Research On Beidou Navigation System Interference Signal Recognition Technology

Aiming at the problem of jamming signal recognition in Beidou navigation system,this thesis takes Beidou B1I signal as the main navigation signal,and after adding interference signal and Gaussian white noise,simulates the received signal at the receiving end of the navigation satellite,and carries out simulation experiments under the condition of jamming signal-to-noise ratio（JSNR）.Because the jamming identification method based on feature extraction is not particularly ideal under the condition of mixed jamming signal-to-noise ratio（JSNR）,this thesis applies the convolutional neural network model to the recognition of jamming signals.In this thesis,the B1I signal is superimposed with eight common jamming signals,and the mixed Gaussian white noise is used to model the signal.The characteristic parameters of the received signal are extracted from the time domain and frequency domain respectively.The characteristic parameters include the following eight kinds:time domain kurtosis factor（k_u）,time domain waveform factor（W_B）,time domain peak factor（C）,time domain pulse factor（I）,time domain margin factor（L）,carrier factor coefficient（G）,normalized average flatness coefficient(F_se),time domain moment partial mean（α）etc.The jamming signal recognition technology based on feature extraction is studied,and three classifiers based on traditional machine learning methods are introduced,including CART decision tree classifier,weighted K-nearest neighbor classifier and BP neural network classifier.Then,according to the shortcomings of BP neural network,an ISSA-BP interference recognition classifier is designed by introducing sparrow algorithm optimization.The experimental results show that the recognition accuracy of ISSA-BP neural network classifier can reach more than 99%under the condition that the jamming signal-to-noise ratio is greater than or equal to 5 d B.Compared with the traditional jamming recognition classifier,it can recognize the jamming signal earlier.When mixing all the samples under the condition of jamming signal-to-noise ratio,the recognition accuracy of each classifier is seriously reduced due to the increase of the number and complexity of the samples,which are 81.6%,89.3%,73.6%and 76.5%respectively.In addition,according to the shortcomings of traditional methods,this thesis designs a convolutional neural network model for interference recognition,which directly normalizes the time domain information of the signal and inputs it into the network for classification.The simulation results show that when the jamming signal-to-noise ratio is-1 d B and above,the interference recognition accuracy of the model can reach more than99%.When mixing all samples,the jamming recognition accuracy is still 94.5%.In contrast,the jamming recognition accuracy of this method is higher than that of the feature extraction method under the same or mixed JSNR conditions.