Estimation Of Power Spectral Density Of Radio Astronomy Signal Based On Sub-nyquist Sampling

The high-resolution spectrum or power spectrum density observation of radio signals helps to understand the fine structure of radio sources,and provides a reliable basis for the forecast and warning of celestial activities and their intensity.The autocorrelation spectrum analyzer is an important equipment in radio astronomy observation.It uses oversampling to improve the anti-noise performance of the system.With the limited sampling frequency of Analog to Digital Converter,the observation bandwidth is limited by the frequency resolution;Compressed sensing technology can effectively reduce the sampling rate of the signal,but requires sparse signal.In order to remove the limitation of sparsity and increase the radio observation frequency while guaranteeing the resolution,the sub-Nyquist sampling and power spectral density estimation methods of the signal are intensively studied.The main research contents are as follows:Firstly,aiming at the problem that the sampling rate of traditional power spectral density estimation is too high,three sub-Nyquist sampling frameworks of compressed sensing theory are introduced: Multi-coset Sampling(MC),Modulated Wideband Converter(MWC),Random Modulation Pre-integrator(RMPI).And the precise time delay between the MC channels make it hard to control,while the MWC and RMPI have been implemented with hardware by scholars.Secondly,for the limitation of compressed sensing on sparsity of signals,the power spectral density estimation based on MC is studied.The analog frontend adopts MC system for compressive sampling,and the backend uses cross-correlation between the outputs of each sampling channel to estimate the power spectral density of the signal in segments.The simulation results show that the compression ratio of compressive estimation is more than the noncompressive estimation,applicable for power spectral density estimation of sparse or nonsparse signals.Third,for the precise time-lag of MC is difficult to achieve with hardware,the power spectral density estimation based on RMPI is studied.The analog frontend uses RMPI to compressive sampling,perform Fourier Transform on each sampling sequence,and then establish cross-power spectrum density equations to realize the power spectral density estimation.Simulation results show that this method can effectively reduce the sampling rate,applicable for power spectral density estimation of sparse or nonsparse signals.Fourth,using the solar radio signals in the frequency range of 55-65 MHz from Yunnan Observatory to verify the two power spectral density estimation algorithms.The experimental results show that both methods can accurately estimate the signal frequency range while reducing the sampling rate.When the same power spectral density resolution and number of sampling channels are selected,the noncompressive power spectral density estimation based on MC is superior to the power spectral density estimation based on the RMPI in estimation speed and performance,but the RMPI hardware circuit is easily to realized make it more practical.