Mobile wireless positioning has recently received great attention. For mobile wireless communication networks, an inherently suitable approach is to obtain the parameters that are used for positioning estimates from the radio signal measurements between a mobile device and one or more xed base stations. However, obtaining accurate estimates of these location-dependent channel parameters is a challenging task. The focus of this thesis is on the estimation of these channel parameters for mobile wireless positioning applications. In particular, we investigate novel estimators that jointly estimate more than one type of channel parameters. We rst perform a comprehensive critical review on the most recent and popular joint channel parameter estimation techniques. Secondly, we improve a state-of-the-art technique, namely the Space Alternating Generalised Expectation maximisation (SAGE) algorithm by employing adaptive interference cancellation to improve the estimation accuracy of weaker paths. Thirdly, a novel intelligent channel parameter estimation technique using Evolution Strategy (ES) is proposed to overcome the drawbacks of the existing iterative maximum likelihood methods. Furthermore, given that in reality it is di cult to obtain the number of multipath in advance, we propose a two tier Hierarchically Organised ES to jointly estimate the number of multipath as well as the channel parameters. Finally, we extend the proposed ES method to further estimate the Doppler shift in mobile environments. Our proposed intelligent joint channel estimation techniques are shown to exhibit excellent performance even with low Signal to Noise Ratio (SNR) channel conditions as well as robust against uncertainties in initialisations.

Abstract The present dissertation deals with the evaluation of multiple-input multiple-output (MIMO) radio channel characteristics from time-division multiplexing (TDM)-switched MIMO channel sounding. The research can be divided into three main areas. First, the impacts of phase noise in TDM-switched MIMO channel sounding on channel capacity are studied. Second, we focus on those impacts on channel parameter estimation using the SAGE algorithm. And in the last part, spatial correlation, channel eigenvalue distribution, and ergodic capacity in realistic environments are analyzed. The rationale behind the first two areas is that most advanced MIMO radio channel sounders employ the TDM technique, which has significant problems from phase noise of the TX and RX phase locked loop (PLL) oscillators causing measurement errors in terms of estimated channel capacity and parameters. We propose statistical models that reproduce the capacity estimates. The effects of the sounding mode (SM), the length of pseudo-random noise (PN) sequence L of the sounding signal, and the system size are disclosed. The distinctive basis is to consider the impact of the actual phase noise in TDM switched MIMO channel sounding, instead of assuming white Gaussian-type phase noise. In a reality, the short-term phase noise component affecting one measurement cycle of a MIMO system plays an important role in the traditional estimators of the radio channel parameters and capacity. We show that the performance impairment is less than that been under the hypothesis of uncorrelated white Gaussian phase-noises samples. The difference is due to the non-vanishing correlation of phase-noise within the measurement cycle. Two approaches to mitigating the impact of phase noise are proposed. The former is the simple and efficient sliding averaging method, where the signal-to-noise ratio (SNR) of the channel impulse response can be increased. The latter is the choice of SM and L, which is more thorough. In the second part, two approaches to mitigating its impact on channel parameter estimation using the SAGE algorithm are also discussed. Besides the sliding averaging, which in general can increase the SNR, the new SAGE algorithm based channel parameter estimation based on the improved signal model accounting for the phase noise in the measurement device is proposed. Finally, the channel eigenvalue distribution and ergodic capacity based on complex hypergeometric functions and their asymptotic characteristics are analyzed. It is shown that the derived theoretical expressions closely approximate the simulated results of the measured finite-dimensional MIMO channels. The spatial correlation and the eigenvalue statistics in frequency selective channels for single and dual polarized antennas are investigated. This knowledge is useful when different MIMO and beamforming techniques are applied.