| In the field of non-cooperative communications,detection,modulation identification,and feature recognition of communication signals form an important basis for radio communication reconnaissance and countermeasures.When studying the detection and analysis of complex high-order modulated signals,how to design real-time and efficient detection equipment to cope with the fast and changing electromagnetic environment,less a priori information and other unfavorable factors becomes a challenging topic at present.In this thesis,the problem of arbitrary fractional sampling rate transformation in noncooperative communication systems and the modulation type identification method based on traditional pattern feature parameter extraction are discussed and studied in depth for real-time and high-speed hardware structures.The main work and contributions of the full thesis include:For the sampling rate conversion of signals,the thesis investigates traditional signal interpolation and extraction structures,fractional multiplier conversion theory and general Farrow filters and their transposition structures.For the problem of flexible selection of signal output sampling rate in data pre-processing of non-cooperative communication signals,a simple structure,real-time configurable and arbitrary multiplicative rate conversion implementation structure based on Farrow interpolation filter structure is proposed.By replacing the lookup table method with the cumulative calculation of the fractional delay parameter and using the fractional delay as a logical control to decide the location of the interpolation points of the input and output sequences,the problem that the Farrow filter structure is not flexible enough and does not support online rate configuration in the current high-speed real-time signal processing hardware is effectively solved.For blind recognition of non-cooperative signals,the thesis makes a comparative analysis of the mainstream modulation recognition methods and studies the problem of feature parameter extraction and selection in blind recognition of communication signals.The thesis proposes a modulation recognition method based on higher-order accumulation and higher-order spectrum features according to the actual signal conditions and hardware environment.Firstly,a carrier frequency estimation algorithm based on the extraction of filtered spectral lines is proposed for the signal before carrier synchronization in blind identification,which amplifies the spectral resolution step by step through multi-stage extraction filters to achieve relatively accurate carrier frequency estimation and residual frequency bias correction.On this basis,reliable feature parameters are extracted,and then a blind identification decision classifier is designed for MQAM,MPSK,and MFSK signals.The implementation results show that the recognition success rate of the modulation recognition method based on the higher-order accumulation and higher-order spectrum features reaches 95% when the signal-to-noise ratio is greater than 8 d B.A communication signal modulation identification and variable rate processing system is designed and constructed,and a high-speed digital signal processing platform with FPGA as the core is built to integrate the signal pre-processing and modulation identification methods proposed in the thesis.The system is verified by acquiring more than ten kinds of modulated signals,and the system can efficiently implement signal rate conversion and realize the rapid identification of signal modulation mode. |
