Nonlinear Signal Processing Techniques for UWB Impulse Radios

Abstract

Impulse Radio (IR) Ultra-wideband (UWB) communication is an attractive potential technology for low-power, low-complexity, and high-speed communications in short range links. One of the main challenges it faces is the highly frequency-selective multipath channel, which generates hundreds of overlapped copies of the transmitted pulse with different delays and amplitudes. In collecting the energy of these multipath components, conventional Rake receivers suffer from high implementation and computational costs due to channel estimation. To circumvent the problem, several non-coherent receivers, energy detection (ED) based receivers, are proposed; however, they come at the cost of degraded performance when narrowband interferences (NBIs) are present. In this dissertation, we present low-complexity, high-performance, non-coherent receiver designs that i) avoid the expensive channel estimation; ii) lower the hardware implement complexity with the use of nonlinear signal processing techniques; and iii) improve the error performance by considering practical imperfections. Firstly, we propose a Kurtosis Detection (KD) operator to replace the square operator in conventional ED; without high sampling rates, the KD based receiver achieves better performances. Secondly, we propose another nonlinear signal processing technique, variance detection (VD), to mitigate NBI effect on the ED-based IR-UWB systems. The lognormal distribution model is introduced and used for derivation of the analytical BER of the VD-based receiver. Thirdly, we propose a unified framework, generalized nonlinear detection (GND), to generalize existing nonlinear detection technologies and further optimize the performances. Our results show that the GND-based receivers have better system performance and stronger ability to resist NBIs than the existing nonlinear detection algorithms. Finally, a blind NBI mitigation technique with a square law (SL) device is presented to mitigate the NBI effects. Our results show that the SL technique can significantly improve the signal-to-interference ratio of the received UWB signal without any prior knowledge of the NBI, and can be implemented in hardware easily; therefore, the proposed SL technique is an implementable and highly effective blind NBI mitigation technique for ED-based IR-UWB systems.