Research On Array Calibration And Direction Finding Of Distributed Sources

In the field of array signal processing,the direction of arrival(DOA)estimation is a hot research topic,as well as an important task of radar,sonar,mobile communication and biomedicine systems.Most conventional estimation algorithms are based on the assumptions that there is no gain-phase errors between the array channels,the position of the array elements is exactly known,and the signal model is a point target or a point source.However,in the practical engineering applications,the performance of the algorithms is degraded severely as these assumptions are difficult to meet.In this paper,in the case that the ideal assumptions are not satisfied,the estimation of array errors and DOA estimation of coherently distributed sources are studied respectively.The main contents of this paper are as follows.Based on the active gain-phase errors estimation algorithm,the analytic expression between the estimation errors of gain-phase errors and the deviation of auxiliary source azimuth measurement is deduced.an active gain-phase errors estimation algorithm using auxiliary sources in unknown directions is proposed.Compared with the conventional calibration algorithm for gain-phase errors,the proposed algorithm does not need to measure the DOAs of the two auxiliary sources,the DOAs of the calibration sources are estimated instead of the actual measurement,so that the additional errors brought by the inaccurate direction measurement can be avoided.Then,the gain-phase errors estimation of wideband array channel is verified by using the measurement data,and the mean and standard deviation of error parameters are given.Based on the data model where the gain-phase errors and position perturbations of array elements both exist,a new method is proposed to calibrate both the gain-phase errors and position perturbations of array elements.In the conventional joint calibration algorithm,the errors is normally estimated by setting a far-field source in known direction.In the practical engineering applications,the direction measurement of the far-field source is not convenient.The proposed method replaces the direction measurement of the far-filed source by measuring the rotation angle of the array to be calibrated,so that it is more convenient to achieve the engineering needs.The one-dimensional DOA estimation of coherently distributed sources is studied.Since the coherently distributed source is characterized by the central angle and angular spread,the two-dimensional search or iterative operation are normally required in the existing algorithms,so that the computational complexity is high.In order to avoid computational complexity and construct a special array structure,based on the uniform linear array,a serach-free rooting algorithm is proposed to estimate the central DOA of coherently distributed sources.The proposed algorithm has both low computational complexity and high estimation accuracy.Moreover,the angles are estimated without any information about the deterministic angular distribution functions of the distributed sources.The two-dimensional DOA estimation of coherently distributed sources is studied.For a centrosymmetric area array,the property that the angular signal distributed weight vector of the coherently distributed source has a symmetric structure is proved.Making use of this symmetric property,an algorithm for the two-dimensional central DOA estimation of coherently distributed sources is proposed.The estimation accuracy of the proposed algorithm is high,and the central directions can be found independently of the deterministic angular distribution functions.However,the algorithm requires two-dimensional search,and the computational complexity is high.Therefore,the proposed algorithm has been improved.Based on a pair of centrosymmetric area arrays,we prove the generalized array manifold matrix has the rotational invariance relation between the two sub-arrays,the improved algorithm uses the sequential one-dimensional searches to replace the two-dimensional search.Both low computational complexity and high estimation accuracy are obtained.