Localization of Low-frequency Source Current Distributions

Electromagnetic (EM) noise field distributions measured externally around electrical and electronic equipment have been used for localization and identification of the noise sources inside the equipment. For high frequencies (more than hundreds of MHz), the EM sources at finite distances have been localized by applying the techniques such as MUSIC [1], SPM [2], and holographic imaging [3], to the EM fields observed by a sensor array. On the other hand, at very low frequencies, the problem of localizing near-field EM sources has been solved for example in bioelectromagnetic inversion problems, such as the MUSIC localization of the current dipoles within human brains [4]. By applying the MUSIC algorithm [4][5] to the magnetic field distributions measured with an array of magnetic vector sensors placed around low-frequency (less than MHz) magnetic dipoles, we have estimated the 3-d locations and orientations of multiple incoherent [6] and/or coherent [7] small current loop sources. For larger current loops whose loop sizes are not negligible, we have extended the MUSIC algorithm to estimate their sizes as well as their locations and orientations [8]. However, the extended algorithm has assumed the shape of the source current loops simply as hexagonal. In this study, we try to estimate the shape of the source current distributions, by applying the GVSPM (Generalized Vector Sampled Pattern Matching) method [9], combined with the MUSIC algorithm.