Complex principal and independent component Analyses of radiated emissions for the time-frequency domain

Two methods are presented for the time- or frequency-domain characterization and diagnostics of unintentional radiated emissions from printed circuit boards. Decompositions produced by principal component analysis (PCA) are compared with those obtained using independent component analysis (ICA). While both techniques achieve first- and higher-order orthogonality of their basis functions, the nature of these functions is found to be radically different, viz., compact vs. extended support in the time domain. Both methods are applied to and compared for out-of-band emissions from realistic wireless transmitting portable device, outside its Bluetooth communication band. The results are supported by calculated field statistics, which are found to differ significantly from Gaussianity, causing significant contributions from higher-order moments in the principal components as opposed to the independent components.