Noise variance estimation based on order statistics in channelized receiver

Traditional method of estimating noise variance with signal present in channelized receiver utilizes quiet channels. Today's EM environment has high signal density and many EM signals, which disable us to find a channel being quiet long enough to obtain enough noise samples. Therefore we proposed a method based on order statistics. New subsets of noise samples are generated by sorting channelized outputs by amplitude at every sample time. After examining the statistical characteristics of the new subsets, an estimator of noise variance is derived by estimating parameters of the subset of minimum samples. The method in essence is a better way to use quiet channels to estimate noise. This method works when there are quiet samples for a sample period compared with hundreds of sample periods which conventional method needs.