关键词: 双谱分析, 特征提取, 分离性, Bhattacharyya距离, 支持向量机

Abstract:

To deal with the problem of emitter identification caused by parameter complexity and agility of multi-function radars, a new identification method based on the bispectrum feature is proposed. This method takes the bispectrum of emitters as the specific emitter feature and the Bhattacharyya distance measurement as the cascade feature extraction algorithm. Bispectrum analysis can restrain the effect of Gauss White Noise on radar signal, and retain the amplitude and phase information. So this paper utilizes this characteristic to enhance the validity of the radar emitter signal feature. Aimed at the problem that the signal bispectrum contains many redundancy bispectrums, the Bhattacharyya distance is introduced to extract the bispectrum cascade feature with best separation as the radar signal feature. Experimental results show that the proposed method is suitable for extracting radar signal characteristics and has a good recognition performance. The average accuracy is not less than 95％ at the signal to noise ratio of a wide range.

Key words: bispectrum analysis, feature extraction, separation, Bhattacharyya distance, support vector machine