Research On Subtle Feature Recognition Of Wireless Transmitter Signal

The research on the subtle features of wireless transmitter signal plays a very important role in communication security and military applications.Mainly through the wireless transmitter wireless communication signal analysis, in order to find the stability of the identification parameters to complete the study of subtle characteristics of the signal. These devices are mainly in the form of pulse modulation to send and receive signals, so the main study is reflected in the characteristics of the extraction of pulse parameters on the content of the work. Intra-pulse parameter characteristics include intentional modulation features and unintentional modulation features. Both of these modulation features are likely to be key features in identifying the emission source. Due to the uncertainties of the intentional modulation feature in the communication process, it is not robust to identify the signal by extracting the intentional modulation parameter feature recognition, which is the reason why most scholars choose the unintentional modulation feature with stability and uniqueness as the identification parameter. Considering the signal recognition in practical application, this paper extracts the envelope feature which may contain the unintentional modulation feature and the instantaneous frequency characteristic which contains the unintentional modulation feature as the signal fingerprint. Based on the characteristics of the data sample, the support vector machine method is used to design the nonlinearity Classifier,and then completed the multi-class signal identification work.In this paper,the flow chart of the signal identification system of wireless transmitter is given. In order to extract the fine features, we must complete the communication pulse extraction, and then complete the pulse signal analysis and extraction of fine features, the final design of the classifier to identify the signal. The research results of each part are as follows:1) For the pulse signal extraction, this paper uses the energy function rough detection and the variance fractal dimension precision positioning method, and realizes the pulse signal interception efficiently and accurately.2) In order to extract the fine features, this paper combines the signal component separation method and the WVD method with HHT to extract the true edge characteristics of single-signal time-frequency information successfully, and effectively extract the time-frequency characteristic which can reflect the subtle characteristic uniqueness of the wireless transmitter feature. In addition, the envelope feature of the differentiable transmitter signal is extracted by using the least squares polynomial model.3) According to the design of the classifier and the characteristics of the fine feature samples, the signal samples are constructed based on the time-frequency characteristics, envelope characteristics and signal labels.In this paper, the multi-class classifier design of the signal is completed by the OVO method based on support vector machine. With the similar model of multi-category civil aircraft wireless transmitter data, with less than 40%of the training test data ratio to achieve a 90% accuracy rate, completed the wireless transmitter subtle feature recognition research.