No abstract

This article aims to study the behavior of Constant False Alarm Rate (CFAR) detectors for a heterogeneous Weibull clutter and its derivatives. CFAR architectures based on exploitation of the Combined Environmental Knowledge Base (CEKB) have been proposed, called Knowledge Based Systems-Maximum Likelihood-CFAR (KBS-ML-CFAR) and KBS-Log-t-CFAR for nonhomogeneous Weibull clutter at general parameters. A CFAR architecture that uses Geographic Information System (GIS) as a Knowledge Base (KB), called KBS-Forward Automatic Order Selection Ordered Statistics-CFAR (KBS-FAOSOS-CFAR) has been proposed for special Weibull parameters. The performances of the proposed detectors have been studied and analyzed by conducting MATLAB simulations. The simulation results show that the KBS-CFAR based on CEKB outperforms the ML and Log-t-CFAR in terms of clutter edge detection capability in nonhomogeneous Weibull clutter case. Compared with other KB, this KBS-CFAR based on CEKB performs well to preserve the probability of false alarm (P f a ) at a desired constant value. For special Weibull parameters, the proposed KBS-FAOSOS-CFAR based on GIS performs better than KBS-Dynamic-CFAR and KBS-Adaptive Linear Combined-CFAR (KBS-ALC-CFAR) in severe interference case. CFAR techniques have been implemented on the ADSP (Advanced Digital Signal Processor) processing board, and the results have been evaluated and discussed.