Research On Reconnaissance And Efficient Implementation Methods Of Polytime Code And Nonlinear Frequency Modulated Radar Signals

With the fast development of radar systems,the radar signals with low probability of intercept(LPI)have been widely researched and employed.Compared with the common radar signals,these radar signals have better stealth abilities.Therefore,a new challenge for radar signal reconnaissance is brought.In recent years,polytime code and nonlinear frequency modulated signals,as two types of LPI radar signals,have received close attention from scholars at home and abroad.On this basis,this thesis studies relevant methods around the radar reconnaissance architecture,signal detection,modulation recognition,parameter estimation and efficient implementation.The main work and research of this thesis are as follows:(1)The radar signal models with polytime code and nonlinear frequency modulation are built.Thus,the modulation theory,time domain characteristics,frequency domain characteristics,time-frequency domain characteristics and autocorrelation function are studied.In addition,the low interception and reconnaissance architecture of two types of signals are analyzed.Lastly,the radar signal processing flow of this thesis is given.(2)The defects of conventional radar signal detection methods are analyzed.Thus,the LPI signal detection method is mainly researched according to the fractional autocorrelation method.Moreover,the signal detection process is designed.It can obtain better detection performance than conventional methods at low signal to noise ratio.(3)The modulation recognition and parameter estimation methods of the polytime code radar signal are researched.Through the analysis of cyclic spectrum characteristics,time domain characteristics and frequency domain characteristics,a modulation recognition method is proposed based on SVM classifier.On this basis,a carrier frequency estimation method for two-state polytime code signals is proposed,which has high engineering application values.(4)The modulation recognition and parameter estimation methods of NLFM radar signals are researched.The defects of discrete polynomial phase transformation method for the modulation recognition are analyzed,and an improved modulation recognition method is proposed,which is more suitable for NLFM signals with variable parameters.According to the cubic phase transformation,the short-time Fourier transform and sinusoidal curve fitting method,the parameter estimation methods of the third-order polynomial phase signal and sinusoidal FM signal are researched respectively.Compared with the traditional method,these methods are simple,efficient,and easy to implement.(5)The hardware implementation of the reconnaissance algorithm studied in this thesis is completed.Firstly,the hardware platform,the software platform and hardware architecture are introduced.Moreover,according to the characteristics of the processor,a hybrid structure model of multi-core collaborative task scheduling is proposed to improve the real-time.Then,the key technologies and processing flow are analyzed,such as the implementation of SVM and SCD.Finally,a three-stage engineering optimization strategy is proposed to ensure the efficient implementation of the algorithm.At last,the effectiveness of the reconnaissance method is verified by simulation.The establishment of a test platform verifies the feasibility,efficiency and engineering applicability of the reconnaissance method in this thesis.