Target Localization in Multipath Propagation Environment Using Dictionary-Based Sparse Representation

Yuan Liu,Guoan Bi,Hongwei Liu,Xiang-Gen Xia,Lu Wang

doi:10.1109/access.2019.2947497

Yuan Liu, Guoan Bi + Show 3 more

Open Access

https://doi.org/10.1109/access.2019.2947497

Copy DOI

Journal: IEEE Access	Publication Date: Jan 1, 2019
Citations: 4	License type: CC BY 4.0

Affiliation: Nanyang Technological University, Xidian University

Abstract

This paper addresses the target localization problem in complex multipath propagation environment for three-dimensional (3-D) radar systems. Firstly, an approach based on the singular value decomposition (SVD) technique is developed to reduce the data dimension and formulate the joint multiple snapshot sparse representation problem in the signal subspace domain. Subsequently, a novel sparse representation based DOA estimation algorithm, combined with alternatingly iterative and dictionary refinement techniques, is proposed. The Cramer-Rao bounds (CRB) for the target DOA and attenuation coefficient estimations of multipath model are derived in closed forms. Experimental results based on both simulated data and measured data indicate that the target localization accuracy can be effectively enhanced by utilizing the proposed algorithm in complex terrain and/or limited snapshot scenarios.

Highlights

It is well-known that one of the critical missions achieved by modern three-dimensional (3-D) radar is measuring the altitude of a target
For performance studies on various scenarios, e.g., those with different reflecting surfaces and/or different numbers of snapshots, several simulations are performed to illustrate the performance of the proposed low-angle target localization algorithm in this subsection
In all the presented simulations, a uniform linear array (ULA) is considered, which comprises M = 16 isotropic elements separated with one halfwavelength, i.e., λ 2 = 0.5 m

Summary

Introduction

It is well-known that one of the critical missions achieved by modern three-dimensional (3-D) radar is measuring the altitude of a target. When tracking a low-angle target, the localization performance of radar system may be affected by the multipath phenomenon from ground surface reflection, i.e., target echoes consisting of a direct path as well as highly correlated multipath [1]–[4]. There have been considerable efforts to deal with this problem in the conventional radar systems since 1970s, for example, [1]–[11]. The key issue of target altitude measurement is to estimate its DOA from the radar echo corrupted by the multipath propagation. The most well-known existing nonparametric DOA estimation methods include subspace-based methods [5]–[8]

Objectives

Results

Conclusion