Study Of Methords Of Frequency-Hopping Signal Estimation Based On Time-Frequency Analysis

The frequency hopping (FH) signal is one of the Spread Spectrum Methods and widely used in modern military communications. For the strong anti-jamming, low probability of intercept and excellent networking capabilities, the FH signal is also a great challenge to communication countermeasure field. Due to the unknown FH parameters, it is difficult to detect the FH signal or obtain the information from the FH signal in communication countermeasure. Therefore, Interception, detection and blind parameter estimation of the FH signal are urgent problems in modern communication countermeasure field. This paper mainly deals with the blind parameter estimation of the FH signal under the framework of time-frequency analysis. It includes the following several aspects.The traditional parameter estimation methods based on time-frequency analysis are analyzed. By comparing the performance, it is found that the computational speed of fast Hartley based STFT is greatly improved. However, a known priori condition is required to determine the appropriate window length in parameters estimation because of its time frequency resolution is poor. Morlet wavelet transform is a muti-resolution transform, but it can not estimate parameters accurately at low SNR when the morlet transform interferes seriously by noises. The performance of SPWVD is excellent, but SPWVD can not satisfy the needs of real-time parameters estimation because of its great amount of calculation.Regarding the defects of the mentioned time-frequency analysis methods, a blind parameter estimation method for the FH signal based on S-transform (ST) is proposed by introducing ST. Considering the distortion of the ridge line at high frequency component, and combining with the statistical property of white Gaussian noise in S spectrum, an indirect band-pass filtering method in time-frequency plane is proposed to smooth the ridge line and decline the influence of noise. The simulation results show that the performance of the proposed method is improved greatly at high SNR. In view of the problem of ST that the width of the window shortens quickly with the increase of frequency, an improved window function method is proposed. In this method, parameter ? ( f) is added into the window function to control the variation tendency. In order to improve the control effect of parameter, the domain, range and monotonicity of ? ( f) are defined, and the curve set that meets the definition is given. Simulation results show that the great improvement on precision of parameter estimation of the proposed method over all the above. Meanwhile, the improvement only concerns with the variation tendency of the window width, and the additional computation cost is low which means the computational complexity is comparative with ST.