Optimum Filtering of Cyclostationary Processes Resulting from Scanning or Sampling of Communication and Control Signals.
MASSACHUSETTS UNIV AMHERST DEPT OF ELECTRICAL AND COMPUTER ENGINEERING
Pagination or Media Count:
Random signal processes which have been subjected to some form of repetitive operation such as sampling, scanning or multiplexing will usually exhibit statistical properties which vary periodically with time. Specifically, a random process, xt, for which the mean, Ext, and the autocorrelation, Ext tau x t, are both periodic in t is termed a cyclostationary process. It is shown that many of the signal processes typically encountered in communication systems and control systems are cyclostationary, a fact often ignored in the analysis of such systems. Using the more accurate cyclostationary model for such processes, a more precise evaluation of system performance can be obtained and signal processors giving superior performance can be designed. In particular, general procedures are developed for finding the optimum periodically-time-varying filters for the continuous waveform estimation problem and the improvement in performance is demonstrated over the best time-invariant filter.