Echo State Network and Echo State Gaussian Process for Non-Line-of-Sight Target Tracking

Abstract

Non-line-of-sight (NLOS) propagation blocks the direct path between the anchors and the agent, therefore, results in large positive bias in localization. Mobile target tracking in a NLOS scenario is much more challenging than static localization because of the need of not only dynamic modeling but also NLOS error mitigation. Gaussian Process (GP) regression is the state-of-the-art machine learning approach that resolves the issue. This article proposes two novel nonparametric strategies for NLOS target tracking, namely Echo State Network (ESN) and Echo State Gaussian Process (ESGP), both of which offers significant enhancement to the tracking performance compared to GP regression. Moreover, ESN is much more computationally efficient than GP regression and ESGP also provides a measure of confidence on the tracking results. Monte Carlo simulations and experiments prove the robustness and effectiveness of ESN and ESGP.