Maximum likelihood and common factor analysis-based blind adaptive spatial filtering for cyclostationary signals

A blind adaptive spatio-temporal filter for unknown cyclostationary signals in noise is derived in two ways by maximizing a constrained conditional likelihood function and by solving a common factor analysis problem. It is shown that specific choices of free constraint-parameters within the resulting beamformer structure yield the existing cross spectral self-coherence restoral (SCORE) algorithms and conjugate cross-SCORE algorithms which blindly adapt and antenna array to extract signals having specified cyclostationarity properties from interference and noise. It is also show that other choices can substantially improve the convergence properties of SCORE algorithms.<>