Performances des détecteurs CFAR dans des environnements gaussiens et non-gaussiens

Abstract

One of the most important problem in radar systems is the automatic target detection while maintaining the probability of false alarm at a desired value. In this thesis, we analyzed the performance of different adaptive detectors CFAR (Constant False Alarm Rate) in the presence of Gaussian clutter and Non-Gaussian (Weibull) clutter for homogeneous and non-homogenous situations (presence of interfering targets or clutter edge in the reference window). In the first contribution of this work, we analyzed the performance of the CA, GO, SO, OSCAGO SOSCA-CFAR detectors and binary integration in a homogeneous and non-homogeneous Weibull environment. We derived closed form expressions for the Pfa of different detectors in the homogeneous case. Simulations in the case of a homogeneous and nonhomogeneous environment, were presented and discussed. In the second contribution, we analyzed the detection performance of the GOSCA, the OSGO and the OSSOCFAR for the concept of MIMO radar in homogeneous and non-homogeneous Gaussian environments. Three schemes have been proposed to generalize these detectors for MIMO radars. We derived closed form expressions. The results for homogeneous and non-homogeneous environments were presented and discussed. In the third contribution, we analyzed the detection performance of the fuzzy detectors FCA-CFAR, FGO-CFAR and FSO-CFAR for a distributed system using fuzzy fusion rules in a homogeneous and non-homogeneous Weibull environment. We derived closed form expressions of the membership functions of each detector. The results are shown and discussed.