Self-critical detection of signals: narrowband and stochastic signal representations

Abstract

The concept of self-critical locally optimum receivers (SCLORs) developed for lowpass signals are extended to the detection of known and stochastic signals in lowpass and narrowband noise. It is shown that the self-critical locally optimum detectors for stochastic signals implement a statistic which depends on an adaptive weighting factor and on the conventional zero-memory non-linearities for both known and stochastic signal detectors. The weighting factor, which depends on the probability density function of the prevailing noise, is responsible for the self-criticism of the detector. It is shown that self-critical detectors are most appropriate for impulsive noise environments.