Abstract:

Historically, the Gaussian assumption for noise and interferences has dominated signal processing. The justification for this assumption is usually given by the well known central limit theorem which allows an elegant, manageable, and meaningful mathematical description of noise. Nevertheless, some communication scenarios are characterized by impulsive properties on behalf of Gaussian noise, so they are non-Gaussian distributed. Channels which are corrupted with impulsive noise can be found in several domains like for example powerline communications, ethernet traffic, underwater signal processing, or as proven in this project also the multiuser interferences of ultra wideband (UWB) communication systems. After describing and discussing impulsive noise models in detail, a comparison of their distributions has been carried out showing the suitability of the different models for solving selected problems. For example, it is well known that the channel capacity with additive impulsive noise channels is always greater than the one of additive white Gaussian noise (AWGN) channels supposed that the signal to noise ratio (SNR) remains unchanged. In this project the capacity of channels corrupted by impulsive noise has been investigated. An example shows the negative effect of the widely used hard limiter (for impulsive noise clipping) and suggests the use of an alternative one called soft limiter allowing no loss of information. Furthermore, UWB systems have been investigated regarding multiuser interferences. It is proven that the multiuser interferences of impulse radio systems are impulsive. Their probability density function is determined analytically and confirmed by simulation results. Taking this result into account a new receiver structure has been developed allowing the suppression of impulsive noise without use of a limiter. The performance of this receiver is proven by simulation.