Pulsar Signal Delay Estimation Method With Anti-noise Ability

Due to the characteristics like extremely stable pulsars' period,X-ray pulsar navigation using pulsars as navigation stars has become an emerging navigation and positioning method.The time offset of the pulse from pulsars arriving at the spacecraft is the most important observation of the navigation system.It refers to the phase delay of the integrated pulse profile with a shorter accumulation time and the standard pulse profile.The reckoning accuracy of this delay determines the quality and positioning precision of the pulsar navigation system.Noise caused by sudden changes of the pulsar rotation period,relativistic influence,planetary activity or other disturbance factors may lead to a decrease in navigation positioning accuracy.Therefore,how to suppress the influence of noise effectively and improve the accuracy of the delay estimation algorithm has always been a hot topic in the research of pulsar navigation technology.Taylor FFT algorithm has been the most commonly used time delay estimation method for pulsar navigation.Its estimation accuracy is independent of the time resolution of pulse signal,and is more affected by signal-to-noise ratio of the profile.Bispectrum can effectively suppress Gaussian noise,so it can achieve higher measurement accuracy than the Taylor FFT algorithm under low SNR conditions.However,the performance of both two algorithms are obtained through a large number of iterative calculations,so the co mputational complexity is high.This paper studied the characteristics of frequency domain and the bispectral domain of pulse profile and analyzed the influence of noise on the estimation accuracy of the two algorithms.Two improved algorithms for suppressing noise interference are proposed in combination with wavelet decomposition:(1)Since the Gaussian white noise has a value of zero in the bispectral domain,the influence of Gaussian no ise can be eliminated by using bispectrum algorithm.By analyzing the mapping of the pulse profile in the bispectral domain,it is known that the high-frequency component of the signal has a very small bispectrum value and is greatly interfered by noise,while the low-frequency component has a high signal-tonoise ratio and a large bispectrum value.Therefore,the high-frequency component can be ignored when using the bispectrum for delay estimation,and only the low-frequency component is used for the experiment.This method first uses wavelet to decompose the pulse profile to obtain the low-frequency component of the signal.This part of the signal contains less noise,which not only preserves the bispectral information of the pulse contour,but also reduces the bispectral point participating in the iterative calculation.Using the low frequency bispectrum to estimatie the time delay can avoid the degradation of the estimation accuracy due to the influence of the high frequency band noise.(2)A fast Taylor FFT algorithm based on low frequency band is proposed.According to the frequency domain characteristics of the pulsar pulse profile,the spectrum shows a convergence trend,and the spectrum amplitude of the low frequency band is much bigger than the high frequency band.In order to reduce the influence of high-frequency components with low signal-to-noise ratio on estimation accuracy,the method first uses wavelet to extract low-frequency components with big amplitude o f spectrum,and rejects high-frequency components that are greatly affected by noise,and uses the fit of the discrete Fourier transform of the two profiles to estimate the time delay.When the signal-to-noise ratio is low,the method greatly reduces the influence of noise on the estimation accuracy and improves the accuracy of the algorithm.And use the low-frequency component can reduce the computational complexity and speed up the algorithm.The experimental data of the paper are the numerical simulation data from the European Pulsar Network(EPN)and the measured data detected by the Rossi X-ray Timing Explorer(RXTE)satellite.The simulations of the two methods are carried out respectively and compare the performance with its classic algorithm.The results prove that both two methods presented in this paper embody the ability of eli minating the influence of noise,and they can greatly reduce the computational complexity of the algorithm and improve the estimation accuracy of the algorithm slightly in the meantime.