Research And Application For Time-Frequency Analysis In Parameter Blind Estimation Of Frequency Hopping Signals

As a branch of spread spectrum communication, Frequency hopping communication has been widely used in the military field, because of its strong anti-jamming and other prominent features. It is particularly important and urgent problem that achieving a higher probability of intercept to capture the full frequency hopping signal in electronic countermeasure, which is also a difficult problem.Frequency hopping signal is typical of non-stationary signal, whose spectrum is a function of time. Since it is not enough to get the information of time domain and frequency domain respectively, therefore, the traditional Fourier transform parameter estimation has been unable to meet the needs. Time-frequency analysis as a new signal processing method, it is more and more widely appreciated in recent years. The basic idea of time-frequency analysis is to design joint function of time and frequency and to describe energy density of signal at time and frequency domain.Firstly, the principle of frequency hopping communication system and the basic theory of time-frequency analysis are introduced in this thesis. Then the type of window function and length on the performance of time-frequency analysis are studied. Regarding of two kind of different situations of high hop frequency-hopping signal and low hop frequency-hopping signal, the two different processing methods are designed separately. High hop frequency-hopping signal is dealt with by general method of time-frequency analysis. In dealing with slow hop frequency hopping signal, a low complexity algorithm of multiple analysis of non-overlapping windows MNOSTFT (Multiple-window None Overlap STFT) is presented, which greatly reduce the complexity of calculation and be conducive to real-time signal processing.Secondly, based on the principles of kernel function, a new kernel function is designed by analyzing characteristics of the frequency hopping signal in ambiguity domain. Entropy measure method is introduced to evaluate the performance of time-frequency analysis, which is used to choose the parameters of kernel function adaptively. Compare with the result of simulation, the new designed kernel function is to meet the requirements and the performance of parameter estimation has been improved.Finally, the conventional reassignment method used to estimate the parameters of frequency hopping signal is researched. Because of its peak value which obtained along the timeline appears great leap, it is not conducive to estimate the hopping time. So a peak smoothing method is given, and based on this method, shift Discrete Fourier algorithm SDFT (Shifted Discrete Fourier Transforms) is introduced to estimate the frequency hopping signal, which is good for improving estimation accuracy of the frequency, and it is to overcome frequency estimation error of frequency hopping signal caused by estimation error of frequency hopping signal period and hopping time in some ways.