Performance Amelioration of Standard Variants of Adaptive Schemes Operating in Heterogeneous Environment

摘要

The ruggedness of emerging single adaptive approach that performs well in all types of operating conditions has led to the development of composite adaptive strategy. In this regard, the fusion of particular decisions of single adaptive schemes through suitable fusion rules can provide a better final detection. Particularly, the fusion of cell-averaging (CA), ordered statistics (OS) and trimmed-mean (TM) procedures can enhance the overall detection performance. Our goal in this paper is to analyze this developed model when the operating environment is heterogeneous. A χ~(2)-distribution with two- and four-degrees of freedom is assumed for the fluctuation of primary and secondary extraneous targets. A closed form processor performance is derived for single pulse detection. The results show that for the non-homogeneous background the new approach is more practical. Particularly in multitarget situations, it exhibits higher robustness as compared to the CA, OS, or TM architectures. Additionally, the novel strategy has a homogeneous performance that surpasses performance of the classical Neyman-Pearson (N-P) detector, which can be employed as a yardstick for the analysis of different techniques in the CFAR world.