DCAR Subarray Location Method in Complex Environment Based on ISGMD-TSELM

Abstract

In distributed coherent aperture radar (DCAR) systems, achieving high-precision subarray localization in complex environments is a fundamental challenge for ensuring the coherence synthesis performance of the radar system. However, low-power calibration requirements result in the calibration signal being submerged in noise signals. A subarray location method based on Improved Symplectic Geometry Mode Decomposition (ISGMD) and Tangent Space Extreme Learning Machine (TSELM) is proposed for a miniature mobile DCAR system. Firstly, the signals' power distribution is optimized to preserve the essential characteristics and coupling relationships among the modes in the constructed symplectic matrix. Secondly, prior information is utilized to calibrate the iterative process of SGMD, enhancing robustness and effectiveness in signal reconstruction and noise separation. Moreover, the model enriches global relations and representation capability of information in high-noise backgrounds by approximating the local relationships of the signal covariance matrix in the Riemannian space, which leads to improved performance in solving nonlinear problems. Consequently, the subarray positioning accuracy is enhanced to satisfy the requirements for achieving full-phase coherence in mobile DCAR systems. This ensures the attainment of the desired signal-to-noise ratio (SNR) gain by the system. Numerical simulation results demonstrate that the proposed ISGMD-TSELM method effectively filters out background noise and feature redundancy in the received signals. When the SNR is −6 dB, the algorithm guarantees a coherent synthesis efficiency of over 70% for the system.