UWB Channel Fading Statistics and Transmitted-Reference Communication
It is well known that Ultra WideBand (UWB) transmission is inherently robust against small-scale-fading (SSF) that arises in multipath scattering environments, due to its large signal bandwidth. However, no model with a physical interpretation exists that relates the variations of received signal strength to the signal bandwidth and general channel parameters, like e.g. the average channel power delay profile. Such a model would be of relevance for e.g. system designers, who have to make tradeoffs between system aspects, like complexity and energy efficiency on one hand, and robustness against small-scale fading on the other hand. In this thesis, a model is presented that allows for such a tradeoff analysis, relating the average power delay profile parameters and signal bandwidth to the statistical properties of the SSF. Additionally, it is shown how the uncoded and coded BER of BPSK modulation can be computed in a closed-form for a given average power delay profile and signal bandwidth. As stated before, UWB communication is inherently resilient against SSF. Unfortunately, coherent receivers become rather complex in the UWB case. In 2002, Tomlinson and Hoctor proposed to combine Transmitted Reference (TR) signaling with an autocorrelation receiver (AcR) for UWB communications, to dispose of the need for channel estimation. Due to the non-linear structure of the AcR, little was known with respect to its behaviour in various situations. This thesis aims to provide better insight in the behaviour of such systems. Not only is the principle of TR UWB communication explained, also several extensions to the TR principle are proposed, which relieve some of its drawbacks. Additionally, novel interpretations for TR UWB systems are presented, which explain the behaviour of TR systems e.g. in the presence of inter-symbol-interference. After understanding the behaviour of TR UWB systems, the design of a high-rate TR UWB system is presented that supports data-rates up to 100 Mb/s, while occupying 1 GHz of bandwidth. Using a combination of trellis-based equalization, multiband processing, turbo equalization and turbo coding, a system is obtained which is moderately complex with respect to digital signal processing and requires an Eb/N0 of only 12 dB to obtain a BER better than 10e-6.
