Target Location And The Realization By Multi-beamforming

The target locating is a classic and important problem of the underwater signal processing. Based on the theory of array signal processing, the paper is devoted to the study of the theories and application technologies of DOA estimation and underwater array signal processing in a systematic and detailed way, especially on the design of the digital multi-beamforming system. The Blind source separation and Independent component analysis is also involved in the paper. The main contributions are as follows:1. The fundamental concepts and theories of array signal processing are discussed. The definition and presentation of the narrowband signal is introduced briefly, and the technology of orthogonal sample is illustrated. Then the statistical model of array signal processing is presented.2. The design process of the multi-beamforming is presented detailedly, which include the orthogonal demodulation and resample technology, the schema of phase shift beamforming, and the effect of the frequency shift and the phase disturbance. The secondary filter is designed to eliminate the influence of the close frequency. The simulation results show that the requirements are implemented.3. The hardware architecture of the system is analyzed, especially on the properties of TMS320C54x and the methodology of software development. The system software is developed and debugged in modularization. The simulation results based on the target board and experiment results in the pool and the lake show that the system running properly. All the functions and parameters required are implemented: 24 digital beamform output correspond to 32 analog input, three different frequency work in the same time and the AGC.4. Based on the ICA analytical method, the theories and some efficient methods of blind source separation(BSS) are studied. It is pointed out that the indeterminacy of BSS is the amplitude and the order of the recovered sources. The ICA estimation principle of Nonlinear decorrelation and Maximum nongaussianity is presented. I showed the relationship of different learning rule, e.g. infomax, MLE, negentropy maximum and mutual informationminimum, and presented a general learning rule. A DOA estimation is developed by the blind estimation of the array manifold. The simulation show that the eXinfomax algorithm can separates 10 mixture of super-gaussian and sub-gaussian signal, and the minimal amplitude of the source is 10-9.