Post date: 15-sep-2014 10:56:29

Title: Heavy-tailed Interference Approximation via Log-cumulants-based Edgeworth Expansion

Date: 06-May-2014

Material: pdf slides

Speaker: Giancarlo Pastor Figueroa (URJC)

Abstract: The Edgeworth expansion approximates a probability distribution function in terms of the cumulants. This expansion is developed within the framework of the First Kind Statistics, where definitions are derived from the Fourier transform, e.g. characteristic functions and cumulants. Alternatively, a similar framework called Second Kind Statistics offers analogous definitions which are derived from the Mellin transform. Although a formalism with such similarity to the existing definitions cannot lead to intrinsically new results, approximation methods within this framework have been understudied. This work introduces the Edgeworth expansion in terms of the log-cumulants, which are the analogous to cumulants in the Second Kind statistics. Remarkably, this new expansion is much more suitable to handle heavy-tailed R^+ distributions which are commonly found in network interference modeling.