Font Size: a A A

Research On High-resolution Direction Of Arrival Estimation And Wideband Beamforming

Posted on:2020-05-14Degree:DoctorType:Dissertation
Country:ChinaCandidate:J S WangFull Text:PDF
GTID:1488306740472804Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
Direction of arrival(DOA)estimation and beam forming are two important research directions in the field of array signal processing,and have been developing rapidly in last half a century,and also have been widely used in many military and national economic fields such as radar,sonar,wireless communication,microphone arrays and biomedicine.Modern high-resolution DOA estimation methods have big performance advantages than traditional ones,due to the applications of modern signal processing theories such as subspace theory,high-order cumulants,compressive sensing,low-rank matrix theory,matrix decomposition,optimization theory,etc.There are three main categories of high-resolution DOA estimation methods:subspace-based methods,sparse methods and grid-less methods.The grid-less DOA estimation methods,which are based on continuous compressive sensing theory,are the latest research hotspots in DOA estimation field.Beam forming techniques are extended to wideband signals in several new applications,in which fixed beamwidth,frequency invariance and broadening nulls are new features of wideband beamformers.In this dissertation,several researches are carried out for high-resolution DOA estimation and wideband beam forming,and the main innovative achievements include:1.To improve the performance of current subspace-based DOA estimation methods with high-order cumulants in the scenario of limited number of snapshots,the denoising method for high-order cumulants matrix using 2q-order correlation subspaces is proposed,and the 2q-MUSIC and the 2q-SS-MUSIC are improved by the denoising procedure.The results of numerical experiments show that,using sparse linear array and planar array,in the overdetermined or underdetermined cases,with small number of snapshots,low signal-to-noise ratio(SNR)and adjacent sources,the improved 2q-MUSIC and the improved 2q-SS-MUSIC have lower root-mean-square errors(RMSE),which prove the improve the effectiveness and good performance.2.To apply sparse signal reconstruction methods to high-order cumulants-based DOA estimation,the method based on fourth order cumulants vector and off-grid sparse Bayesian inference is proposed.It does not need the prior information of the number of sources,suits for any geometry of sensors array,can achieveO(M~4)sources at most(M is the number of sensors),and has better performance in the scenario of adjacent sources,while the disadvantage is the high computational complexity.The results of numerical experiments show that,in the cases of different grid-interval,adjacent sources,SNR,underdetermined situation,planar array and non-circular sources,the proposed method achieves lower RMSE in most situations,which prove the effectiveness and the good performance.3.To apply the grid-less DOA estimation methods to the scenarios of nonlinear arrays,the one-dimensional grid-less DOA estimation method for two-dimensional arrays and two two-dimensional grid-less DOA estimation methods(the least square method and the weighted atomic norm method)for uniform rectangular arrays(URAs)or sparse rectangular arrays(SRAs)are proposed.The results of numerical experiments show that,using nonuniform circular array,the RMSE of the first method approximately approach the Cramer-Rao Bound(CRB)in most cases of number of snapshots and SNR,and the first method can detect the adjacent sources with angle interval 5°with snapshot-number 50 and SNR 0 d B,which is better than other compared methods;using URA,the least square method and the weighted atomic norm method approximately approach the CRB in most cases of number of snapshots and SNR;Using SRA,the weighted atomic norm method approximately approach the CRB in most cases of number of snapshots and SNR,and can distinguish the adjacent sources with angle interval 2°with snapshot-number 200 and SNR 6 d B,which is better than other compared methods.The above results prove the effectiveness and the good performance of the proposed methods.4.In the field of wideband array signal processing,a wideband grid-less DOA estimation method for two-dimensional arrays is proposed,and the methods of designing frequency invariance and nulls for eigenfilter-based wideband beam forming are proposed.The results of numerical experiments show that,using nonuniform circular array,first method has lower RMSE in the cases of large frame-number and high SNR,and can distinguish adjacent sources with smaller angle-interval;Using uniform linear array,the proposed wideband beam forming method can achieve very low minor lobe level,constant beamwidth in the main lobe and nulls with-80 d B in angle range[28°,32°].The above results prove the effectiveness and the good performance of the proposed methods.
Keywords/Search Tags:array signal processing, direction of arrival estimation, beam forming, subspace, high-order cumulants, compressive sensing, sparse signal reconstruction, grid-less, sparse arrays, low-rank matrix, frequency invariance, nulls
PDF Full Text Request
Related items