Representation and Estimation Techniques for Cyclostationary Random Processes

Many communication and control systems employ signal formats that involve some form of periodic processing operation. Familiar examples are signals produced by samplers, scanners, multiplexers, or modulators. Very often these signals can be modelled as cyclostationary processes, i.e., processes whose statistical properties, such as mean and autocorrelation, fluctuate periodically with time. Filters designed to extract signals of this type from a noise background can exhibit dramatically improved performance when the periodic nature of the statistics are taken into account, rather than using the more conventional time-average statistical approach. Some techniques for solving for the optimum filter and a video signal example are discussed.