Multiple-input multiple-output data detection and channel estimation in flat fading environments

The multiple-input and multiple-output (MIMO) technique has emerged as a key feature for future generation of wireless communication systems mainly because of its potential huge channel capacity. This dissertation addresses two practical issues in design of an MIMO receiver in flat fading environments: MIMO data detection and channel estimation.; For MIMO detection, Markov chain Monte Carlo (MCMC) simulation techniques are used to obtain Bayesian estimates (soft information) of the transmitted symbols. Unlike the previous reports that widely use the statistical inference to estimate a posteriori probability (APP) values, this dissertation proposes alternative statistical methods by viewing the underlying problem as a multidimensional Monte Carlo integration. It is shown that this approach leads to results similar to those that would be obtained through a proper Rao-Blackwellization technique and thus it is concluded that the proposed methods are superior to those reported in the literature. It is also noted that when the channel signal-to-noise (SNR) is high, the MCMC simulator experiences some very slow modes of convergence and thus accurate estimation of APP values requires simulation of very long Markov chains, which may be infeasible in practice. Two solutions to this problem using the theory of importance sampling are proposed. Extensive computer simulations show that both solutions improve the system performance greatly. The proposed detectors are compared with the best existing detectors, including the well-known list-version sphere decoding (LSD) algorithm, and a superior performance is observed. Further research is conducted to explore the cause of the performance gap, and as a consequence a better implementation of LSD algorithm is proposed.; On channel estimation, pilot-aided estimation methods are explored for fast flat fading MIMO channels. Two different piloting schemes are investigated. The first scheme uses pilot symbols that are time multiplexed with data symbols. This is referred to as a pilot inserted (PI) scheme. The second scheme uses pilot symbols that are added to and transmitted simultaneously with data symbols. This is referred to as a pilot embedding (PE) scheme. The linear minimum mean square error (LMMSE) estimators of both schemes are developed. It is shown that for the optimum performance both PI and PE schemes should transmit space-time pilot symbols that form orthogonal matrices. The PI and PE schemes are also compared for the capacity they can offer. It is found that in low SNR, PE scheme has a slightly higher capacity; however, in higher SNR, PI scheme offers a better capacity. Moreover, an iterative joint channel estimation and data detection method is proposed and used to corroborate the above findings in a computer simulation set-up.