Sub array technique applied to reducing the number of sensors in blind source separation

This work describes an efficient technique to separate electromagnetic sources, which present very small frequency difference (ΔF = 0.41%). The proposed technique needs fewer sensors than sources to collect the data. The case of sensors array with elements distant λ0/4 is presented, but can be extended to λ0/2. It is applied the independent component analysis (ICA) method. An ICA algorithm based on the maximization of nongaussianity and symmetric orthogonalization (FastICA) is used. The appropriate measuring system is mounted to demonstrate the procedure used for data acquisition. Three different antennas were used as independent sources. An array of two planar dipole antenna connected to a radio frequency analyzer was used as sensors in the measuring system. Experimental results show that it is possible the identification of several sources with fewer sensors.