Spectrum Sensing Method Of Measurement And Control Link In Complex Electromagnetic Environment

With the development of aerospace technology,more and more communication equipment is begin to appear,which makes the types of electromagnetic interference more and more complex.Due to the characteristics of the electromagnetic environment,electromagnetic signals present a state of interweaving in time and space,which may lead to the problem of mutual interference when carrying out different communication tasks in space.In addition,the scramble for spectrum resources and all kinds of artificial interference also make the interference signal and transmission channel more complex.Aerospace TT & C system is a key link in the development of aerospace and aviation,and its security communication is related to the development of relevant research and the guarantee of national defense security.Faced with the problems of complex interference and aliasing in frequency band,aiming at the situation of large dynamic range of interference to noise ratio and low interference to noise ratio,it is necessary to analyze whether there is interference in the frequency band and the intensity of interference.In this paper,the deep learning method is used to study the spectrum sensing and interference signal strength estimation method.Firstly,the measurement and control frequency band and its electromagnetic environment are analyzed,and the interference forms in the environment are studied and simulated.Combined with the range of interference to noise ratio and interference aliasing,the data set needed by the project is generated.Secondly,the neural network method is introduced into the spectrum sensing technology for analysis.In order to improve the performance of the energy sensing algorithm and the cooperative sensing algorithm based on the maximum eigenvalue and energy,the spectrum sensing methods based on convolutional neural network and cyclic neural network are studied respectively.Meanwhile,the advantages of the two networks are combined,and an improved feature fusion neural network CNN-GRU is proposed.The results show that the spectrum sensing method based on neural network can achieve more than 90% detection probability in the range of-12 d B INR and above,and the detection probability of fusion neural network in the range of-15 d B INR and above can reach more than 90%.Finally,the interference intensity estimation method based on neural network is studied.The interference to noise ratio estimation method based on covariance matrix eigenvalue and M2M4 interference to noise ratio estimation method have low accuracy in the case of low interference and multiple interference,and are more sensitive to signal types.Therefore,the method of interference to noise ratio estimation is studied by using the advantage of feature extraction of neural network.This method improves the estimation accuracy of interference to noise ratio significantly,and has little difference for different signal types,The improved fusion neural network improves the estimation effect on the basis of single network.