A GLRT Detector for Correlated Targets

Paper in proceeding, 2025

The detection problem for correlated targets is considered. A target model is employed that encompass a time varying target amplitude described as a correlated Gaussian stochastic process. The likelihood ratio test (LRT), Detection of Correlated Targets (DCT) depends on the target amplitude variance and we present a method to derive the variance and arrive at a generalized LRT-statistic called the Generalised Detector for Correlated Targets (GDCT). We show that the LRT test statistic for the problem converges to the classical matched filter as the correlation time goes to infinity and to the energy detector as the correlation time goes to zero. An example mimicking a moving target detection problem is presented where the result illustrates that the GDCT performance is only slightly inferior to the correctly specified DCT and is superior if the DCT has a miss-specified target amplitude variance.