OPTIMUM DISTRIBUTED CONSTANT FALSE ALARM RATE DETECTION OF WEAK SIGNALS

A set of sensor detectors receive observations that may contain a zero-mean random signal of unknown amplitude. These observations also contain additive narrow-band non-Gaussian noise-plus-clutter with an unknown scale parameter. A constant false alarm rate distributed signal detection scheme is proposed that is optimized under weak signal conditions. This scheme assumes that a set of noise-plus-clutter-only reference samples are available at each sensor, and that these are used to provide an estimate of the unknown scale parameter. Each sensor processor uses the locally optimum test based on observing the scalar quantity that results from dividing the observation under test by the sensor's scale parameter estimate. © 1995, Acoustical Society of America. All rights reserved.