Blind and informed cyclic array processing for cyclostationary signals

Limiting the analysis to the exploitation of the second order statistics of the complex data, the optimal Spatio-Temporal (ST) receivers in stationary contexts are Linear and Time Invariant (TI). However, for (quasi)-cyclostationary observations, it is now well-known that the optimal ST complex receivers become (poly)-periodic (PP) and, under some conditions of non circularity, Widely Linear (WL). Using these results and the fact that PP filtering is equivalent to FREquency SHifted (FRESH) filtering, the purpose of this paper is to present a new ST, PP and WL receiver structure, very useful for applications such as passive listening or source separation, taking into account the potential (quasi)-cyclostationarity and non circularity properties of the observations. This new cyclic receiver may be implemented either blindly or from the a priori knowledge or estimation of the useful signal steering vector. The performance computation of this new cyclic receiver shows off the great interest of the latter in cydostationary contexts and its great capability of interferences rejection even from a one sensor reception.