Blind adaptive spatiotemporal filtering for wide-band cyclostationary signals

An algorithm that blindly adapts spatiotemporal filters to extract from sensor array data one or more desired signals having known cyclostationarity properties is presented. This method, referred to as the spectral self-coherence restoral (SCORE) method, can accommodate multiple signals of interest, multiple interferers, and unknown noise, does not require complicated searches, and requires knowledge of only a cycle frequency of the desired signal. The SCORE method is reinterpreted using the framework of canonical correlation analysis and extended to accommodate wide-band received data that do not adhere to the narrow-band assumption. Background information on cyclostationarity is presented, and canonical correlation analysis is explained briefly. The SCORE method for wide-band array data is from the perspective of canonical correlation analysis. Results of computer simulations are presented to illustrate the performance of the method and to compare it with that of the minimum-variance distortionless response beamformer