Separation of electromagnetic sources by the method of independent component analysis

This work describes how to separate two electromagnetic sources with slight frequency difference and two sources with same amplitude, phase and frequency. The case of sources closely separated for example by λ₀/4 is presented. It is applied the method of independent component analysis (ICA). An ICA algorithm based on the maximization of the negentropia and symmetric orthogonalization (FastICA) is used. An appropriate measuring system is mounted for the acquisition of data to be used in the ICA application. The sources used are two half wavelength dipole antennas, designed for 1.800 GHz. The correlation coefficient is used as a parameter to compare how much the estimated signal is similar to the original one. The limit of the difference between the frequencies the signals, that the ICA can separate, is evaluated through simulations. Signal reconstruction is required when the sources are equal before performing the separation process. The reconstruction of the sources after separation, using the ICA, is compared with original sources and it is in good agreement.