基于子块积分循环双谱的雷达辐射源信号识别

Radar Emitter Signal Recognition Based on Sub-Block Integral Cyclic Bispectrum

循环双谱性能优越,能用于雷达辐射源信号的识别,但是其数据量庞大。循环双谱对角切片法丢失了大部分信息。证明了循环双谱的对称性和周期性,在此基础上提出了子块积分循环双谱,该方法不但能有效地减小数据量,而且保留了大部分的循环双谱信息。然后用Fisher判决率(FDR)选择鉴别能力较强的子块积分循环双谱。仿真条件下,对比分析了子块积分循环双谱与循环双谱对角切片法的识别效果,结果表明新方法的识别性能远优于对角切片法。

Cyclic bispectrum could be used as a feature to recognize the radar emitter signal for its excellent properties such as noise immune and information abundant,but it has large amount of data.The existing cyclic bispectrum diagonal slicing,technique loses too much valuable information.This paper proves the symmetry and periodicity of cyclic bispectrum,and then proposes the Sub-block Integral Cyclic Bispectrum（SICB） method.This new method can not only reduce storage requirement effectively,but also retain most of the cyclic bispectrum information.Fisher discriminant rate（FDR） is used for feature selection.Under simulated conditions,the recognition performance of this SICB method is compared with that of the diagonal slicing method.The result indicates that SICB is superior to the diagonal slicing method.