先验失配条件下的知识辅助检测器稳健性分析

充分利用探测环境的先验信息是提高雷达探测能力的有效途径之一。先验信息必须在雷达检测算法设计阶段确定下来,因此先验信息与当前探测环境之间可能存在不一致性。以复合高斯杂波中的、利用纹理分量先验信息的知识辅助(KA)检测器作为研究对象,首先建立了该检测器检测性能与先验分布参数失配之间的量化关系,然后根据给定的杂波探测环境模型参数,分析了先验模型失配对检测性能的影响。分析结果表明:知识辅助检测器的稳健性与当前探测环境模型参数有关。进一步给出了先验模型失配的容许区间,当先验模型参数在这个区间内,知识辅助检测器性能优于不使用先验信息的检测器性能。

Robustness analysis of knowledge aided detector under prior mismatched conditions

Smart use of the prior information of detection environment is one of effective approaches to improve the performance of radar detection. Prior information is given at the design stage of radar detector, and the resultant difference between the prior model and current detection environment may be existed. In this paper, the knowledge aided (KA) detector using the texture component prior information in compound Gaussian clutter is considered. The quantized relationship between the detector performance and the mismatched prior model parameters is proposed, and the impact of the mismatched prior model on the detector performance is analyzed, with given detection environment model. The analyzed results show that, the robustness of knowledge aided detector is related to the model parameters of current detection environment. Furthermore, we propose the allowable region of the prior model mismatches, and when the prior model parameters are located in this region, the knowledge aided detector outperforms the convention detector without using prior information.