An optimal non-parametric filter for high interference harmonic signals

In this paper the separation of periodic harmonic signals from noisy interference mixture is proposed. The optimal filter is designed using minimum mean square technique by Amplitude and Phase Estimations (APES) method. The filter is designed such that they have unit gain at the frequencies of the harmonics of the desired signal and suppress everything else. The design has to overcome the shortcomings of previous designs while retaining their desirable properties, like the ability to cancel out interfering signals. The design reduces to a number of well-known designs under certain conditions. The proposed filter has able to separate signals with signal to distortion ratio (SDR) 7dB greater than previous designs.