Signal Processing for Ultra Wideband Transceivers

PhD Student 
Christoph Krall

 

 In this thesis novel implementation approaches for standardized and non-standardized ultra wide-band (UWB) systems are presented. These implementation approaches include signal processing algorithms to achieve processing of UWB signals in transceiver front-ends and in digital back-ends. A parallelization of the transceiver in the frequency-domain has been achieved with hybrid filterbank transceivers. The standardized MB-OFDM signaling scheme allows parallelization in the frequency domain by distributing the orthogonal multicarrier modulation onto multiple units. Furthermore, the channel's response to wideband signals has been parallelized in the frequency domain and the effects of the parallelization have been investigated. Slight performance decreases are observed, where the limiting effects are truncated sidelobes and filter mismatches in analog front-ends. Measures for the performance loss have been defined. For UWB signal generation, a novel broadband signal generation approach is presented. For that purpose, multiple digital-to-analog converters are used in an array to achieve flexible (adaptive) signal generation. Firstly, the converters in the array are assumed to be perfectly aligned to the clock signals, such that no mismatch spectra occur. Secondly, time offsets are introduced in the converter model and a compensation algorithm is presented. A digital predistortion of the signals, to compensate for the mismatch spectra, is presented and implemented, which achieves a reduction of the mismatch spectra by app. 20 dB. Furthermore, receiver architectures for the standardized IEEE802.15.4a signaling scheme, which is a pulse-based signaling scheme, are investigated. A comparison of three receivers in single and multi-user environments is presented. It is seen that the receiver proposed in this thesis has superior performance in the multi-user case, because it uses spreading information present in the standardized UWB signals. To reduce the distortions encountered in non-coherent receiver architectures at high data rates, a novel equalization algorithm for nonlinear receiver front-ends is presented. The nonlinear second-order equalizer can be optimized and computed according to a minimum mean squared error (MMSE) criterion. It is found that the nonlinear equalizer is a generalization of the linear equalizer equations. The solution is compared to an iterative learning algorithm (LMS), which shows asymptotic convergence to the presented solution. The presented equalizer improves the uncoded BER floor by a factor of 20.  

 

This thesis is supervised by Gernot Kubin, Klaus Witrisal.