Research On Key Technologies For Fingerprint Identification Of Communication Equipment

The rapid development of communication and computer brings many application scenarios to the fingerprint identification technology of communication device.This technology has important significance in the fields of electronic countermeasures and wireless network security,and has been widely studied in recent years.The fingerprint identification technology of the communication devices is able to identify devices via build a relationship between the devices and the signal emitted by the devices.The relationship is mainly based on the fingerprint features of the signal emitted by the device that are different from those of other devices.In this thesis,serval key technologies of the communication device fingerprint identification are studied,including signal acquisition,signal feature extraction and classifier design of communication device.The main work focuses on the fingerprint identification algorithm research of transient signal identification of device,fingerprint recognition of communication device based on ensemble learning and deep learning.All of the proposed algorithms are verified by the measured data,the main work includes the following aspects:(1)For the transient signal of communication devices,an algorithm that utilizes the short-term periodicity of the signal is given.In the process of mapping transient signals into probability distributions,the feature extraction method is improved by using the short-term periodicity of the intercom transient signals,and the difference between the two features is quantized by symmetric KL.The experimental results are shown in Among the 6 walkie-talkie signals measured,the average recognition rate exceeded 98%.(2)Feature extraction of steady state signals is achieved.The characteristics of selfencoder and four kinds of integral bispectrum features are analyzed and visualized.The experiments are carried out on the cell phone signal data in combination with Softmax regression model.Experiments show that the combined integral bispectrum features have better discrimination than other features.(3)Research on individual identification technology of communication devices based on Ensemble Learning.The individual identification scheme of communication devices based on eXtreme Gradient Boosting(XGBoost)classification model is studied.The stacking classification model is constructed,and the Gradient Boosting Decison Tree(GBDT)model with different parameters is used as the primary learner of the stacking classifier.The recognition result of the measured data is combined,and the recognition rate of the stacking classification model is improved by about 2% compared with the GBDT classification model.(4)The steady state signal recognition of mobile phone based on deep learning is studied.The Deep Neural Network(DNN)is designed and constructed to further feature extraction and recognition of the integral bispectrum features,and the output is compared with other classifiers.The results show that the average recognition rate of the six mobile phones after using the DNN structure is over 99%.