基于BP神经网络的导航信号欺骗干扰检测

Navigation signal spoofing interference detection based on BP neural network

全球导航卫星系统的开放服务给生命安全应用程序带来了巨大便利的同时,也存在许多隐患,例如缺乏安全防护以及信号的脆弱性,极易被恶意用户进行欺骗干扰,而这类问题的干扰不同于压制式干扰容易被检测出来,其隐蔽性极高,危害性极大,难以被一般的仪器和算法检测出来。针对这些问题提出一种基于BP神经网络的有监督的机器学习方法进行欺骗干扰检测,BP神经网络是按照误差反向传播的多层前馈神经网络,采用梯度下降法计算目标函数最小值,采用伪距、载波相位、多普勒频移、时钟频漂和信噪比等观测值进行神经网络训练,再将新的信号输入训练好的神经网络进行分类测试,从接收者操作特征曲线的结果看出,此方法的分类效果达到83%,说明此方法具备较高的检测概率,可以进一步研究。

While the open service of the global navigation satellite system(GNSS)brings great convenience to life safety applications,there are many hidden dangers,for example,the lack of security protection,the signals are vulnerable and prone to being deceived and interfered by malicious users. This type of interference,however,unlike suppressed interference which is easy to be detected,is extremely concealed and harmful,which makes it difficult to be detected by general instruments and algorithms. In view of the above,a supervised machine learning method based on BP(back propagation) neural network is proposed to perform spoofing interference detection. The BP neural network is a multi-layer feedforward neural network with error back propagation(EBP). As for the BP neural network,the gradient descent method is used to calculate the minimum of the objective function. The observations of pseudorange,carrier phase,Doppler shift,clock frequency drift and signal-to-noise ratio(SNR)are used to train the neural network. And then,the new signal is input into the trained neural network for classification test. From the results of receiver operating characteristic(ROC) curve,it can be seen that the classification effect of this method reaches 83%,which indicates that it has high detection probability and can be further studied.