An optimal spatio-temporal filter for extraction and enhancement of multi-channel periodic signals

Filtering methods have been widely used for extraction of signals in both time and space. Recently, two multi-channel filters have been proposed which can be applied for extraction of multi-channel periodic signals. In these filters, the harmonic structure of periodic signals is exploited. The filters were based on the periodogram and the LCMV beamformer, respectively. The periodogram-based filter is unsuitable for multi-source scenarios whereas the LCMV-based filter has an erratic filter response behaviour at high SNRs. We propose an optimal filtering method which is useful in multi-source scenarios and which has a nicely behaving filter response for a larger range of SNRs compared to the LCMV-based filter. Our simulations show that our method solves the high SNR issue of the LCMV-based filter and that the proposed filter is applicable to real-life signals.