High-Precision LoS Localization Using Composite Nakagami-m Log-Normal Model

Overview & Synopsis

This paper presents a high-precision localization method tailored for line-of-sight (LoS)-dominated wireless environments. Unlike conventional received signal strength (RSS) techniques that indiscriminately utilize multipath components, our approach leverages LoS path characteristics through a refined Composite Nakagami-m Log-Normal statistical model. We demonstrate that the LoS signal amplitude follows a Nakagamim distribution (equivalent to Gamma-distributed energy), while its scale parameter exhibits log-normal variation due to freespace path loss. This dual-scale model accurately captures both small-scale fading and distance-dependent attenuation unique to LoS propagation. For practical implementation, we develop an expectation-maximization (EM) algorithm enhanced with Langevin Monte Carlo (LMC) sampling, enabling efficient maximum likelihood estimation. Experimental validation using Quadriga-generated LoS channel data confirms consistent high accuracy, outperforming standard RSS methods in comparable LoS conditions. Although primarily designed for LoS scenarios, the framework offers theoretical insights for future extensions to mixed propagation environments. The solution remains practical for existing infrastructure, providing immediate benefits for 5G/6G systems in LoS-predominant deployments such as urban canyons and millimeter-wave cells.