Researches On Key Techniques Of Signal Detection And Channel Estimation In MIMO OFDM Systems

Multiple-input multiple-output (MIMO) can be used in orthogonal frequency division multiplexing (OFDM) systems to not only offer the potential of high spectral efficiency and the promising to exploit maximum spatial resources, but also lessen the severe effects of frequency-selective fading. Therefore, MIMO-OFDM is the one of key techniques of the next genernation mobile communications system.We firstly survey the research status of signal detection and channel estimation for MIMO-OFDM systems, list a variety of concentrated problems and fundamental questions, and summarize the main research contents of this thesis.We propose a novel near maximum-likelihood (ML) Chase detection scheme for MIMO systems in the second part of this thesis. The scheme is to reduce the dimension of the detection problem dynamically by performing early detection on some sufficiently reliable symbols based on the signal-to-interference-plus noise ratios of their linear estimates, and produce a list of candidate symbol vectors based on a heuristic reliability measure defined in terms of the bit log-likelihood ratio (LLR) of the linear estimate. And then we derive a low preprocessing complexity expression for calculating the pre-layer bit LLR. Our simulation results show that the proposed detector can offer more attractive performance-complexity tradeoffs than the recently proposed detector.Again, we propose a low complexity near-ML Vertical Bell Labs Layered Space-Time (V-BLAST) algorithm to improve the poor performance of tranditional V-BLAST algorithm. In the first detected layer, the algorithm construct a list of LLR-based candidate error vectors to increase the diversity order and decease the probability of error propagation, and then apply parallelizable reduced-complexity sub-detectors to the remaining layers after cancelling the contribution of each candidate vector of the list. Simulation results show that the proposed method can approach near ML performance but complexity is the same to that of V-BLAST, and become a viable technique for real-time high-date rate applications. Next, we derive the baseband signal model of distributed antenna systems (DAS) in a matrix form, and propose a subspace-based blind channel estimation method, where one side of the link constitutes a largely spaced antenna array. The approach gives estimations of all channel responses by minimizing some quadratic form built on the orthogonality between channel and noise subspaces, which does not necessarily require that the number of receive antennas should be no less than that of transmit antennas. And then we derive the close-form equation of variance marix and mean square error, and prove the proposed method is an optimal channel estimation one whose performance can close the minimum constrains Cramer-Rao Bound (CRB).Finally, we propose an iterative channel estimation approach based on the parametric channel modeling and inter-carrier interference (ICI) mitigation algorithm for MIMO-OFDM systems operating on frequency-selective fading channels. Since wideband transmission results in a sparse multipath channel, the iterative channel estimation employs lowpass filter in the time domain to lessen the noise components continuously, while ICI mitigation use frequency filter to reduce the interference impact. Simulation results demonstrate the usefulness of the proposed algorithm on frequency-selective fading channels.In the pilot-symbol aided modulation (PSAM) channel estimation, pilot symbols facilitate channel estimation, but reduce the transmit energy for data symbols under a fixed total transmit power constraint. We analyze the the effect of pilot-symbol-aided channel estimation error and correlated fading characteristic between the antennas on the performance and capacity, and derive the close-form expression of optimal pilot-to-data power ratio (PDR) in OFDM and correlated MIMO-OFDM systems. Theoretical and simulation results show that implementing the optimal PDR in an actual system should prove relatively straightforward, since there is a surprisingly broad range of correlation.