Detection And Parameters Estimation Of Frequency Hopping Signals Based On Time-frequency Analysis

Frequency hopping(FH) communication is one of the main types of spread spectrum communication, because of its strong anti-interference ability, anti-interception ability, the capability against multi-path fading and other advantages, it has been widely used in both military and civilian communication field, but it also put forward severe challenges to communication reconnaissance at the same time. Time-frequency analysis(TFA) method is an effective tool to analyze FH signals that are typical non-stationary. Therefore, this dissertation focuses on the TFA methods which are applicable for FH signals, and then FH signals blind detection and blind parameters estimation are also researched. The main contents and innovations are given as follows:1. The global wave TFA methods suitable for FH signals are researched. The applications of various time-frequency representations(TFRs) for FH signals analysis are emphatically studied, and the performance of various TFRs is simulated and compared qualitatively in terms of time-frequency concentration and cross-term interference suppression capabilities. Then the performance of TFRs is compared quantificationally by utilizing the information entropy, and the computations of some typical TFRs are given to evaluate their advantages and disadvantages synthetically.2. The blind detection algorithms for FH signals are researched. Firstly, as the FH signals have texture features in the time-frequency diagram, the texture features can be extracted by gray level co-occurrence matrix(GLCM). So the background noise can be removed through the separation of texture features, and this method can be also used in the colored background noise. Then a procedure to detect FH signals is given, which eliminates the salt-and-pepper noise by morphological filtering, labels the connected components in the time-frequency diagram to get the location information, and removes the frequency-fixed and burst interference by means of clustering, so the FH signals can be detected when the clustering result exceeds the user-defined threshold. Simulation results show that the proposed algorithm can detect the FH signals effectively even when the Signal Interference Noise Ratio(SINR) is low.3. The blind parameters estimation algorithms for FH signals are researched. Firstly, a blind parameters estimation algorithm for low-speed FH signals using image processing method is proposed, which is based on the research of the proposed blind detection algorithm. This algorithm extracts and modifies the edge information from the FH pattern according to the clustering result, then the parameters can be estimated through the modified edge information. This algorithm can estimate the parameters effectively with the existence of frequency-fixed and burse interference, and it also works at low SNR. Then with regard to high-speed FH signals, the disadvantages of the global wave TFA methods are analyzed and the local wave TFA methods are introduced, and thus a blind parameters estimation algorithm of FH signals using local characteristic-scale decomposition(LCD) is proposed. In the algorithm the FH signals are iteratively decomposed into several intrinsic scale components(ISCs), and some ISCs are deleted which are regarded as noise components, and then an analysis sequence is derived from the maximum instantaneous amplitude of the denoised signal. So the hop rate and hop timing of the FH signals can be estimated by performing wavelet transform and Fourier transform on the analysis sequence, and therefore the frequencies can be estimated. Simulation results show that this algorithm is suitable for high-speed FH signals and can estimate the parameters accurately.