP(LBP).Further,we use node selection(NS)method to reduce the complexity of GAI-BP detection and propose NS-GAI-LBP detection at last.Simulation results illustrate that NSGAI-LBP detection accelerates the convergence and reduce the complexity nearly without performance degradation in comparison to GAI-BP detection.Subsequently,this paper has researched on the method of improving the performance of BP detection and summarized predecessors’ ideas on this topic.Ideally,The performance of BP detection should be similar to the performance of maximum likelihood(ML)detection which is the widely recognized optimal algorithm.Unfortunately,there are numerous loops in the factor graph(FG)of BP detection.Hence,it is impossible for BP detection to calculate precise posterior probability distribution.Fortunately,some scholars has proposed the method to reform the channel matrix by QR decomposition.Based on this method,they attempted dividing regions and passing messages between regions to eliminate loops in the FG.But this method is unfriendly to massive MIMO systems for the high complexity of QR decomposition.Additionally,error floor phenomena limits the performance of GAI-BP detection in high signal-to-noise ratio(SNR)region.In order to alleviate this phenomena,we propose forced convergence(FC)method for GAI-BP detection to avoid ”partial” trapping sets.Hence,the proposed FC-GAI-BP detection outperforms GAI-BP detection.Finally,we study on how to design an efficient iterative MIMO detection and polar decoding receiver.Afterwards,a module which can transform estimated symbol means into bit soft messages is very significant for hard-output MIMO detectors,because soft messages should flow between the MIMO detector and the polar decoder in the IDD receiver.And this module includes a lot of searching operations and transforming operations between binary codes and Gray codes.In order to reduce the complexity of this module,we propose a kind of treestructured Gray codes.Simulation results show that with the aid of proposed tree-structured Gray codes,the complexity of this module is reduced without performance degradation.