Research On Array Calibration And Direction Finding Based On Measured Data

DOA estimation is a main research of array signal processing. And the direction finding technology of spatial spectrum based on the super-resolution algorithm is an important research point in DOA estimation. Since the super resolution direction finding algorithms with multiple signal classification thought was brought forward, because of its superior excellent performance, it have received abroad attention and research. However, these algorithms need the accurate array steering vector as a premise. In practice application, variety non-ideal factors make the error between the ideal steering vector and the actual large. As a result, the algorithm performance deteriorates greatly. Sometimes, the algorithm even stop working. From the measurement of the actual steering vector to array calibration, a large number of methods were proposed. This speeds up the algorithm from theory into practice. The practical application of the algorithm is the focus of research in recent years. In this dissertation, the performance of the super-resolution algorithm based on the practical data is studied.Firstly, errors that may occur during propagation in real environment are modeled in this dissertation. Based on the amplitude and phase characteristics based on the measured data, the form and category of existing errors are discussed.Then, because that the super-resolution algorithm requires the number of signal sources to be known, several classic methods to estimate the number of sources are presented. By analysing the distribution characteristic of the eigenvalues for the measured data, several improved algorithm are studied. Simulation and real data experiments results verify effectiveness of the improved algorithm.Next, two kinds of super resolution direction finding algorithms, MUSIC and PM, are introduced. Combined with the form of the errors presence in the measured data, the influences of amplitude error and phase error on the direction finding algorithm of multiple signal class are discussed, respectively. Moreover, based on the characteristics of direction finding algorithm and conditions of the experiment platform, two methods to get the actual steering vector are studied. The direction finding experiments by using the proposed method to obtain the actual steering vector are performed with the measured data of the single source and dual sources. At last, for the single error model which are amplitude and phase error or mutual coupling and several errors model that different types of error are existing at the same time, several classical calibration algorithms are listed. Their performance in simulation and real data experiments are studied. The test results of the measured data shows that these algorithms are only applicable to the presence of specific error conditions and not suitable for that where there are different error types at the same time. Due to a variety types of error existing in measured data, the correction algorithms don’t improve the performance.