Extraction of a source signal whose kurtosis value lies in a specific range

In many applications extraction of source signals of interest from observed signals is a more feasible approach than simultaneous separation of all the source signals, since the latter often costs lots of computing time and often is not necessary. If the desired source signals have some specific properties, then we can exploit these properties to design effective source extraction algorithms. This letter proposes an algorithm, which extracts the desired signal with a priori knowledge about its statistics. That is to say, if we know the range in which the kurtosis value of the desired signal lies, we can use this algorithm to extract it. The validity and performance of the proposed approach are confirmed through computer simulations and experiments on real-world ECG data.