Generalized structure of blind adaptive frequency-shift filter for signal extraction

A generalized structure of a composite Blind Adaptive FREquency SHift (BA-FRESH) filter is proposed in this paper. By exploiting the spectral correlation of cyclostationary signals, the BA-FRESH filter does not requires a reference signals nor the statistical knowledge of the desired signal. It is capable of separating desired signals from spectral overlapping interfering signals by knowing the cyclic frequencies of the signals. Theoretical derivations of the optimum filters used in different variations of the structure are obtained and computer simulation results show that their performance are very satisfactory.