Inverse of linear periodically time-varying filtering

An inverse filtering scheme for linear periodically time-varying systems is proposed. The scheme makes use of the property that there exist correlations between frequency-shifted versions of a cyclostationary signal. It successfully separates spectrally overlapped shifted versions of a signal by utilizing the correlations between a wide sense cyclostationary (WSCS) process in different frequency bands. The proposed scheme outperforms previous inverse filtering techniques and can be extended to the inverse problem for multi-intrinsic frequency linear periodically time-varying systems. Simulation results agree well with the theoretical development