Clutter edge detection and estimation of field parameters in radar detection

In this paper we propose three detectors which decide between non-homogenous and homogenous hypotheses of a Gaussian field. We derive the optimal Neyman-Pearson bound and also ALR, GLR and HLR tests for this problem. The simulation results show that the ALR detector is so close to the Neyman-Pearson bound and the performance of ALR and HLR detectors is better than that of GLR. These results show that the edge of non-homogeneousness can be approximately modeled by a discrete uniform random variable, (i.e., this assumption causes more signal processing gain than that of deterministic model). In this paper, we also derive an ML estimation of non-homogeneousness edge. Finally, we propose an extended CFAR method for radar target detection.