Air-to-Ground Big-Data-Assisted Channel Modeling Based on Passive Sounding in LTE Networks

Abstract

In this paper, a novel approach of channel modeling based on big data analysis is proposed that is applied to extract air-to-ground channel models from down-link signals collected by using an Unmanned Aerial Vehicle (UAV) in operating Long-Term-Evolution (LTE) networks. In this approach, the most "sensitive" channel parameter to the UAV height is chosen based on a feature selection algorithm from a parameter set consisting of nine channel parameters calculated from channel impulse responses. In the case considered here, the K-factor is found to be the most height-sensitive parameter. The behavior of the mean of K-factor is modeled as a piece-wise function against height which demonstrates a break point that is determined by assessing the contribution of height-dependent samples to the overall entropy. The residuals of subtracting the mean K-factor are statistically modeled. The results illustrate that the proposed big-data-assisted approach is applicable to provide accurate description of channel statistics versus the variables of interests.