| Ultra-wideband (UWB) technique has been the subject of extensive research in recent years due to its unique capabilities and potential applications, particularly in short-range multiple access wireless communications. Despite of many advantages over traditional narrow band systems, UWB impulse radio is sensitive to timing jitter effect. In this thesis, we devote to design a high speed UWB receiver which is robust under timing jitter. The basic idea is that the received signal is over-sampled within the sampling window and its maximum value or maximum absolute value is selected as the decision variable and to be compared with a threshold. A main advantage of this scheme is that its performance will not degrade if the timing jitter is shorter than half of the window length. Therefore, the proposed scheme is robust against timing jitter and no precise synchronization between the transmitter and receiver is required. However, how to determine the optimal threshold value is a critical issue for this new scheme. In this thesis, we propose a simple stochastic approximation (SA) approach to adjust the threshold recursively. The approach is based on a version of SA known as the Kiefer-Wolfowitz (KW) algorithm with expanding truncations and randomized differences. Corresponding to two different decision-making structures, two SA algorithms are presented and their convergence properties are analyzed, respectively. The proposed algorithms are effective in threshold optimization and the convergence rate is fast, as demonstrated by the numerical results.;Prior to timing jitter robust receiver design, a measurement campaign was carried out on indoor UWB signal propagation in order to characterize the UWB indoor channel and provide a simulation platform. Channel parameters are analyzed based on measurement data, including path loss, RMS delay spread, amplitude distribution and correlation properties.;Finally, we extend the threshold optimization algorithm to solve a more general N-state distributed estimation problem. We combine multiple observations of a signal process via the maximum function for decision-making and find out that the optimal decision function can be implemented by means of thresholds under suitable technical conditions. We propose here a training sequence based algorithm for threshold adjustment. The algorithm is a variation of the Kiefer-Wolfowitz algorithm with expending truncations and randomized differences. Convergence of this algorithm is also established. |
