基于神经网络和SVM的GPS干扰类型识别

全球定位系统（GPS）干扰信号类型的识别是采取有效抗干扰手段的先决条件。针对7种典型的GPS干扰信号，提取了包括高阶统计量在内的8个特征，设计了反向传播（BP）神经网络分类器和多项式支持向量机（SVM）分类器，实现了干扰信号类型识别。仿真结果表明，两种分类器均具有较高的正确识别率和较好的热噪声鲁棒性，特别是在干噪比（JNR）为3dB时，平均正确识别率可保持在94％以上。

Recognition of GPS interference signals based on neural network and SVM

Global Positioning System（GPS） interference recognition is the prerequisite of effective anti-interference measures. In this study, eight features, including higher-order statistics, were extracted, and Back Propagation（BP） neural network classifier and polynomial Support Vector Machine（SVM） classifier were employed for recognition of seven typical types of GPS interference signals. Simulation results demonstrated that both classifiers performed well with high recognition rates and robustness when thermal noise existed, and when Jammer-To-Noise Ratio（JNR） was 3 dB, the average recognition rates still remained above 94%.