Efficient Calculation of Meta Distributions and the Performance of User Percentiles

TL;DR

This paper introduces a new binomial mixture method for efficiently calculating meta distributions in wireless networks, improving accuracy and simplicity over existing approaches by leveraging moments of conditional distributions.

Contribution

The paper proposes a novel binomial mixture approach for directly computing meta distributions from moments, addressing inefficiencies of traditional methods.

Findings

The binomial mixture method provides a simple linear transform of moments.

It offers improved efficiency and robustness over standard approaches.

The method enhances the analysis of user performance percentiles in wireless networks.

Abstract

Meta distributions (MDs) are refined performance metrics in wireless networks modeled using point processes. While there is no known method to directly calculate MDs, the moments of the underlying conditional distributions (given the point process) can often be expressed in exact analytical form. The problem of finding the MD given the moments has several solutions, but the standard approaches are inefficient and sensitive to the choices of a number of parameters. Here we propose and explore the use of a method based on binomial mixtures, which has several key advantages over other methods, since it is based on a simple linear transform of the moments.