Performance analysis of generalized modified order statistics CFAR detectors

Abstract

Generalized order statistic cell averaging (GOSCA), generalized order statistic greatest of (GOSGO), and generalized order statistic smallest of (GOSSO) CFAR detectors are proposed. Each of them has its own advantages according to radar environment situations so that most effective radar detector can be chosen at any unpredictive situation. Their performance formulas in terms of the false alarm probability and detection probability are derived. From the performance analysis, the GOSCA CFAR detector is the best in a homogeneous situation, the GOSGO CFAR detector in clutter region near the clutter edges, and the GOSSO in the clear region close to the clutter boundary and interfering targets situation. A new window structure to eliminate the fatal problem in GOSSO CFAR detector in clutter regions is also proposed. The false alarm probability of the proposed window structure performance is compared with conventional window structure.