Characterisation and Modelling of Indoor and Short-Range MIMO Communications
MetadataShow full item record
Over the last decade, we have witnessed the rapid evolution of Multiple-Input Multiple-Output (MIMO) systems which promise to break the frontiers of conventional architectures and deliver high throughput by employing more than one element at the transmitter (Tx) and receiver (Rx) in order to exploit the spatial domain. This is achieved by transmitting simultaneous data streams from different elements which impinge on the Rx with ideally unique spatial signatures as a result of the propagation paths’ interactions with the surrounding environment. This thesis is oriented to the statistical characterisation and modelling of MIMO systems and particularly of indoor and short-range channels which lend themselves a plethora of modern applications, such as wireless local networks (WLANs), peer-to-peer and vehicular communications. The contributions of the thesis are detailed below. Firstly, an indoor channel model is proposed which decorrelates the full spatial correlation matrix of a 5.2 GHzmeasuredMIMO channel and thereafter assigns the Nakagami-m distribution on the resulting uncorrelated eigenmodes. The choice of the flexible Nakagami-m density was found to better fit the measured data compared to the commonly used Rayleigh and Ricean distributions. In fact, the proposed scheme captures the spatial variations of the measured channel reasonably well and systematically outperforms two known analytical models in terms of information theory and link-level performance. The second contribution introduces an array processing scheme, namely the three-dimensional (3D) frequency domain Space Alternating Generalised Expectation Maximisation (FD-SAGE) algorithm for jointly extracting the dominant paths’ parameters. The scheme exhibits a satisfactory robustness in a synthetic environment even for closely separated sources and is applicable to any array geometry as long as its manifold is known. The algorithm is further applied to the same set of raw data so that different global spatial parameters of interest are determined; these are the multipath clustering, azimuth spreads and inter-dependency of the spatial domains. The third contribution covers the case of short-range communications which have nowadays emerged as a hot topic in the area of wireless networks. The main focus is on dual-branch MIMO Ricean systems for which a design methodology to achieve maximum capacities in the presence of Line-of-Sight (LoS) components is proposed. Moreover, a statistical eigenanalysis of these configurations is performed and novel closed-formulae for the marginal eigenvalue and condition number statistics are derived. These formulae are further used to develop an adaptive detector (AD) whose aim is to reduce the feasibility cost and complexity of Maximum Likelihood (ML)-based MIMO receivers. Finally, a tractable novel upper bound on the ergodic capacity of the above mentioned MIMO systems is presented which relies on a fundamental power constraint. The bound is sufficiently tight and applicable for arbitrary rank of the mean channel matrix, Signal-to-Noise ratio (SNR) and takes the effects of spatial correlation at both ends into account. More importantly, it includes previously reported capacity bounds as special cases.