Study On Theory And Applied Techniques Of 2-D Super-resolution Direction Finding Algorithm

In order to improve the angle resolution of the passive detector and make it have the capability of measuring multiple simultaneous signals, application of super-resolution DOA estimation algorithm to passive detector is studied in this dissertation. Several problems urgently to be solved are primarily researched, including estimating the number of signals, array error calibration in actual system and fast realization of the algorithm. Each subject is studied in detail, and corresponding calibration methods are proposed. The software and hardware of the direction estimation processor is designed and realized in this dissertation, and the experiments in the practical system prove these proposed methods valid and effective, and some indexes are tested in the system.In terms of source number estimation, based on the reality that noise in the system is correlated colour noise, two methods of source number estimation are proposed, one is determining the number of sources based on diagonal loading to the covariance matrix, the other is method of source number detection based on modified K-means clustering. The first method prove that the effect of small eigenvalues on the estimation of source numbers can be reduced by diagonally loading to the corresponding correlation matrix, then the information theoretic criteria method is available. The second method construct a distinguish criteria according to the difference between the signal and noise eigenvalues, and then the signal and noise eigenvalues are classified into two kinds using K- means clustering method. Both of the methods can be applicable in colour noise field, and their performance prefer to the Gerschgorin disks method. Computer simulation results and the available measured data confirmation prove the proposed methods valid.In terms of array error calibration, according to the reality that direction estimation performance degrades significantly when channel uncertainty exists, a calibration method for array antenna channel uncertainty using instrumental sensor is proposed. The method obtains the estimation of the channel uncertainty errors by processing the data information of the instrumental sensor's signal, so the influence of channel uncertainty is restrained, and the performance of DOA estimation is improved through errors compensation. According to antenna array position error, a correction method based on genetic algorithm is proposed. The method needs a instrumental source whose precise position is not necessary, and obtains the joint estimation for the antenna position error and direction of source. The method does not increase arithmetic time, because the correction can be done off-line.In terms of fast realization of 2-D DOA (Direction of arrival—DOA) estimation, a fast algorithm for 2-D DOA estimation based on multi-stage wiener filter(MSWF) is proposed. Utilizing the forward recursion of the multi-stage wiener filter, we can get the signal and noise subspace, so the calculation is reduced effectively. In order to reduce the calculation of peak value searching in 2-D DOA estimation, a new searching method based on large and small step size is proposed. The roughly estimated value is obtained using large step size searching, and then the precise value is obtained by small-step-size searching in the region of rough value. The calculation of peak value searching in 2-D DOA estimation is drastically reduced by using this method.In terms of design and implementation scheme of direction finding processor, according to the characteristic that data's dynamic range is very wide and the precision needs very high, four chips of high performance float DSP TS101 are used to realize the MUSIC (Multiple Signal Classification—MUSIC) algorithm, so the MUSIC algorithm can be realized in real time. The experiments in the practical system prove these proposed methods valid and effective, and some indexes are tested in the system. The design and implementation of the direction measuring system settle a basis for developing new type of passive detective system and engineering application of MUSIC algorithm.