Separação e identificação de múltiplas fontes eletromagnéticas usando o método de análise em componentes independentes e o coeficiente de correlação.

Abstract

Currently the use of wireless has significantly increased the amount of electromagnetic sources (EM) and hence the concern on identification and separation of multiple sources physically closed. This situation is the main focus of this research, where is described the case of separation of two EM radiating sources showing small difference in frequency (ΔF), same amplitude and phase. This is accomplished without prior knowledge of the sources or the environment in which they are inserted. Experimental results, in which the sources are separated by distance of d = λ0/10 is presented. Then, there are shown the separation and identification of multiple sources with ΔF = 0.63%, generated by different equipment. A study of TV channels (digital and analog) spectrum sensing is developed to cognitive radio. For all cases presented above the method of the independent component analysis (ICA) was applied and the FastICA algorithm used for its implementation. The limitations of the traditional procedures for similar sources and spatially very close are shown. A new procedure of data collection is proposed based on the mathematical reconstruction of signals by measuring its phase and amplitude. Results for the influence of the number, positioning and arrangement of sensors, in the ICA response are presented. The correlation coefficient is used to quantify the ICA response, by comparing how much the estimated signal is linearly similar to the original one. These values can reach up to 0.99. The results are considered promising, when taking into account the complexity of the measurement, the small frequency difference, and the number and proximity of the sources.