Statistical power consumption assessment of mobile access networks at country level

TLDR: This paper enrich and validate the existing analytical model for power consumption at sites level in a scale of a country, and proposes a statistical model based on linear regression to estimate the total power consumption.

Abstract: In this paper, we tackle the problem of assessment of the electrical power consumption of mobile access networks at country level. We focus on radio access sites hosting 2G and 3G base stations and their corresponding technical environment. First, we enrich and validate the existing analytical model that was proposed for power consumption at sites level in a scale of a country. We show that the fine-tuned analytical modeling yields interesting results only at global level but with strong a priori constraints on the impacting parameters. Then we propose a statistical model based on linear regression to estimate the total power consumption. The statistical model is tested and validated on real measurement from a whole network of European country. We show that, the proposed model is very efficient to predict the total power consumption of the entire studied network with a small training dataset and reduced number of input parameters. Indeed, with a training set of 30 sites (6% of sites), the global power consumption of the entire network is estimated within a 10% of margin at 95% of confidence level.