Detection of multiple signals by the significance test

A description is presented of the significance test approach to statistical detection, and its form for the detection of multiple signals in Weibull clutter is derived. The significance test is a statistical measure indicating how likely it is that N multiple observations were generated by the underlying noise model, i.e. the case in which no signal is present. The less likely the observations are to have been generated by the noise the more likely they are to have been generated by the signal. The actual measurements are the joint tail probabilities associated with the magnitude of N envelope values. The authors derive the distribution function of the test statistic, from which they are able to calculate false alarm probability as a function of threshold setting. The performance of the significance test is compared to that of the linear, square law, and M-out-of-N detectors. The authors propose a method for combining the most desirable properties of the significance test and M-out-of-N techniques