Joint Estimation Techniques Of Spectrum And DoA Based On Sub-Nyquist Sampling

Spectrum sensing and spectral analysis are crucial in application fields such as electronic reconnaissance,spectral management and control,and cognitive communication.The receiver aimed at spectrum sensing should have a large instantaneous bandwidth so that high-speed sampling with the Nyquist rate will bring great pressure to the hardware implementation.At present,spectrum sensing with sub-Nyquist sampling techniques can effectively alleviate the problems caused by high sampling rate,and monitor a large spectrum range in real time.In addition,spectrum sensing combined with array techniques can not only obtain the spectral distribution,but also estimate the direction of arrival(DoA)of signal source at the same time.Therefore,spectrum sensing based on both array and sub-Nyquist sampling techniques can acquire their advantages.However,the current researches still have problems such as insufficient sensing performance,complex receiver structure,and excessive overhead of hardware resource.To overcome these problems,based on array and sub-Nyquist sampling techniques,this dissertation systematically studies blind spectrum sensing and DoA estimation Techniques with sub-Nyquist sampling,including various receiver structures and corresponding algorithms,where ”blind” means sensing terminals do not know the distribution of the spectra of signal sources(the support information of sources).From the perspective of signal modeling,the research contents of this dissertation can be divided into two categories: one is to transform the carrier frequency by multi-coset sampling structure,and combine array techniques to establish a signal model containing both carrier frequency and DoA information;the other is to transform the support information by modulated wideband converter(MWC)structure,and build a signal model including support information,carrier frequency,and DoA.The main contributions of this dissertation are listed below:1.Based on the multi-coset sampling structure and linear array,for the scene where the spectra of signal sources are arbitrarily distributed,a signal model is built through the characteristics of both multi-coset sampling structure and linear array having similar signal models.A joint estimation algorithm of carrier frequency and DoA based on subspace decomposition and two-dimensional covariance matrix expansion method are proposed.The algorithm jointly estimates the carrier frequency and DoA by constructing a two-dimensional pseudo-spectrum,and then reconstructs the spectra of signal sources.This joint estimation algorithm has better performance and no parameter pairing problem.In the case of sparse delay pattern and sparse array,the proposed expansion method can greatly enlarge the size of the covariance matrix,thereby improving the perceived performance of the algorithm in extreme scenarios with multiple signal sources.2.Based on MWC structure and linear array,for the scene where the spectrum of the signal source exists independently in one sub-band and there is at most one signal source in the same sub-band,an array receiver structure based on the same single-channel MWC structure is designed.A signal model in the form of a third-order tensor is established by specifically aliasing spectra of signal sources.A blind spectrum sensing and DoA estimation algorithm based on tensor decomposition and a folded frequency estimation method are proposed.The designed receiver takes advantage of the MWC structure that can trade sampling rate for channels,which effectively reduces the complexity of the receiver structure and the overhead of hardware resources.The proposed algorithm transforms the high-dimensional estimation problem into multiple paired one-dimensional estimation problems.The folded frequency of the signal source can be estimated from its reconstructed sub-band spectrum by using the folded frequency estimation method,and the carrier frequency and DoA are respectively calculated based on the estimated support information.Theoretical analyses and simulation results show that: increasing the number of equivalent channels of a single-channel MWC structure can improve the robustness of the algorithm;increasing the number of antennas can effectively improve the performance of DoA estimation.3.Based on MWC structure and linear array,for the scene where the spectra of signal sources are arbitrarily distributed,the receiver structure is improved by connecting each antenna to a different single-channel MWC structure.Two blind spectrum sensing and DoA estimation algorithms respectively based on compressed sensing(CS)and subspace decomposition are proposed.How to judge cross-band signal sources reasonably is analyzed by estimated parameters,and the folded frequency estimation method is improved for cross-band signal sources.Theoretical analyses and simulation results show that: the number of antennas determines the maximum number of identifiable signal sources in one sub-band and DoA estimation performance;the total number of equivalent channels determines the total maximum number of identifiable signal sources and robustness of the algorithms;the proposed algorithms can identify more signal sources and have better robustness under the same hardware resources.4.Based on MWC structure and L-shaped array,for the scene where the spectra of signal sources are arbitrarily distributed with two-dimensional DoA,a L-shaped array receiver based on different single-channel MWC structures is designed,and a blind spectrum sensing and two-dimensional DoA estimation algorithm based on subspace decomposition is proposed.The algorithm constructs a two-dimensional pseudo-spectrum for each subband,so that one signal source will only create an obvious spectral peak on the pseudospectrum of the corresponding sub-band and will not affect the other pseudo-spectra,so it can perceive multiple signal sources at the same time.To enhance the recognition ability under complex scenarios,a dynamic identification method for multiple signal sources is also proposed.Theoretical analysis shows that a reasonable setting of the total number of equivalent channels can ensure that the two-dimensional pseudo-spectra do not interfere with each other,thus ensuring the performance.Simulation results verify that the proposed algorithm has great multi-signal sensing ability and parameter estimation performance.