Research On Blind Detection Techniques Of Frequency Hopping Signals In High Frequency Channel

Frequency-hopping communication is increasingly widespread applications in HF communication, with the strong anti-jamming ability, low probability of interception, anti-fading capability and multi-site networking capability. So it becomes a hot research in communication confrontation for the reconnaissance of FH signals. As the third party of non-cooperation, how to detect the FH signals quickly and efficiently is the prerequisite of subsequent processing, and it’s the key technology of FH signals reconnaissance in complex electromagnetic environment. This paper focuses on the HF FH signals blind detection technologies. The main contents and innovations are given as follows:1. The problem of preprocessing in HF wideband reception has been studied. Firstly, paper analyses the complex electromagnetic surroundings in HF wideband reception and points out the prominent factor which affects the performance of detection is that the noise floor is not flat in band. Then this paper introduces two noise floor estimation algorithms based on low pass filter and morphological filtering, and improves the algorithms by using Gaussian structural elements. These algorithms make the background noise approximately white after whitening processing which laid a solid foundation for subsequent HF FH signals blind detection.2. The problem of determining the existence of FH signals on the condition of strong interference is researched. According to the probability distribution of the spectrum between different signals are different on the basis of channel detection algorithm, this paper proposes two types of FH signals existence detection algorithms based on the entropy and weighted standard deviation of the frequency component spectrum. Firstly, two statistical probability distribution characteristics of the single frequency component’s spectrum entropy and the weighted standard are defined in this paper, and the spectral entropy and the performance of weighted standard deviation of different signals types are also analyzed. Then the FH signals existence detection model of spectral entropy and weighted standard deviation based on characteristics of the spectral entropy and weighted standard deviation are given. The simulation results show that the algorithms can determine whether there are FH signals effectively without any prior information, which provide the basis for further follow-up detection.3. The problem of FH signals blind detection based on gray-scale time-frequency spectrogram is studied. Firstly, the paper proposes a FH signals blind detection algorithm based on improved symmetrical co-occurrence matrix by bringing in symmetrical co-occurrence matrix threshold segmentation algorithm. The algorithm processes the gray-scale time-frequency spectrogram according to the neighborhood distribution characters and the space location information of FH signals hops. Then the FH signals blind detection algorithm based on secondary morphological filtering is proposed. The gray-scale time-frequency spectrogram is processed with gray morphological filters from frequency component and time component respectively, according to the gray-scale morphology characters of FH signals hops in the time-frequency spectrogram. Simulation results show that the algorithms can eliminate the effects of noise and interference signals and extract the FH pattern from gray-scale time-frequency spectrogram completely. Furthermore the algorithms can avoid the defects of energy binarization, which are simple and easy for engineering implementation.4. The problem of FH signals blind detection based on direction-frequency spectrogram is studied. Aiming at the situation that multiple FH signals exist in the received bandwidth simultaneous, the paper improves the existing algorithms by introducing the signal direction information. Firstly, the definition of direction-frequency spectrogram is given and the feasibility of FH signals detection in direction-frequency spectrogram is analyzed. Then a detection algorithm based on direction analysis is proposed according to the direction distribution of signals in direction-frequency spectrogram, and binary segmentation direction-frequency spectrogram and feature extraction by HMT are processed respectively. Then according to the conclusions in the above chapters of the research findings, the direction-frequency spectrogram can be improved by combining with the time-frequency spectrogram, which can reduce the noise direction’s interference. Simulation results show that the algorithm can apart the signals effectively and eliminate the complicated steps for removing the jamming. By combining the time-frequency preprocessing, signals’ direction detection probability is improved and the false alarm probability is declined.