A SIRV-CFAR adaptive detector exploiting persymmetric clutter covariance structure

This paper deals with covariance matrix estimation for radar detection in non-Gaussian noise modeled by spherically invariant random vector (SIRV). In many applications, it is possible to assume a particular structure for the clutter covariance matrix: this is the case for instance for active systems using a symmetrically spaced linear array or pulse train. In this paper, we propose to use the particular persymmetric structure of the matrix to improve performance in term of detection. In this context, we provide a new adaptive detector and derive its statistical properties as well as its statistical distribution. Moreover, the high improvement of its detection performance is demonstrated on experimental ground clutter data.