Reseach On Pattern Synthesis And Direction Of Arrival Estimation For Sparse Array

Array signal processing is one of very important research subarea in signal processing. It affects people’s lives in everywhere and has very extensive application, such as Radar, sonar, wireless communication, seismology, medical detection, electronic warfare and malfunction diagnosis. Antenna array technology provides an effective idea to detect and deal with the received signal from any angles. Compare with single antenna which is limited by directivity and bandwidth, the beampattern is obtained by adjusting the amplitude and phase distribution in the antenna array. Recently, the conditions that how to extract the target information from the low SNR and restrict the pattern with narrow scanning beam without increasing the gains are required highly in more and more fields. However, many methods and applications are limited to use the uniform array, or there are lots of difficulties and challenges in matching some specific pattern. In this paper, some new algorithms based on sparse array are developed and proposed to solve these problems. These methods can reduce the complexity and cost of the antenna radiation greatly, also overcome and remedy the insufficient of the traditional methods effectively in DOA estimation and pattern synthesis. What’s more, a lot of theoretical analysis and algorithm simulation are carried out. The main work and innovations of this dissertation are summarized as follows:Firstly, this article starts from the basic theory of array signal processing. Some basic model of antenna array, which are uniform linear array, planar array and circular array, are described. The traditional pattern synthesis methods are introduced. Then, DOA estimation model is further studied within the basic theory of compressed sensing. The convex optimization is studied in the application of pattern synthesis. These two theories are paved the way for subsequent chapters.Afterwards, the spatial spectrum estimation and global searching methods are summarized in the application of the DOA estimation. In order to reduce the computational complexity and improve the convergence of global optimization, a method based on genetic algorithm and fuzzy discrete particle swarm optimization is proposed for estimating the DOA and signal power simultaneously. The parameters of the DOA and signal power are represented as a cost function, the solution is solved by utilizing the proposed optimization method. There are lots of problem needed to be solved in DOA estimation, such as ambiguity and mutual coupling. The ambiguity mechanism of the uniform linear array and nonuniform linear array are discussed deeply, a DOA estimation method with resolving manifold ambiguity through adding a semi-circular substrate is studied corporately based on the basic of planar substrate. With a lot of simulations, this method can eliminate the trivial and non-trivial ambiguity completely.This article also discussed the method of the signal feature extraction in application of the harmonic time reversal signal and sparse antenna pattern synthesis comprehensively. The matrix pencil method and convex optimization in pattern synthesis are introduced and the drawbacks are discussed. The optimization problem of the array element location and excitation is converted into feature extraction problem in harmonic signal. Moreover, the Filter Diagonalization method(FDM) is introduced as spectrum estimation in quantum mechanics. The method is proposed to solve the reconstruction problem of the nonuniform linear array pattern through FDM. Then, this proposed method is applied in planar array. It only needs little priori knowledge to reconstruct the desired pattern. In actual radar antenna applications, the accurate steering array is got hardly due to the array error. A convex optimization method based on reweighted l1-norm is proposed to do the pattern synthesis. This method can reconstruct the beampattern within maximum sparse antennas.Subsequently, aiming at reducing the peak sidelobe level, we further continue to study the pattern synthesis method. As we know, the array position optimization is non-linear problem. It is hard to be obtained in statistic searching optimization. According to the feature that the array factor and array excitation are a pair of Fourier transfer, a hybrid optimization method is proposed to synthesis pattern fast in thinned pattern. Moreover, this method is extended to two-dimensional planar array. This method not only reduces the peak sidelobe level and control the null, but also minimizes the array number. Then, this method is applied in sum and different pattern. We propose antenna correction with pattern synthesis in special pattern, which has very extensive application in deep space detection and satellite communication.Finally, the conformal array is known as conforming with carriers, so the shadow effect, mutual coupling and polarization lead to pattern synthesis low performance. A convex optimization using model-based parameter estimation is proposed to sparse the array and synthesis the pattern in conformal array. The isolated pattern in arbitrary location is obtained by using the model-based parameter estimation method. Then, the reweighted l1-norm optimization method is utilized to design the beampattern. This method not only can sparse the array layout, but also synthesis the beampattern.