Parameter Estimation Of Signals In Ultra-wideband Non-uniform Sampling

With the increasing complexity of the electromagnetic environment in electronic warfare,the range of frequencies that electronic reconnaissance receivers need to cover is increasing,which requires higher sampling rate of ADC chips.Subsequent signal processing and data transmission and storage are also faced.huge pressure,so the traditional Nyquist sampling theorem has become a bottleneck for parameter estimation of wide-band electronic reconnaissance signals.In view of the above problems,this paper first proposes a wide-band electronic reconnaissance process based on compressed sensing theory,and combines ultra-wideband non-uniform sampling technology with compressed sensing theory.Aiming at the problem of basis mismatch in practical applications,the spectrum estimation algorithm under gridless sparse representation is proposed.Then,based on the large bandwidth LFM signal in wide-band electronic reconnaissance,three parameter estimation algorithms based on discrete sparse dictionary are proposed.The task of estimating the parameters at the under-Nyquist sampling rate in non-reconstructed conditions is realized.The main research contents are as follows:1.This thesis analyzes the shortcomings of the electronic reconnaissance system under the traditional Nyquist framework for parameter estimation of wide-band electronic reconnaissance signals,and proposes a wide-band electronic reconnaissance process based on compressed sensing,which can greatly reduce the sampling rate and hardware requirements,and proposes a ultra-wideband non-uniform sampling model,capable of simultaneously compressing and sampling signals.2.Aiming at the problem of basis mismatch in compressed sensing applications,a method for spectrum estimation of non-uniformly sampled signals under gridless sparse representation is proposed.Firstly,a continuous parameterized dictionary is established to perform gridless sparseness of wideband electronic reconnaissance signals.The simulation results show that the application of the atomic norm denoising online spectrum estimation under Nyquist framework and the spectrum estimation of the non-uniform sampling signal under the compressed sensing framework are simulated.The results show that under certain conditions,the spectrum estimated task of multi-frequency complex sinusoidal signal,LFM signal and BPSK signal with smaller wideband can be solved by converting the atomic norm of the non-uniformly sampled signal into a semi-definite programming problem under the gridless sparse representation of the continuous parameterized dictionary proposed in this thesis.3.For the case that the LFM signal with larger bandwidth cannot obtain the optimal sparse representation under the gridless sparse representation and DFT basis and affect the parameter estimation effect,three kinds of LFM non-uniform sampling signal parameter estimation algorithms based on discrete sparse dictionary are introduced: Waveform Matching,Down-chirp and DFRFT-based algorithms are simulated and analyzed separately.The Waveform Matching algorithm is simple in principle,but in the case where the parameter estimation accuracy requirements are the same,the number of atoms required for the sparse dictionary in the Down-chirp algorithm is less,but both require a priori information of the frequency parameter of the LFM signal;the parameter estimation algorithm based on the DFRFT dictionary can not only solve the above problems,it also has excellent anti-noise performance due to its insensitivity to noise.4.Using the built ultra-wideband non-uniform sampling hardware platform,the non-uniform sampling measured data of the radar signal generated by the signal source is obtained,and the spectrum estimation algorithm under the gridless sparse representation,the Waveform Matching algorithm,the Down-chirp algorithm,the DFRFT-based parameter estimation algorithm is further verified by simulation.