Research On Interference Signal Perception And Characteristic Estimation Of Space TT&C Link

Space TT&C is widely used in military field,and satellite countermeasure has become the main field of space information investigation and countermeasure.It is of great significance to study the interference sensing and automatic identification of space TT&C link on this basis.The research on interference signal perception and identification have made a great achievement after years of development,but most of the methods is only able to detect and inhibit specific interference types,which has some limitation.At the same time,most of the feature extraction methods for interference type recognition are the traditional pattern recognition method that requires manual extracting characteristics of interference,which is of great complex.In this paper,we study the method of interference recognition and parameter estimation based on energy detection and convolutional neural network(CNN).The main research contents include interference signal pattern analysis,interference detection,interference classification and interference parameter estimation.Firstly,this paper analyzed the time and frequency domain wave form involved five kinds of interference signal,including audio interference,narrowband interference in the same band,swept frequency interference,spread spectrum interference and rectangular pulse interference.Secondly,this paper successfully detect the existence of interference signal through energy detection and CFAR algorithm.The research shows that the accuracy make a great achievement in a high JNR(jammer-to-noise ratio)environment,while the performance of algorithm degrade a lot when JNR reduced.So a combined detection between energy detection and CNN is published in this paper.The experiment result shows the CNN achieve a better result than energy detection in a low JNR environment.Thirdly,this paper extract 15 feature of 5 kinds of interference signal through CNN,then reducing the dimensionality of feature from 15 to 2 through multidimensional scaling(MDS)in order to facilitate the analysis.The result shows the extracted feature is robust and separable.Next we classify the 5 kinds of single interference and 15 kinds of simultaneous interference through Softmax classifier,the experiment shows that the classification accuracy of single interference reached nearly 100% and the simultaneous interference is nearly 96% in the-5dB~15d B JNR.Finally,we estimate the center frequency of narrow band interference end bandwidth of broad interference by Chebyshev filter and fourth order energy detection algorithm instead of second order energy detection.The result shows that the estimate accuracy achieve above 96% in the-5dB~15dB JNR.