Research On Array Multi-parameter Estimation Algorithm Based On Parallel Factor Model

Direction of arrival(DOA)estimation is a key point of array signal process and has motivated considerable concerns of researchers recently.Parallel factor(PARAFAC)analysis,which utilizes the uniqueness of PARAFAC decomposition to obtain the parameter estimations,is a common method for the problem of DOA estimation of received signals and has been investigated for numerous engineering fields.Usually,conbining the compressed sensing theory with PARAFAC model can partly reduce the computional complexity and capacity for data storage of PARAFAC method.This paper combines the trilinear decomposition,Quadrilinear decomposition and parallel profiles with linear dependencies(PARALIND)decomposition of PARAFAC model with compressed sensing theory,respectively,and can obtain the multi-parameter estimation for different scenes with low computational complexity.The research of this paper on DOA estimation algorithm via compressed sensing PARAFAC framework has great significance in theory and application.The main work of this paper is organized as follows:1)A compressed trilinear decompositiom based two-dimensional DOA(2D-DOA)estimation algorithm is proposed for acoustic vector-sensor arrays.The proposed algorithm first compresses the received data according to the compressed sensing theory,and then the 2D-DOA can be estimated from the acoustic vector-sensor matrix after trilinear decomposition.The proposed algorithm can be applied to arbitrary acoustic vector-sensor arrays and can obtain automatically paired 2D-DOA estimation with no need for sparse recovery process.Due to the compression process,the proposed algorithm has lower computational complexity than traditional PARAFAC method,while it's angle estimation performance is close to the latter and outperforms the estimating signal parameters via rotational invariance techniques(ESPRIT)algorithm and propagator method(PM).2)A compressed sensing trilinear model based 2D-DOA and frequency estimation algorithm for L-shaped array is proposed.This method first compresses the multi-delay outputs data of received signals with partitioning according to the compressed sensing theory,then obtain the angle and frequency estimation via trilinear decomposition and sparse recovery.Compared with traditional PARAFAC method,the proposed algorithm needs lower computational complexity and smaller capacity for data storage.It can obtain automatically paired 2D-DOA and frequency estimation and resentful for both uniform and non-uniform L-shaped array.In addition,the angle and frequency estimation performance of the proposed algorithm is close to the traditional PARAFAC method,and outperforms the ESPRIT algorithm and PM algorithm.3)A compressed sensing quadrilinear model-based algorithm is proposed for 2D-DOA estimation of electromagnetic vector array.Owing to the compression process,the proposed algorithm needs smaller storage capacity and has lower computational complexity,compared with the traditional quadrilinear method.The angle estimation performance of this method is close to the traditional quadrilinear method and better than ESPRIT algorithm.Moreover,it can distinguish closely-spaced sources and achieve their angle estimations effectively.4)A compressed sensing PARALIND decomposition model-based algorithm is proposed for 2DDOA estimation of coherent sources of uniform rectangular array.The proposed algorithm combines the compressed sensing theory with PARALIND model and owing to the compression process,the proposed algorithm holds the properties of lower computational complexity and smaller capacity for data storage,compared with traditional PARALIND decomposition algorithm.It also works well to acquire automatically paired azimuth angles and elevation angles.The angle estimation performance of the proposed algorithm is close to the traditional PARALIND method,and better than the forward backward spatial smoothing-ESPRIT(FBSS-ESPRIT)algorithm and FBSS-PM.