The Research On Joint Multiple Parameter Estimation In Array Signal Processing

Array signal processing plays a very important role in modern signal processing. It has a good prospect of application in many fields, such as radar, sonar, mobile communications, etc. By sampling and processing signal both in time domain and in spatial domain, the information of interest contained in the signal can be exploited fully, thereby interference can be suppressed more effectively, and so the capacity of system can be improved by introducing array signal processing algorithms. Compared with many traditional techniques, the methods based on array signal processing can improve both the accuracy and the resolution of direction finding. At the same time, those parameters, such as frequency, relative time-delay and polarization of incident signals are very important for confirming the velocity, distance and characters of targets, so the joint direction and those parameters estimation are important too. This paper aims at the development of high-resolution and robust parameters estimation methods, involving direction-of-arrival (DOA) and frequency in different array antennas, and then verifies these methods by computer simulations. The antenna array for research includes Uniform-Linear-Array (ULA), Uniform-Circular-Array (UCA) and L-shaped array. The main work can be summarized as follows:The fundamental theory of the array signal processing, including the mathematical model of array system and the common array model, is addressed. Then, the analysis of the classical subspace based algorithms, including the MUSIC algorithm and the ESPRIT algorithm, are presented.The background and knowledge of PARAFAC technique are introduced which mainly focus on analysis of 3-dimension data. Two important methods of decomposition of PARAFAC model are introduced: TALS (Trilinear Alternating Least Square) and COMFAC (COMplex parallel FACtor analysis). The problem of bind 2D-DOA estimation is addressed with L-shaped array. Based on shift invariance property, an improved method for estimating 2D-DOA is presented. The algorithm uses the array geometries to construct a matrix and then obtain the required signal subspace via the eigen decomposition of the constructed matrix. Our algorithm has much better 2D-DOA estimation performance than conventional ESPRIT algorithm, and it can identify more DOAs than conventional ESPRIT algorithm.The next section links joint angle and frequency estimation problem to the trilinear model and derives a novel blind joint angle and frequency estimation algorithm. Angle and frequency are obtained based on trilinear decomposition of a trilinear model, which is constructed based on oversampling the system output. The proposed algorithm has better performance, and supports small sample sizes. The useful behavior of the proposed algorithm is verified by simulations. The feasibility of applying the proposed algorithm to the other arrays, including uniform circular array and L-shaped array, is discussed. Then, the proposed algorithm is proved to be applicable to uniform circular array and L-shaped array.