Prof. Dr.-Ing. Werner Henkel
We address noise which is impulsive, non-Gaussian, and follows a heavy-tailed distribution. In wireless communications, often noise is considered as an additive white Gaussian noise (AWGN). In reality, receivers have omnidirectional capability and it receives many signals which are not only AWGN, it has non Gaus- sian components. Most commonly used heavy-tailed distributions are Middleton’s models (Class A, B, and C), the Symmetric Alpha-Stable distribution, the Gaussian mixture distribution, and the generalized Gaussian distribution. For the detection of signals in the presence of impulsive noise, these models are used to model the effect of impulse noise in wireless receivers. In Middleton’s Class-A model, a single antenna system is considered. Recently, McDonald extended this model for two antenna system. In our study, we extended Middleton’s Class-A model to make it suitable for multiple-antenna systems. In order to detect and mitigate the effect of impulsive noise, a model for a multiple-antenna system is adopted while this is an accurate model for the thermal noise present at the receiver and measured noise from different noise sources. In addition, an ignition circuit is designed which gen- erates impulse noise similar to the one of car ignition. Measurements are taken from different sources and using measured data, analytical models for the statistics of impulse noise including voltage histograms, probability density functions (pdfs) of voltages, Gaussian-Gaussian model, generalized Gaussian distribution have been developed, where the Gaussian-Gaussian can be seen as for class-A model. The Levenberg-Marquardt algorithm (curve-ftting tool) has been used to demonstrate the approximation of measured data with various model functions. To estimate re- quired parameters, this curve-fitting tool is used, which gives more precise results. Finally, MATLAB simulations have been performed to examine the effects of this non-linearity for both Middleton’s Class-A model, Gaussian mixture, and general- ized Gaussian distribution.
[Report] Wireless Impulse Noise Modeling