Bayesian parametric GLRT for knowledge-aided space-time adaptive processing

In this paper, the problem of detecting a multichannel signal in the presence of spatially and temporally colored disturbance is considered. By modeling the disturbance as a multichannel auto-regressive (AR) model and treating the spatial covariance matrix as a random matrix, a parametric generalized likelihood ratio test (P-GLRT) is developed based on a Bayesian framework. The resulting P-GLRT, which is denoted as the knowledge-aided P-GLRT (KA-PGLRT), employs a fully Bayesian principle and performs a jointly spatio subtemporal whitening process. The KA-PGLRT detector is able to utilize some prior knowledge through a colored loading step between the prior spatial covariance matrix and the conventional estimate of the P-GLRT. Simulation results verify that die KA-PGLRT detector yields better detection performance over other parametric detectors.