Adaptive DCS-SOMP for Localization Parameter Estimation in 5G Networks

Abstract

In this work, we model a 5G downlink channel using millimeter-wave (mmWave) and massive Multiple-Input Multiple-Output (mMIMO) technologies, considering the following localization parameters: Time of Arrival (TOA), Two-Dimensional Angle of Departure (2D-AoD), and Two-Dimensional Angle of Arrival (2D-AoA), both encompassing azimuth and elevation. Our research focuses on the precise estimation of these parameters within a three-dimensional (3D) environment, which is crucial in Industry 4.0 applications such as smart warehousing. In such scenarios, determining the device localization is paramount, as products must be handled with high precision. To achieve these precise estimations, we employ an adaptive approach built upon the Distributed Compressed Sensing—Subspace Orthogonal Matching Pursuit (DCS-SOMP) algorithm. We obtain better estimations using an adaptive approach that dynamically adapts the sensing matrix during each iteration, effectively constraining the search space. The results demonstrate that our approach outperforms the traditional method in terms of accuracy, speed to convergence, and memory use.