Research And FPGA Implementation Of Key Algorithm In The Real-time Spectrum Analyzer

With the development of technology,the modern radar system and the communication network use pulse,frequency hopping,CDMA,adaptive modulation and other digital radio frequency technology.Given the behavior characteristics of the current digital radio frequency signal,time is becoming an indispensable information in modern electronic and communication system test.Therefore,it is necessary to search for the new method of spectrum analysis to trigger capture and real-time spectrum analysis the various kinds of abnormal signal,transient signal and weak signal in strong noise background.The real time spectrum analysis system is used to detect the weak signal in strong noise background and to analyze the real time spectrum of non-stationary signal analysis.In this paper,the main contents are as follows:(1).Through the analysis of the basic principle and needs of function of real-time spectrum analysis system,the real-time spectrum analysis system includes signal acquisition module,real-time trigger module and baseband digital signal processing module is designed,and the realization of the basic principle of each system module is researched.(2).The signal sampling theory,Analog to Digital Converter(ADC)performance and the time interleaved parallel sampling technique is introduced in this chapter of signal acquisition module.In the time interleaved parallel sampling technique,the sine fitting algorithm is used to retrieve the channel mismatch errors.Besides,according to the characteristics of the three error,the adder,multiplier and FARROW structure all-pass filter are combined to analyze signal correction method in detail.(3).According to the demand of weak signal spectrum analysis of abnormal signal and transient signal and strong noise background,three triggered ways of the trigger which are Power trigger,level trigger,frequency mask trigger,and the cycle timing DDR2 storage mechanism are designed in this paper.They are used to achieve effective real-time trigger and capture of the signal.In order to avoid the loss of signal,the overlapping frame short stepping sliding can solve the window effect in spectral analysis effectively.(4).Based on the problem of spectrum analysis and time frequency resolution in the traditional real time spectrum analysis and the contradiction between large computation and real time in high resolution spectrum analysis of abnormal signal and transient signal,a new spectrum detection scheme has been implemented in this paper.In this scheme,the sampling window,whose window length approaches the stationary signal,is used to sample the transient signal.Then the sampling signal for multiple cache,and the cache according to the read delay polyphase filter sampling signal and the sum of each cache signal,filtering results obtained large signal sets and phase signal reconstruction with spectral information;The spectrum zoom and time resolution-independent is obtained by windowed Chirp-Z Transform(CZT);At last,the influence of side lobe error and noise error is solved by using cumulative average and threshold analysis.(5).Combined with the engineering practice,the FPGA implementation of real-time spectrum analysis system of this thesis is designed.Aimed at the optimization challenges of timing and power consumption faced by FPGA implementation,in this paper,an optimization scheme,combining Register Transfer Level(RTL)logic optimization,chip pin electrical standard setting,synthesis and timing constraints and EDA synthesis with the layout and routing software settings,has been put forward.Furthermore,the implementation of the optimization scheme can be used to optimize the timing and power optimization design system.Through the optimization above,the minimum time margin of the system is increased from-3.3ns to 0.759 ns and the system's power consumption is reduced by 45%.Through the timing simulation of the system,the weak signal detection,the transient signal analysis and real-time mode analysis performance are verified.The simulation results show that for the weak signal detection under strong noise background,under the condition of-30 dB SNR,the maximum frequency measurement error and the relative amplitude error of the system are 7KHz and 2.5%;For transient signal whose duration is 1.25?s,the maximum time measurement error and frequency error is 0.575?s and 38 KHz respectively;In real-time analysis mode,the system measurement error of maximum frequency and relative amplitude measurement error is 137 KHz and 0.72% respectively.Except for the initial calculation of delay about 16.384?s,in this paper,the processing performance of continuous data is real-time.