A Study Of Signal Detection And Iterative Receive Technology In MIMO Systems

In Multiple-Input Multiple-Output (MIMO) and MIMO-Orthogonal Frequency Division Multiplexing (MIMO-OFDM) systems, receivers usually need to solve the problems of high dimensional signal detection and iterative parameter estimation, which leads to prohibitively high complexity. The iterative parameter estimation technique can obtain asymptotically optimal performance and reduce the complexity of joint parameter estimation. On the purpose of pursuing good tradeoff between performance and complexity, the signal detection in V-BLAST (Vertical-Bell Labs lAyered Space-Time) systems and the iterative parameter estimation in MIMO-OFDM systems are studied respectively. The main contributions of the dissertation are summarized as follows:1. By exploiting the minimum mean square error based sorted QR decomposition (MMSE-SQRD) of the channel matrix, a sphere constraint QRD-M algorithm and a sphere constraint Stack algorithm were proposed for V-BLAST systems. Accordingly, the strategies of updating sphere radius were developed for the proposed two algorithms. The newly proposed algorithms not only outperform the conventional algorithms, but also dramatically reduce the computational complexity. Moreover, they can achieve almost the same performance as the SE sphere decoding (SE-SD) algorithm while visiting much fewer nodes in the tree-search procedure.2. Base on the maximum likelihood (ML) metric and the maximum a posteriori (MAP) metric, a breadth-first low complexity list detection (BrF-LCLD) algorithm and a reduced complexity QRD-M (RC-QRD-M) algorithm were presented for coded V-BLAST systems. Furthermore, the original sorted QR decomposition (SQRD) algorithm of the channel matrix was modified to improve performance. Compared with the ML and MAP QRD-M algorithm, the proposed algorithms efficiently reduce the computational complexity at the price of slight performance loss. In addition, they are quite suitable to practical implementation and maintain fixed computational complexity.3. Using the Variational Bayesian EM (VBEM) algorithm, an asymptotically optimal Bayesian iterative receiver with joint channel estimation and signal detection/decoding was firstly derived for MIMO-OFDM systems. And then, a space-frequency domain (SFD) combined recursive variational Bayesian channel estimation (SFD-RVBCE) algorithm was proposed for the novel VBEM iterative receiver. By making some reasonable approximations, a frequency domain RVBCE (FD-RVBCE) algorithm was finally obtained. Theoretical analysis and simulation results demonstrate that the proposed receiver not only achieves near optimal performance, but also shows better performance and lower computational complexity than the existing VBEM iterative receiver.4. Based on the decomposed signal model, several soft-decision-directed recursive channel tracking algorithms were developed to reduce complexity. Based on the original signal model, the corresponding simplified algorithms were derived subsequently. The proposed algorithms completely avoid matrix inversion by performing channel prediction on time domain and recursive channel estimation on frequency domain. Theoretical analysis and simulation results show that they can significantly reduce the computational complexity, and approach the performance of the excellent Turbo-Kalman and Turbo-RLS algorithms. In addition, they can be applied to different environment after small modifications due to their common recursive tracking framework, therefore are of benefit to hardware implementation.