Research On The Methods Of Parameter Blind Estimation For Frequency Hopping Communication Signals

As a kind of spread spectrum communication technology, frequency hopping communication has been extensively used in the military and civilian fields, because of its strong anti-jamming and anti-intercept properties and networking ability. In frequency hopping communication system, code sequence is used to control hopping of the carrier frequency, so frequency hopping signal is a typical of multi-component non-stationary signal. In the time domain, frequency hopping signal is a multi-frequency shift keying signal. In the frequency domain, it is a random hopping signal in a wide frequency band on equal or unequal intervals. In analysis of this non-stationary signal, the Fourier transform is unable to obtain the joint information of time domain and frequency domain. Time-frequency analysis is a kind of signal processing method which can describe energy density at different times and frequencies. So it gets extensive attention in recent years. In the paper, time-frequency analysis technology is mainly used to study the parameter blind estimation in frequency hopping signal.The method of time-frequency analysis can be divided into linear time-frequency analysis and bilinear time-frequency analysis. In Linear time-frequency distribution, including short time Fourier transform and wavelet transform. Wigner and pseudo-Wigner distribution, smoothed pseudo Wigner distribution and Butterworth distribution belong to bilinear time-frequency distribution. This paper mainly study from the following aspects:Firstly, although FFT-based algorithm for the spectrogram reassignment is able to improve the time-frequency resolution of short time Fourier transform, the algorithm complexity is higher. And the performance of time-frequency distribution is affected by the type of window function and length. So, recursive algorithm that has been proposed by the previous is used to realize the parameter blind estimation in frequency hopping signal. And in the process of parameter blind estimation, the performance of these algorithms is analyzed and compared by simulation. The results show that the algorithm complexity of recursive algorithm is lower than FFT-based algorithm, and is not affected by the length of window function. In terms of anti-noise ability, when signal-to-noise ratio is greater than OdB, this two algorithms both can realize the parameter blind estimation.Secondly, information entropy criteria which can be used to measure the merits of time-frequency distribution is described. In order to improve time-frequency concentration, the information entropy criterion is used to optimize the parameter information of smoothed pseudo Wigner-Ville distribution, And smoothed pseudo Wigner-Ville distribution which has been optimized is applied to realize the parameter blind estimation in frequency hopping signal, can get the correct estimation of frequency hopping parameters when the signal-to-noise ratio is not less than OdB. In order to realize the estimation of the frequency hopping parameters when the signal-to-noise ratio is lower, a algorithm that wavelet transform with singularity detection performance is used to detect hopping point of time-frequency ridge of smoothed pseudo Wigner-Ville distribution is proposed. Frequency-hopping duration and frequency-hopping timing can be estimated. The results of simulation show that the estimation of frequency hopping parameters can be get when the signal-to-noise ratio is not less than-1dB.Finally, the characteristics of slow frequency hopping signal and Winger-Ville Distribution with excellent time-frequency concentration for single frequency component signals, the signal decomposition method which is used to estimate slow frequency hopping signal parameters is proposed. When signal-to-noise ratio is low, frequency refining method is used to improve accuracy of frequency estimation. The results of simulation show that the method is effective for parameter estimation in slow frequency hopping signal, and reduce the signal to noise ratio threshold and computational complexity. The frequency refining method is also applied to estimate the frequency-hopping frequency of fast frequency hopping signal, the accuracy also can be increased.