非高斯背景下基于Sigmoid函数的信号检测

Signal Detection Based on Sigmoid Function in Non-Gaussian Noise

针对非高斯背景下的弱信号检测问题,该文提出一种基于Sigmoid函数的信号检测(SFD)方法。首先依据混合高斯模型对非高斯背景建模,在此基础上系统研究了参数k与SFD的检测性能以及检测特性的关系,确定了k的最佳的取值,并指出SFD在检测性能达到最优的同时也具有恒虚警特性。其次通过固定k值得到了一种新的非参量检测方法,较传统的匹配滤波性能有明显提升。最后进行仿真分析验证了SFD的有效性和优越性。

To solve the problem of weak signals detection in non-Gaussian background,a method based on Sigmoid function is proposed which is named Sigmoid Function Detector(SFD).Firstly,the non-Gaussian background is modeled as a mixed Gaussian model.Based on this,the relationship between parameter k and SFD's performance and characteristics are systematically analyzed.It is pointed out that SFD will be a constant false alarm detector when its detection performance is optimal.Secondly,a new non-parametric detector is proposed via fixing the parameter k,which has significant improvement over matched filter.Finally,simulation analysis is carried out to verify the effectiveness and superiority of SFD.