2-D DOA Estimation Based On Manifold Separation

Array signal processing means that several sensors form a sensor array in the space.After the array received the signal,it will sample and extract the signal parameter.Direction of arrival(DOA)estimation is a common problem in array signal processing.The classical multiple signal classification(MUSIC)algorithm can be used to eigenvalue decompose the covariance matrix of snapshot data to get the signal subspace and noise subspace,then to estimate the DOA of signals by utilizing the orthogonality between these two spaces.In general,spectral search is used to implement the DOA estimation.The classical MUSIC algorithm in DOA estimation has high resolution and stability.In order to obtain higher estimation accuracy,a smaller step of spectral search will be used,especially in joint estimation of azimuth and elevation.However,it will lead to an unacceptable amount of computation.People have tried a lot harder to reduce the amount of computation,with producing a series of excellent algorithms.The root-MUSIC algorithm uses polynomial rooting and the LS-ESPRIT algorithm utilizes two same sub-arrays to receive the signals,both of which can reduce the amount of computation without requiring spectral search.While the manifold separation technology(MST)make the geometric structure of array not restrict to a certain mode,and it can also transform two-dimensional(2D)DOA estimates into binary polynomial rooting.However,these algorithms have some limitations,which are going to be discussed in this thesis.Above all,to reduce the computational complexity of the DOA estimation algorithm is a vital research direction.This thesis proposes an efficient algorithm to compute the 2D spatial spectrum,which consists of two main stages.The first stage is offline processing,which calculates the sampling matrix by measuring the array response at a number of locations,and then calculates the relevant auxiliary matrix needed in the online processing.The second stage is online processing,which include calculating a truncated coefficient matrix and applying partial 2D DFT to calculate the 2D spatial spectrum.This algorithm is parallel and its complexity is almost independent of the number of antennas,so it can be applied to parallel computing systems.This thesis has two innovations as following:(1)For the calculation of the coefficient matrix,this thesis proposes a one-step method to calculate the coefficient matrix,which obviously simplifies the calculation.This thesis also proves the symmetry property of the coefficient matrix,which shows that we can get all the coefficients needed in the algorithm just by computing a quarter of the coefficients.(2)According to the feature of the coefficient matrix where non-zero coefficients are only concentrated in the upper left corner,this thesis proposes a partial 2D DFT to calculate the 2D spatial spectrum,which can effectively reduce the amount of calculation.A large number of simulation experiments show that compared with the classical MUSIC algorithm,the proposed algorithm can reduce the amount of computation when their estimated performance is almost the same.