Space-time adaptive radar detection of distributed targets

This paper deals with radar detection of spatially distributed targets embedded in Gaussian noise with unknown covariance matrix. Precisely, noise returns are modeled as Gaussian vectors independent from range gate to range gate; they share the structure of the covariance matrix, but have possibly different power levels. A set of secondary data, free of signal components, is used to estimate the correlation properties of the disturbance. No a priori knowledge about the spatial distribution of the target scatterers is assumed. An adaptive detector is conceived and its performance is evaluated also in comparison with a previously proposed detection structure. Remarkably, the proposed receiver ensures CFAR-ness with respect to either the structure of the covariance matrix and the power backscattered from each cell. A performance assessment confirms its validity to operate in a heterogeneous environment.