Research On Sparse Array-based Direction Finding Structures And DOA Estimation Algorithms

As array signal processing technology develops,the requirements on direction of arrival(DOA)estimation accuracy become higher and higher.Using more antennas to improve the estimation accuracy is a direct solution,but the number of front-end circuit chains is increased.As a result,the system complexity and hardware cost dramatically increase.To address this issue,an effective way is sparsely placing the antennas,which can reduce the number of antennas without sacrificing the array aperture.Meanwhile,the number of degrees of freedom can be guaranteed by exploiting the auto-correlation information of the received signal.This kind of arrays is called the sparse array.However,there still exist some problems in the sparse array now,e.g.,holes in the coarray of coprime arrays,inflexibility of antenna positions,and weakness in processing high instantaneous bandwidth signals.Therefore,this thesis aims at solving these problems,and can be divided into four parts according to the research contents.First,there exist holes in some sparse array configurations,which will reduce the degree of freedom.The array interpolation algorithms based on matrix completion can fill the holes,thus improving the degree of freedom.However,the noise in covariance matrix is not suppressed,where the estimation accuracy is affected.Aiming at solving this problem,this thesis firstly improves the conventional matrix completion-based virtual array interpolation algorithm and obtain the hybrid denoising approach,where the noise is suppressed and the estimation accuracy is improved.Then,direct denoising approach is proposed based on hybrid denoising approach,where the relationship between the covariance matrix with respect to the physical array and virtual array is exploited.The two steps in hybrid denoising approach are combined into one step.As a result,the computational complexity is reduced.In addition,the vectorization and redundancy reduction operation are no longer required.Numerical simulations are performed to validate the corresponding performance.Then,according to the applications of the compressive sensing(CS)technique in DOA estimation system,data compression-based DOA estimation structure using coprime array can reduce the system complexity and hardware cost while the estimation accuracy is guaranteed,but this structure has no generality and relevant performance analysis.To address this issue,a generalized compressed sparse array(CSA)structure is developed in this thesis,and an improved algorithm is also proposed.Besides,the Cramer-Rao bound(CRB)for the CSA with respect to the DOA and corresponding existence conditions are derived,based on which the degrees of freedom of CSA is further derived and validated in detail.A meaningful conclusion is that for the CSA with the L-element sparse array being the receive array and having M<L front-end circuit chains,it can achieve a higher number of degrees of freedom than the conventional M-element sparse array.In addition,the M-channel CSA has a higher array aperture,thus having a higher estimation accuracy than the conventional M-element sparse array.The superiorities of the proposed CSA are examined by numerical simulations.Furthermore,in order to obtain a high number of degrees of freedom,one-bit receiver needs more antennas and front-end circuit chains,thus increasing the system cost.To overcome this problem,the one-bit quantization technique is combined with the data compression techinique in the third part of this thesis.A novel DOA estimation structure,referred as compressive one-bit DOA estimation structure,is proposed.In addition,two estimation algorithms for the proposed scheme,i.e.,the iterative compressive measurement based multiple signal classification(CM-MUSIC)algorithm and the iterative CS-based algorithm,are also proposed in this part The iterative CM-MUSIC algorithm is suitable for the situation where the number of sources is low and the requirement on estimation accuracy is high.On the other hand,the iterative CS-based algorithm can be used in the situation where the number of sources is high and the requirement on estimation accuracy is low.Numerical simulations are conducted to assess the performance of the proposed scheme.Finally,since we can transmit signals with coprime carrier frequencies to generalize an equivalent coprime array,the coprime-frequencies based uniform linear array(CF-ULA)structure can further reduce the number of essential antennas.However,for the CF-ULA structure,a group of additional phase is induced when the reflected signals with different carrier frequencies are received.The performance affected by the additional phase has not been analyzed.In addition,as the number of antennas goes up,the system complexity of the CF-ULA structure is then increases.To solve these problems,the performance when CF-ULA is used for DOA estimation is analyzed by deriving the corresponding spatial correlation coefficient and CRB expression.An important result is that the additional phase can improve the angular resolution and the degree of freedom.Then,a data compression-based CF-ULA structure and the corresponding DOA estimation algorithms are then proposed in this thesis.The proposed scheme combined the data compression technique and the CF-ULA structure,where the system complexity is effectively controlled and the estimation accuracy is not degraded.The effectiveness of the proposed scheme is examined by numerical simulation.