Bayesian Compressive Sensing Enabled EMI Source Localization in Shielding Enclosures From Complex and Phaseless Near-Field Scanning

Abstract

Electromagnetic interference (EMI) can disrupt the operation and reliability of electronic devices, thus effective shielding and detection of EMI sources are crucial to mitigate these issues. However, locating EMI sources in shielding enclosures is a challenging task due to the complex electromagnetic (EM) environment. Compared to conventional methods that demand dense spatial sampling and lengthy data acquisition, this article proposes an efficient approach to locate EMI sources based on the compressive sensing (CS) of complex or phaseless near-field (NF) data. The internal EMI sources are modeled as equivalent dipoles in the enclosure through the use of numerical Green's function (NGF). With the help of the CS framework, the required NF measurements can be largely reduced. By further resorting to the reciprocity theorem, a large amount of CPU cost involved in the NGF evaluation can be saved, thus accelerating the construction of the matrix equation that connects the equivalent dipole sources and the NF data. To solve this highly ill-posed inverse problem arising from high-coherent under-sampled NF data, a multitask Bayesian compressive sensing (MT-BCS) algorithm is employed, facilitating reliable reconstruction of sparsely distributed EMI sources. Moreover, to meet the need for more accurate and convenient magnitude-only measurements in high-frequency scenarios, this work proposes a phaseless mapping technique to linearize the nonlinear problem caused by phaseless data, and further develops a tailored MT-BCS strategy to enable robust and efficient reconstruction from limited phaseless NF measurements. Numerical and experimental results demonstrate the effectiveness, robustness, and high efficiency of the proposed method. This novel methodology offers powerful capabilities for practical EMI diagnostics and other EM inverse problems.