Research On DOA Estimation Algorithm For Non-ideal Conditions

By processing the received signals,direction-of-arrival(DOA)can obtain the needed important information such as angles and positions.Therefore,it has drawn increasing attention in the applications of many systems,such as radar,sonar and communication systems.Under ideal conditions,DOA estimation algorithms have been developed to a high level.However,different from the DOA estimation under those ideal assumptions,when obtaining the needed information under the actual conditions in applications,DOA estimation has to experience the non-ideal observation space.Thus,the environments of the signals,the conditions of the arrays,and the harsh requirements for the fast processing in practical applications,lead to the fact that the non-ideal conditions become the bottleneck from theoretical technologies to practical applications.As a result,the process that the theory technologies of algorithms become the practical applications,is the process that they can adapt to the complex practical environments and the realistic demands.Under the premise of adapting to the practical environments and meeting the demands of applications,to achieve DOA estimation with high accuracy,this thesis researches on the DOA estimation problems for the main forms of the non-ideal environments of signals and noises,the experienced non-ideal conditions and requirements when processing the signals,and the non-ideal environments of the arrays.The specific problems include DOA estimation for low signal-to-noise ratio(SNR),DOA estimation with colored Gaussian noises,DOA estimation under the demands for fast signal processing,DOA estimation in the presence of unknown mutual coupling,DOA estimation in the presence of gain-phase errors,and so forth.This thesis carries out the researches for the following four parts:Firstly,for the conditions of low SNR and colored Gaussian noises,DOA estimation algorithms with high accuracy are researched.The model for received data is established,the statistical properties of the second order and the fourth order are summarized.In addition,the principles of the sparse signal reconstruction for the widely used l1-norm,and the classical algorithms for low SNR and colored noises,are analyzed.Making full use of the potential advantages of the convex relaxed sparse DOA estimation technique,aiming to solve the sensitivity problem of these algorithms to colored noises,a high order sparse DOA estimation algorithm with low dimensional measurement matrix is proposed.Meanwhile,for the proposed algorithm,the feasibility of the high order sparse technology and the rationality of the parameter selection are analyzed.The effectiveness of the technology in the proposed algorithm is verified by simulations.Secondly,under the condition of the practical demand for fast signal processing,DOA estimation algorithms with high accuracy are researched.Different from the treatments of constant speed or low speed under the assumptions of ideal signal processing environment,practical applications are usually required to obtain the exact needed information in the fastest speed.Based on the reservation of the advantages of the convex relaxed sparse DOA estimation algorithms,aiming at the problem that the degree of the increased speed of the reduced complexity algorithms for the l1-norm is tiny as they do not change the principle of the signal reconstruction in essence,a reweighted smoothed l0-norm based sparse DOA estimation algorithm is proposed.The proposed algorithm includes the design of continuous function with reweighted coefficients,and the design of fast sparse signal reconstruction.Meanwhile,for the proposed algorithm,the corresponding scheme of the parameter selection and the analysis of the computational complexity are provided.The proposed algorithm is two orders of magnitude faster than the widely used l1-norm based sparse algorithms.Besides,for the DOA estimation problem with the situation of multiple measurement vectors(MMV),a joint smoothed l0-norm based fast sparse DOA estimation algorithm is proposed.Also,the extension applications and the complexity of the proposed algorithm are analyzed.The proposed algorithm has a good adaptability to the MMV conditions,and can deal with the colored Gaussian noises.Simulations verify the effectiveness and the quickness of the proposed algorithms.Compared with the l1-norm based sparse DOA estimation algorithms,the proposed algorithms can both provide higher accuracies of DOA estimation in a wide range of SNR,and can both make the computation speed one hundred times faster.Thirdly,in the presence of unknown mutual coupling,DOA estimation algorithms with high accuracy are researched.The received data model considering the effects of mutual coupling is established,and the corresponding representative DOA estimation algorithms are analyzed.To further improve the accuracy of DOA estimation with unknown mutual coupling,aiming at the problem that the refinements of the dictionary in sparse algorithms make the algorithms invalid because of the increased correlation between the grid cells,a sparse DOA estimation algorithm with effective refinement process is proposed.Besides,the received data model for noncircular signals is established.Taking advantage of the useful information in the structure of the noncircular signals,a noncircularity based reweighted sparse DOA estimation algorithm with double restraints is proposed.Simulations verify that the proposed algorithms both have superiorities.At last,in the presence of gain-phase errors,DOA estimation algorithms with high accuracy are researched.The corresponding signal model is set up,and the representative DOA estimation algorithms are analyzed.In the DOA estimation algorithms based on the iterations,the unknown gain-phase errors lead to the initial error matrix being randomly set as a unit matrix or a random matrix.Furthermore,the required computation time of the iterations based sparse DOA estimation algorithms with gain-phase errors is enormous.To solve these problems,making use of the structure of the gain-phase errors,a scheme for the error matrix estimation and a fast sparse DOA estimation algorithm are proposed.In addition,aiming at the important application of DOA estimation,i.e.the DOA estimation for multiple-input multiple-output(MIMO)radar,under the circumstance that gain-phase errors are considered in both of the transmit array and the receive array,the problem of DOA estimation with high accuracy is researched.To achieve the DOA estimation in this condition,a high order error matrix estimation algorithm for the joint transmit-receive array with gain-phase errors is proposed.Meanwhile,the proposed algorithm can deal with the colored noises.The applicable conditions and estimation performance are analyzed.The advantages of the proposed algorithms are respectively verified by simulations.