Persymmetric Adaptive Detection With Reduced-Dimension Approach

Abstract

For target detection in Gaussian noise with unknown covariance matrix, usually sufficient training data are needed to form a nonsingular estimate of the noise covariance matrix. However sufficient training data could not be satisfied in practice. Aiming to design a detector with small training data size, this letter proposes a reduced-dimension detector by incorporating the persymmetric structure of received data. Based on the persymmetric structure, a reduced-dimension approach of subspace transformation-based detector is adopted. Then, the criterion of the generalized likelihood ratio test is used to design the detector. The proposed detector is shown to possess the constant false alarm rate property with respect to noise covariance matrix. Compared with existing detectors, such as conventional polarization-space-time generalized likelihood ratio detector, persymmetric polarization-space-time generalized likelihood ratio detector and subspace transformation-based detector, simulation results demonstrate that the proposed one has better detection performance, especially in the circumstance of limited training data.