Statistical Characterization of Wireless Interference Signal Based On UWB Spectrum Sensing.

Ultra-wideband (UWB) technology offers the potential for unparalleled support of short-range broadband communication over a multi-gigahertz spectrum and are expected to enable several applications with extreme requirements in future wireless networks. Enabling these systems in the unlicensed spectrum requires efficient co-existence management and adequate understanding of the characteristics and spatio-temporal dynamics of interference signals over the multi-GHz bandwidth. This paper investigates the suitability of Gaussian, Middleton canonical class A, symmetric alpha stable and Gaussian Mixture distributions for modelling radio frequency interference from systems in the UWB spectrum based on measurements. We evaluate the closeness of fit of the distributions to measured interference data and provide insights on the applicability of these models for characterizing interference in the UWB spectrum. Results show that the Gaussian Mixture distribution (GMD) yielded the best fit to the measured interference evaluated with Kullback-Leibler (KL) divergence below 0.05. Results also show that interference signals generated from the GMD agree closely with the measurements.