Research On Feature Extraction And Parameter Estimation Of Non-stationary Signals Based On HHT

In modern life, wireless communication network has become an important way of information transmission. However, its open transmission environment results in signals more vulnerable to eavesdropping and deciphering. Therefore, a vital part of wireless communication network evolution is how to ensure the security of information. In recent years, widespread attention has come to physical security techniques which focus on the signal waveform designing.To improve the security and reliability of wireless communication system, a wireless communication signal which lacks of cyclostationarity and ergodicity was designed. It means neither statistical mean or autocorrelation values of the signal observed in different moments or time intervals is the same. Moreover, the signal characteristics get more and more unpredicted as the number of observation data increasing. This means that statistical characteristics of the non-stationary signal cannot be estimated by long-term observation and analysis, thus achieving anti-reconnaissance and anti-interception capabilities. To describe and evaluate the properties of non-stationary signals with time-varying characteristic, time-frequency analysis methods are needed. Hilbert-Huang Transform(HHT) is a new time-frequency analysis method, which is a process of signal smoothing. In HHT, the frequency of original signal is characterized by instantaneous frequency.In this study, Chirp signal, a typical non-stationary signal, was analysized by HHT to analyze the detection performance of HHT. Then anti-detection capability of the proposed non-stationary communication signals were also verified by HHT. The main works of this thesis could be divided into the following sections:(1)Based on the understanding of the HHT theory, the definition of the DS(Degree of Stationarity) was studied. Simulation results proved that DS could quantitatively describe the stationarity of Chirp signal.(2)For some typical signals, the time-frequency analysis performance of HHT was analyzed and compared with several common time-frequency analysis methods, such as Short Time Fourier Transform(STFT), Wigner-Ville Distribution(WVD) and Wavelet Transform(WT). Simulation results demonstrated the excellent performance of HHT in describing the time-frequency characteristics of signal.(3)In the study of HHT-based parameter estimation of Chirp signals, the impact of sampling frequency, frequency value, crossing-components and noise on estimation performance was studied with two different processing algorithms. Simulation results showed that HHT could achieve good performance in estimating single component Chirp signal without noise.(4)To verify the anti-detection performance, the proposed non-stationary communication signals with different parameters were detected and analyzed based on HHT analysis. Simulation results proved that HHT has a good detection performance for the non-stationary communication signals with less time-varing carrier frequencies and without noise.