Bayesian Subspace Signal Detection in Partially Homogeneous Environment With Insufficient Training Data

Partially homogeneous environment (PHE) is a typical non-homogeneity that radar systems frequently experience. Because there are generally insufficient training data to generate a reliable estimate for the unknown covariance matrix, signal detection in PHE can be challenging. In order to address this issue, we apply Bayesian theory in this paper by assuming that an inverse Wishart distribution governs the noise covariance matrix. Subsequently, we introduce an efficient Bayesian detector. The results of numerical examples demonstrate that the proposed detector has significantly better detection performance.