非高斯混响背景下一种自适应CFAR检测器的研究

在浅海波导中混响是主动声纳系统的主要干扰，为了抑制混响，通常采用宽带发射信号和大孔径发射、接收阵。这两种方法都减小了混响的分辨单元．因此经波束形成和匹配滤波后提高了信混比。然而这也带来了负面影响，即分辨单元内混响包络的分布偏离了传统假设的瑞利分布，出现了高尾的非瑞利分布，导致匹配滤波器虚警概率提高。为了得到恒虚警检测器，需要对混响进行更精确的建模。通过理论和实际数据分析表明，利用多变量椭圆轮廓分布可以很好的建模浅海混响的统计特性。在混响建模为多变量椭圆轮廓分布模型的基础上，利用GLRT理论可以得到一种自适应CFAR检测器，通过实验数据分析，自适应CFAR检测器较之常规匹配滤波器，其信混比（SRR）可提高3dB。

An adaptive CFAR detector in non-Gaussian reverberation environment

Active sonar systems operating in shallow water are most significantly hindered by increased reverberation power levels. To offset the resulting loss in detection performance, systems have been developed with wider transmission bandwidth and larger aperture transmitting and receiving array. Both of these approaches reduce the size of the resolution cell from which reverberation results and therefore, increase the signal-to-reverberation power ratio after beamforming and matched filtering. However, increasing bandwidth or array size can have the adverse effect of producing reverberation with a statistical distribution that is more heavy-tailed than the traditionally assumed Rayleigh distribution at the matched filter envelope to cause an increase in the probability of false alarm. Reverberation＇s statistical distribution can be modeled well by using the Multivariate Elliptically Contoured distributions. Here, an adaptive CFAR detector, based on the reverberation MEC distributions, can be derived form the GLRT theory. By analyzing the experiment data, it is found that the detection performance can be enhanced by about 3 dB in reverberation environment.