Spread-Out Overlapping Sources by Independent Component Analysis for Location Positioning

The problem of spread-out overlapping sensors and sources of electromagnetic fields is discussed, and the way to distinguished them having same amplitude, phase, and frequency is presented. The case of sources separated by small distance d <;<; λ is studied. To do so, the method of independent component analysis is applied. In this work an algorithm based on maximizing the negentropy and symmetric orthogonalization is used. A measurement setup is proposed to construct the database for the algorithm. Two half wavelength dipoles, at 1.8 GHz, are considered as sources. A high gain broadband antenna is used to receive the incoming signals. The correlation coefficient is used to evaluate how much the estimated signals using the independent component analysis are similar to the originals signals. Correlation coefficient high as 0.98 is achieved for sources placed λ/10 apart. The results obtained by independent component analysis are promising considering the complexity of the measurement setup and the proximity of sources.