Typical Communication Interference Signal Identification Method Based On Improved SVM And DBN

In the current electronic information warfare,communication jamming methods are complex and diverse.If different jamming signal types can be quickly and accurately identified,corresponding measures can be taken to ensure one's own communication,which is of great significance to modern electronic information warfare.Traditional machine learning algorithms have been applied to the classification and recognition of typical communication interference signals.With the continuous development of machine learning theories and methods,more and more scholars have carried out research work on corresponding improved algorithms.The main research contents of this paper are as follows:1.Research on the classification and recognition methods of typical communication interference signals based on characteristic parameters and improved Support Vector Machine(SVM).Six feature parameters are extracted from the interference signal to construct a six-dimensional feature space;in view of the problem that the penalty coefficient and the kernel function coefficient in the traditional SVM algorithm need to be selected manually,which causes the recognition accuracy to decrease,the improved gray wolf algorithm is used to find the parameters of the SVM Excellent;input the feature matrix into the improved SVM,and the simulation experiment verifies that the improved SVM has better classification and recognition performance than the traditional SVM,especially under the low signal-to-noise ratio,the interference signal recognition rate of the improved SVM is significantly improved.2.Research on the classification and recognition methods of typical communication interference signals based on automatic learning features of Deep Belief Networks(DBN).Designed a DBN network that satisfies the identification of interference signals,carried out DBN network model parameter optimization experiments,and verified the good predictive learning ability of DBN for the classification and identification of interference signals through simulation experiments;on this basis,a multi-dimensional scaling-depth confidence network was proposed(Multidimensional Scaling-Deep Belief Networks,MDS-DBN)method,the training prediction time of the network is reduced,and the classification and recognition accuracy is improved,which verifies the effectiveness of the MDS-DBN method.3.An experimental test platform based on GNU Radio software and USRP hardware was built to verify the effectiveness of the improved SVM and MDS-DBN for the identification of typical communication interference signals and the practicality of engineering,and further improve the classification and identification of typical communication interference signals.