Coupled Block-Term Tensor Decomposition for Near-Field Localization in multi-static MIMO Radar Systems

Abstract

This paper presents a high-resolution coupled rank-(L<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">r</inf>,L<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">r</inf>,1) block-term decomposition-based near-field localization scheme for multi-static MIMO radar systems. The proposed COBRAS (COupled Block-term decomposition for multi-static RAdar Systems) algorithm uses the exact wavefront model to estimate the target location parameters in 3D space and can be applied to arbitrary array geometries. Compared to the far-field models, the exact near-field wavefront model allows exploiting the distance information for high-accuracy positioning. Moreover, we consider a system with massive antenna arrays, which increases the Fresnel region and expands the range of the near-field assumption. The COBRAS algorithm includes the initial tensor decomposition of the data and further post-processing steps that allow extracting the location parameters. Additionally, we compare the performance of different rank-(L<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">r</inf>,L<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">r</inf>,1) block-term decomposition algorithms and demonstrate how the employment of coupling improves the localization performance compared to the non-coupled solutions.