Independent Component Analysis of the Cyclostationary Signals in the Transmission Line

Abstract

The blind separation of the cyclostationary signals using a single channel measurement in transmission line is proposed. The waveform identification of the partial pulse responses is evaluated for additively mixed sources of the data, inter-symbol interference and crosstalk. A matrix model of a composed random vectors is considered. The proposed estimation procedure is based on pre-processing using principal component analysis of measured data by digital oscilloscope and following independent component analysis. The component analysis allows diagnostics of signal integrity using eye-diagram patterns in additional dimension of sources and interference.