Signal processing for state estimation during sensor failure using mixed signal separation technique

One of the main source of concern in signal processing involves separation of unknown source signals from their linear mixtures when the transmission channel characteristics are unknown. In most cases this reduces to estimating the useful signal from a set of observable signals with noise. This paper discusses the structures that are used in separating instantaneously mixed signal sources. It is assumed that the only information available a priori is the set of observable signals. The separation of signals is done using independent component analysis. The proposed method is used in sensor signal processing for estimating the variations of a signal during faulty sensor situation.