Adaptive Radar Detection for Extended and Distributed Targets without Assignment of Secondary Data

In this paper we attack adaptive detection of extended and multiple point-like targets, embedded in Gaussian noise, without assignment of secondary data; otherwise stated, we try to get rid of the usual assumption that a distinct set of data, sharing the spectral properties of noise in the cells under test, but free of signal components, is available. At the design stage we resort to the GLRT and to an “ad hoc” two-step design procedure, possibly exploiting an upper bound on the maximum range dimension for extended targets and on the maximum number of targets for multiple point-like ones. The proposed detectors have been compared based on prediction capabilities, CFARness, and computational complexity, also in comparison to existing adaptive algorithms.