In this paper we derive the optimum linear algorithm for the separation by an array of a given number of cyclostationary signals of the type encountered in digital communications. Spatially and temporally uncorrelated noise is considered at the input to each sensor. It is proved that the optimum signal separator can be expressed as a matrix transfer function in the z-domain. It is intended that this algorithm be used as a benchmark to test the performance of adaptive schemes. Therefore, bounds on the noise estimation covariance are obtained assuming perfect knowledge of the number of sources as well as of their timing and DOA.

Peer Reviewed