Research On Multidimensional Array Signal Processing Based On Compressive Sensing

Ocean acoustic tomography,based on the principle of "sound wave energy travels in the ocean for long distances" and "the sound velocity of sea water is a sensitive function of seawater temperature",has become a large-scale ocean temperature monitoring in the context of global warming.An efficient way to measure the temperature rise of the earth indirectly.Due to the multi-path propagation characteristics of the shallow sea,if a point source is placed in the sea area to be tested,the hydrophone array at the receiving end will observe the arrival signals arriving at different angles at different times.The arrival angle of the arrival signal reflects the propagation path of the water layer passing through different depths.The arrival time reflects the cumulative influence of the water temperature of the water passing through the sound wave.Therefore,the angle information and the arrival time information can be used to invert the sound velocity profile of the sea area.Therefore,the high-precision separation and recognition of the sound path of the received signal and the joint estimation of the angle-delay are forward problems of high-quality ocean acoustic tomography.The double array architecture expands the point source of the transmitting end into a transmitting array,and expands the estimated data of the sound path to the "emission angleangle of arrival-propagation time" three-dimensional domain while expanding the observation data to the multidimensional signal.The high-resolution multidimensional array signal processing algorithm under the dual-array architecture,such as the high-order smooth dualarray sound path separation algorithm,can clearly and clearly distinguish the acoustic path that cannot be recognized due to the similar arrival angle and propagation time in the threedimensional estimation space.A joint estimate of the launch angle-angle of arrival-propagation time is identified.The algorithm utilizes the fourth-order cumulant enhancement algorithm for noise robustness and the number of separable sound path paths,and uses three-dimensional smoothing to process coherent signals in multidimensional array signals to obtain highresolution recognition and estimation accuracy and Gaussian noise environment.High robustness and low dependence on the number of elements.However,the disadvantage is that the use of statistics and three-dimensional smoothing requires extremely high computational cost.In addition,the smoothing operation causes aperture loss and is not adaptable,and it is prone to insufficient smoothing or smoothness.Therefore,improving the current multidimensional array signal processing algorithm has important practical significance for the development and application of the dual array architecture.Compressed sensing theory,as the latest achievement of information theory,provides a complete theoretical basis for the reconstruction of sparse vectors,and also provides a new solution to the traditional array signal processing problem.This thesis studies the dual array architecture and compressed sensing theory.Firstly,it introduces the research background of the subject,then introduces the dual array architecture and its related algorithms,and focuses on the shortcomings of the sound path separation technology under the dual array architecture,and puts forward the performance requirements for the theoretical results of this thesis;The main content of compressed sensing theory and its related algorithms are introduced,which provides the theoretical basis for the subsequent chapters,and also puts forward the technical route requirements for the theoretical results of this thesis.Based on the analysis of the dualarray architecture signal model and the compressed sensing theory,the research work has been carried out from the following aspects:1.The space-time sparsity characteristics of the acoustic raypath under the dual array architecture are analyzed.Based on the traditional method of multidimensional data dimensionality reduction technology,combined with the mathematical model of compressed sensing theory,a dual array architecture based on compressed sensing is proposed.The multidimensional signal model is analyzed,and the two application preconditions of the signal model satisfying the theory of compressed sensing are analyzed,which provides a feasible basis and theoretical basis for the subsequent algorithm.Finally,the key steps to solve the problem of sound path separation in dual array architecture by using compressed sensing theory are summarized.2.The characteristics of greedy tracking algorithm for solving compressed sensing reconstruction problem are analyzed.The algorithm of double array compressed sensing raypath separation based on orthogonal matching pursuit is proposed.The dictionary optimization technique is introduced for the characteristics of over-complete dictionary atomic correlation.A dual-array compressed sensing raypath separation algorithm based on perceptual dictionary is proposed.The performance characteristics of the above algorithms are analyzed,and the performance of the above algorithms is tested and verified by simulation experiments.Finally,the effectiveness of the above algorithm is verified by the actual water tank test and ocean test.3.The characteristics of the convex optimization algorithm for solving the compressed sensing reconstruction problem are analyzed.The weight matrix is introduced for the difference between the norm optimal problem and the norm optimal problem.Then the double array raypath separation algorithm based on weighted convex optimization is proposed.The performance characteristics of the above algorithms are analyzed,and the performance of the above algorithms is tested and verified by simulation experiments.Finally,the effectiveness of the above algorithm is verified by the actual water tank test and ocean test.Finally,the thesis summarizes the research content,analyzes and compares the advantages and disadvantages of the above algorithms,and points out the direction and problems to be further studied in order to apply the theory of compressed sensing more effectively.