Adaptive Detection with Constant False Alarm Ratio in A Non-Gaussian Noise Background

A class of adaptive detectors with constant false alarm ratio (CFAR) for weak signals’ detection in additive non-Gaussian noise background is investigated. Although locally optimal detector (LOD) has the optimum detection performance, there are some disadvantages, such as complicated detection structure, poor adaptability, and difficulties in achieving CFAR in a time-varying noise background. In order to solve this problem, a cumulative distribution detector (CDD) is first proposed. However, CDD requires an accurate estimation of the cumulative distribution function for the non-Gaussian background, which limits its application. Therefore, a detector based on sigmoid function (SGD) is described. The performance of SGD is analyzed in detail, and then, an optimal SGD is obtained. In addition, by taking a simple approximation of that, a CFAR detector based on SGD is finally obtained, and the simulation results show that it has similar detection performance compared with LOD but provides relatively high efficiency and adaptability.