CFAR at the end of instrumented range

Abstract

The constant false alarm rate (CFAR) detection method is commonly used in radar signal processing systems and is described in textbooks. However, there is an implementation problem at the beginning and the end of range samples where data within one of the sliding average windows is missing. The common implementation method is to assume the values in the missing window to be zeroes. That causes the false alarm rate to increase at the two ends of the range samples. The problem is not normally noticeable unless weather clutter exists in those regions. As a result, abnormal false alarm rates will appear as "weather breakthrough". In this paper, two approaches are described as solutions to the problem of the common method. The clamped threshold algorithm freezes the last valid threshold and extends it to the range boundary. The second approach allows the sliding window to shrink as it is limited by the range boundary. The two new approaches and the common method are evaluated with real data and simulations. The results indicate that the clamped threshold algorithm is the preferred approach.