Off-grid Multi-Source Passive Localization Using a Moving Array

TL;DR

This paper introduces a new passive localization method using a moving array and sparse representation, achieving higher accuracy and robustness at low SNR with fewer observations.

Contribution

It develops a novel on-grid compressive sensing approach with a majorization-minimization technique for finer and more robust passive localization.

Findings

Effective at low SNR conditions

Requires fewer observing positions

Achieves finer target position estimation

Abstract

A novel direct passive localization technique through a single moving array is proposed in this paper using the sparse representation of the array covariance matrix in spatial domain. The measurement is constructed by stacking the vectorized version of all the array covariance matrices at different observing positions. First, an on-grid compressive sensing (CS) based method is developed, where the dictionary is composed of the steering vectors from the searching grids to the observing positions. Convex optimization is applied to solve the `1-norm minimization problem. Second, to get much finer target positions, we develop an on-grid CS based method, where the majorization-minimization technique replaces the atan-sum objective function in each iteration by a quadratic convex function which can be easily minimized. The objective function,atan-sum, is more similar to `0-norm, and more sparsity encouraging than the log-sum function.This method also works more robustly at conditions of low SNR, and fewer observing positions are needed than in the traditional ones. The simulation experiments verify the promises of the proposed algorithm.