Study On Key Technologies Of Efficient Direction-of-arrival Estimation Based On Uniform Circular Array

DOA(Direction of Arrival) estimation, as one of the main branches of the phased array signal processing, has a wide application prospects both in military and civilian services. After decades of wide and profound research, DOA estimation theory has already been well developed and many excellent algorithms have emerged. Abundant well-known algorithms represented by MUSIC(Multiple Signal Classification) and ESPRIT(Estimation of Signal Parameters via Rotational Invariance Techniques), the two main subspace-class methods, have been developed in super-resolution DOA estimation area, but it is challenged to implement them in real engineering practice. One of the main reasons is that there are obvious errors between the real system model and the ideal system model. For example, the channel gain/phase inconsistency and electromagnetic mutual coupling of a real array will severely degrade the performance of the DOA estimation algorithms, and thus make them lack of robustness. Another reason is that MUSIC can be applicable to all kinds of arrays, but spatial spectral peaks searching should be involved. Generally the computation load is heavy and it is not suitable for real-time system. ESPRIT is an efficient method since it is based on a close-form algebraic solution, but the array should has a shift invariant structure which greatly restricts its applications. Some methods can be used to decrease the dimensions of the angular search or to transform an irregular array to a shift invariant structure, but unavoidably the transformation will introduce mapping error or truncation error. Sometimes the transformation itself is limited to a specific structure array.Taking the UCA(Uniform Circular Array) as the main research object, the dissertation has done a comprehensive study on array calibration and DOA estimation when channel inconsistency and mutual coupling errors are presented. A new robust joint calibration method is designed to estimate the mutual coupling and channel inconsistency error parameters for a UCA. After a detailed analysis of several blind DOA estimation methods based on RARE(Rank Reduction) technique, a blind DOA estimation algorithm via modified ESPRIT is proposed for a UCA in presence of strong mutual coupling. Also the dissertation aims to design some efficient DOA estimation algorithms which are easy to be implemented and applicable to an array with arbitrary geometry. Two new DOA algorithms are proposed which are named IESPRIT(IESPRIT,Iterative ESPRIT) and GIESPRIT(GIESPRIT,Grid IESPRIT) respectively. Quite satisfactory estimates can be obtained when the two new methods are applied to a UCA for 1D or 2D DOA estimation. What's more, the new methods are extendible and can be applied to irregular arrays, and thus extend the application scenarios of the classic ESPRIT.The main work and innovation of the dissertation can be concluded as following:(1) Simplified 2D UCRB(Unconditional Cramér-Rao Bound) expression has been strictly proved for stochastic signal models by extending the conclusion of 1D DOA estimation model.(2) PM(Propagator Method) can be used to estimate the signal or noise subspace without executing the decomposition of the covariance matrix. In the dissertation a pseudo-covariance matrix based PM for non circular signals is proposed, also a high(three) order cumulant based PM for non Gaussian signals is presented. The performance is analyzed and compared by adopting the concept of algebraic space distance when all the PMs are used to estimate the subspaces. Because of their strong ability of suppressing colored Gaussian noise, the accuracy of the estimated subspaces as well as the DOA estimates can be greatly improved.(3) A robust calibration method is proposed for a UCA to estimate the mutual coupling and the channel gain/phase inconsistency parameters jointly. In the DFT(Discrete Fourier Transform) beam space, the two types of error parameters are decoupled first, and they should satisfy an algebraic linear equation. So with a single signal source, the joint calibration can be done through two trials, and the impinging direction of the two trials needn't to be calibrated beforehand. The new algorithm provides a simple, reliable and robust UCA calibration method in practice.(4) A blind DOA estimation algorithm via a modified ESPRIT is proposed for a UCA in presence of strong mutual coupling. In the DFT beam space, the manifold vector of the UCA under the error model is transformed to a Vandermonde structure vector so that a modified ESPRIT method can be applied to get the blind DOA estimate. Specifically four methods to select the subarrays and two methods to estimate the mutual coupling parameters are presented. It is shown that the new method can differentiate closely spaced signals while conventional MUSIC or other RARE based blind estimation methods will fail.(5) A modified DOA estimation algorithm named IESPRIT(Iterative ESPRIT) is proposed for an array with arbitrary geometry. The theory of the new method is analyzed and the convergence theorem of the iteration is proved. Also two implementation schedules are introduced. IESPRIT extends the classic ESPRIT algorithm to be applicable to an irregular array. The new algorithm is based on solving an algebraic equation by iterative method, so it gains computational efficiency over MUSIC method. It adopts a closed “Angle Lock Loop” block which can lock the direction of the SOI(Signal of Interest) accurately. Therefore it provides a new solution to the MTT(Moving Target Tracking) application.(6) Another modified algorithm named GIESPRIT(Grid IESPRIT) is proposed to get 2D DOA estimate for a UCA. GIESPRIT extends IESPRIT to more complicated system model and can reduce the iteration times efficiently. Based on the IESPRIT, the whole spatial space is separated by a proper number of grids, and in each grid IESPRIT is applied to get the DOA estimate. A simple azimuth and elevation matching method is presented and the convergence of the iteration is also proved. The new method can be extended to an array with arbitrary geometry for 2D DOA estimation.Thus the dissertation has completed the studying of all the main aspects of DOA estimation especially for a UCA. It covers the subspace estimate, the calibration of the array, blind DOA estimation method under error model and two new extendable, efficient DOA estimation algorithms. Large amount of simulations are carried out and the effectiveness of the new methods is verified.