Efficient PWS-Based Computation of the Sensing Matrix for Computational Radar Imaging Systems

Abstract

This letter presents an efficient method to compute the sensing matrix of computational imaging (CI) systems based on the plane wave spectrum (PWS) representation of electromagnetic waves. CI has recently been proposed as a promising paradigm to overcome some of the limitations of conventional radar imaging systems (mainly related to low acquisition speeds). As CI resorts to antennas that radiate spatially incoherent patterns to compress the scene information, the system complexity is transferred to the signal processing layer. CI systems are characterized by means of the so-called sensing matrix, which properly accounts for the spatially incoherent behavior of the antennas. The conventional method used to compute the sensing matrix is based on a point-by-point propagation between the aperture fields and the imaged scene, which is significantly time consuming. To address this limitation, this work proposes a PWS-based method, which enables to perform a plane-to-plane propagation. This results in a drastic improvement in the time required to compute the sensing matrix, without affecting its accuracy or the quality of the retrieved radar images. The proposed method has been experimentally validated with a CI system using frequency-diverse antennas.