Communication Station Signal Fingerprint Identification Technology Research

Communication Station Signal Fingerprint Identification Technology through different radio equipment due to differences in hardware performance of the signal on the launch signal fingerprint information to distinguish and identify the signal to determine the signal from which device to achieve the device monitoring,tracking and so on.In this paper,different types of the same operating mode of the communication station,from a number of perspectives of transient signal fingerprinting methods,through the unstable conditions in the radio to obtain transient signals,the purpose is to build effective signal fingerprint features.The classifier model is designed to achieve the resolution and identification of different communication station signals.The main contributions and innovations of the dissertation are as follows:(1)A transient signal extraction method is proposed.The method includes three parts:signal preprocessing,transient signal starting point positioning,and transient signal end point positioning.In the signal preprocessing part,a filter is designed to filter the received signal.The starting point detection method based on short-term energy detection was studied in the detection part of transient signal start point.The signal is extracted from the received signal by the method of transient signal extraction Transient signal.(2)Research on transient signal fingerprint extraction method.Firstly,the high-order moment R/J features of the Hilbert transform envelope of transient signals are studied from the time-domain envelope.The experimental results of the measured signal show that the R/J feature is highly effective at high signal-to-noise ratio but has poor anti-noise and interference capability.Then,a method based on complex Morlet wavelet for feature extraction of transient signal envelop fingerprint is proposed.This method performs polynomial fitting on the signal envelope extracted from the wavelet and uses the coefficients of the fitted polynomial as the signal fingerprint feature.The experimental test of the measured radio signal shows that the fingerprint feature has higher efficiency and stronger anti-noise capability.Finally,the method of time-frequency analysis for extracting time-frequency energy distribution characteristics of transient signals as transient signal fingerprints is studied.The experimental results show that this feature satisfies the uniqueness and invariability of the transient signal fingerprint,has a high degree of discrimination for different radio signals and is less affected by noise.(3)Combining the advantages and disadvantages of various features,this paper proposes a fusion fingerprint feature as a classification basis for the recognition of radio signals.Experimental results show that for different radio signals,the fused fingerprint features have stronger discrimination recognition ability than a single feature.In addition,an integrated classifier model is designed based on SVM and ensemble learning method.Validate the performance of the integrated classifier through experiments,Experimental results show that the integrated classifier model has stronger classification recognition and anti-noise performance than a single classifier.