An experimental study of the effect of human activity on the alpha-stable characteristics of the power-line noise

In recent works it has been recognized that alpha stable distributions are convenient characterizations of some noise properties in the power line. They naturally include in their formulation both background noise and impulsive components. The intended applications of this modeling approach are twofold. First, it allows to better understand the statistical properties of the noise. Second, it allows to synthesize the noise process in a more realistic way, as compared to commonly used Gaussian models. In this work we focus our attention on the former application. We study the relation between human activity and the parameter values of the alpha stable probability distribution. In our research methodology we collected a large number of noise samples in three real power distribution networks: a typical town house, a university office and a semi-industrial workshop. Then, we applied some parameter estimation algorithms to the collected noise traces and observed the time evolution of the parameter values. It is shown that noise in all considered scenarios presents characteristics that deviate from the Gaussian assumption and this behavior depends on the level of human activity. This result should be taken into consideration for the realistic representation of transmission impairments in the design and evaluation of power-line communication (PLC) systems.