Time-frequency estimation for cyclostationary signals

This paper presents estimators of the Wigner-Ville spectrum for wide sense cyclostationary (WSCS) processes. It is shown that the smoothing kernel is a mixed continuous time/discrete time function under the cyclostationary assumption. For estimators of WSCS processes it is shown that the tradeoff between variance reduction and blurring of the nonstationary features of the process does not exist. The Cohen´s class smoothing kernel is given for finite observation lengths of the process, and smoothed pseudo-Wigner distribution type kernels are illustrated in examples