Study On Multi-target Detection For Underwater Vehicle

Multi-target detection is the key technique of the underwater acoustic signal processing, and it plays an important role in many research fields, such as parameter estimation, system modeling and target identification. It has many engineering applications, ranging from military surveillance to mobile communications. Especially for the long-distance detection of underwate vehicle system, multi-target detection is a key technique to break through. Various multi-target detection methods are thoroughly studied in this thesis. A series of experiments are conducted in water tank and lakes, which validate the proposed methods correct and efficient.1. Several multi-target detection methods based on narrow band signals with Gaussian white noise are introduced, which are called RAIC, EIT, MEGM and GDE methods. The theories and algorithms of these methods are discussed. The detection performances of these methods are investigated respectively. Furthermore, some factors affecting the detection performances of these methods, such as the interval of angles, the snapshots and the number of array sensors, are investigated via computer simulations. The performances of these methods are evaluated. The methods based on eigenvalue grads have a risk of false alarm. A modification is introduced to improve the Probability of False Alarm (PFA) performance of EGMs. Computer simulation results show that the modification is efficient in PFA control, and brings only tiny loss of number-detection probability in very low SNR. The statistical detection performances of the aforementioned methods are compared and evaluated. The conclusions are given in this thesis in detail. The aforementioned methods except GDE require high SNR to distinguish those targets in spatial colored noise. Using the eigenvalues modification algorithm, the aforementioned methods can estimate the number of sources at low SNR in colored noise. But the correlation coefficient of the colored noise affects the detection performances of these methods.2. The typical wideband method, coherent signal subspace method (CSM) is studied, including the theory of CSM and the method of computing the focusing matrices. Simulations are conducted to test the performance of the different focusing matrices and focusing frequency .TCT focusing matrices outperform other focusing matrices at the expense of computation. The optimal focusing frequency is chosen as...