Research On Joint Estimation Algorithms Of Array Signal Parameters And Its Applications

Array signal parameters estimation techniques have played a fundamental role in manyapplications involving radar, sonar and communications. Howerer, with the development of arraysignal estimation algorithms in practical applications, the investigation on robust and high accuratearray signal parameter estimation algorithm have received great interest. Joint multiple parameterestimation of multiple sources for the conventional sensor array and the vector array, and thecorresponding estimation of multiple targets in bistatic MIMO-radar are investigated in thisdissertation. This research of the dissertation maily consists of the following parts:The two dimensional direction of arrival and frequency estimation algorithm is investigated. Inthis paper the theory of parafac factor is applied for the joint estimation of2D-DOA and frequency.First the joint2D-DOA and frequency estimation using the parafac factor algorithm for L-shapedarray is proposed. The output signal of the array antennas is analyzed and it has trilinear modelcharacteristics. The frequency and2D-DOA can be estimated from trilinear model decomposition.Then the joint2D-DOA and frequency estimation using parallel factor quadrilinear decomposition foruniform square array and double linear array is studied. The output signal of the array antennas isanalyzed and it has quadrilinear model characteristics. The frequency and2D-DOA can be estimatedfrom the matrices via low-rank decomposition which utilizing the uniqueness of the parallel factordecomposition. The method, which does not need eigvalue decomposition or singular valuedecomposition is an iterative algorithm and it can convergence over certain iterative times. Theadvantages of the method require no searching spectral peak or pairing parameters. In constrast withthe conventional methods, the algorithm has higher precision estimation of parameters and works wellunder small sizes.The joint angle and frequency estimation method for arbitrary acoustic vector sensor array isinvestigated. The main advantage of the vector-sensors over traditional scalar sensors is that theymake use of more available acoustic information and they can estimate multi-parameter of the soundwave. We drive the model of the received data for arbitrary acoustic vector array. Two algorithms ofmulti-parameters estimation have been proposed. DOA-matrix method has the advantage of lowcomputational complexity and can be applied in source orientation occasion which needs strictlycomplexity limit. The parallel factor quadrilinear decomposition algorithm does not require searchingspectral peak and can pair parameters automatically. Compared with the algorithm of ESPRIT and the parallel factor trilinear decomposition, The parallel factor quadrilinear decomposition algorithm hashigh precision in parameter estimation and supplys a new choice for source orientation.The algorithm of the direction of departure(DOD) and direction of arrival(DOA) estimation ofmulti-target for bistatic MIMO radar is studied. We propose a reduced-dimension multiple signalclassification (MUSIC) algorithm. The algorithm reduces the dimension of2D-MUSIC and the DOAcan be get from searching spectral peak of the1D-MUSIC. The DOD steering vector can be estimatedfrom the relationship of DOA and then DOD can estimated via the least square method. The proposedalgorithm can avoid the high computational cost within two-dimension MUSIC (2D-MUSIC)algorithm and has very close performance to2D-MUSIC algorithm. We illustrate that thereduced-dimension MUSIC algorithm has better performance than ESPRIT algorithm and pair theparameters automatically and can work well in the other irregular array manifolds.The joint estimation of2D-DOD and2D-DOA for bistatic MIMO radar with L-shaped array anduniform circular array is studied. Most of works develop models for uniform linear array which onlyidentifies one-dimensional angle. The L-shaped array can estimate two dimensional angle and is muchcloser to actual situation. Uniform circular array is a commonly array and can also estimate twodimensional angle. But the signal processing is more complicated because the direction vectors do nothave the Vandermonde characteristics. The algorithm of2D-DOA and2D-DOD for L-shaped anduniform circular array using the parallel factor analysis is presented. The proposed algorithm canestimate four-dimensional angle which uses the structure characteristics of L-shaped and circulararray and least square method. The advantages of the algorithm are that it has higher parameterestimation precision than ESPRIT algorithm and is close to Cramro-bound(CRB). The estimatedparameters pair automatically and it can works well under small sizes.The four dimensional angle and Doppler frequency estimation for bistatic MIMO radar withL-shaped array is studied. The algorithm of multi-parameter joint estimation based on DOA-matrixmethod is proposed. The algorithm construsts the matrixs according to the DOA-matrix and derivesthe formula of joint four angle and Doppler frequency which use the relationship between theeigenvalue and corresponding eigenvector. The close-form solution can be obtained. Compared withESPRIT algorithm the proposed method only needs once eigenvalue decomposition and reduces thecomputation load. The performance of the proposed algorithm is very close to ESPRIT. It can pair theparameters automatically and eliminate the effect of the spatial colored noise. Simulation resultsverify its good performance.