Exploiting Channel Correlations for NLOS ToA Localization With Multivariate Gaussian Mixture Models

Abstract

In this letter, we develop a Bayesian probabilistic approach for time-of-arrival (ToA) localization in non-line-of-sight (NLOS) channels, where multivariate Gaussian mixture models (GMM) are used to approximate the joint distribution of channel bias values and harness the channel correlations. Using over-the-air measurements from a proprietary localization system, we numerically demonstrate that the proposed algorithm outperforms both a counterpart Bayesian probabilistic approach that does not take into account channel correlations and the well-known non-linear least square (NLS) optimization method.