ORDER STATISTIC SIGNAL PROCESSOR FOR ESTIMATING VALUES OF SIGNAL PARAMETERS AND DEMODULATING AMPLITUDE-, PHASE-, AND FREQUENCY- MODULATED SIGNALS

The order statistic signal processor obtains continuing estimates of amplitude, phase, or frequency of an input signal based on order statistics. By way of definition, the P'th order statistic for a set of samples from a continuously distributed process is the P'th largest of the samples. Where a signal parameter fluctuates with time as a result of contamination by noise and interference and where the noise fluctuation rate is much greater than the rate at which changes in a signal parameter occur, an order statistic associated with a set of signal parameter samples can serve as an estimate of the signal parameter. The order statistic signal processor is based on an iterative process whereby the prior estimate of a signal parameter, i.e. amplitude or phase, is subtracted from each of a plurality of signal parameter samples obtained over a period of time. The differences are converted to quantities of fixed magnitude with signs corresponding to the differences. The converted differences are summed, the sum is biased, and the biased sum is then integrated to provide a new estimate of the signal parameter. The signal parameter estimate obtained in this way corresponds to a particular order statistic for the signal parameter sample set that is obtained each time the process repeats. A particular order statistic is selected by adjusting the bias level.