Research On Direction Of Arrival Estimation Method For Underwater Acoustic Signal Based On Compressive Sensing

Direction of arrival(DOA)estimation is one of an important research content of array signal processing,which has a wide application prospect in underwater acoustic signal processing.Compressive sensing(CS)theory has injected new ideas into signal processing field,which breaks through the constraint of traditional Nyquist sampling frequency.DOA estimation based on CS theory has become a new research direction and has important theoretical and realistic significance for detecting,locating and tracking of the underwater acoustic signal.Under the background,DOA estimation of underwater acoustic signal is investigated based on the CS theory framework in this dissertation.The main research achievements are summarized as follows:Aiming at testing theory achievement value and demand for practical application scene in this dissertation,the cause of ship radiated noise is researched and spectrum characteristic is analysized.The corresponding mathematical model is established.The waveform and spectrum of the ship radiated noise are simulated,which act as underwater acoustic signal of theory application in this dissertation.Aiming at higher measurement cost of MMUSIC and ES-DOA in traditional array signal processing because of more hardware processing systems and greater amount of calcualation,the DOA estimation algorithms are proposed based on CS framework.CS array(CSA)structure is obtained by sampling array output and simultaneously compressing data,which can reduce the number of front end circuit and measurement hardware cost.On the basis of the CSA,two algorithms of CSA-MMUSIC and CSA-ES-DOA are proposed.Theoretical analyses indicate that two algorithms can decrease software computational burden.Simulation experiments verify that two algorithms can use fewer sampling data to achieve the similar estimation performance to the traditional algorithm,and have better DOA estimation performance for coherent and adjacent underwater acoustic signal by using fewer snapshot,lower signal noise ratio(SNR)and fewer number of measurement(M).Aiming at greater storage for random measurement matrix and difficult hardware implementation,structure method of deterministic matrix is proposed based on optical orthogonal codes(OOC).Firstly,the basis matrix is constructed by combining OOC and Singer structure.The deterministic matrix OOC-B is obtained by expanding basis matrix using Bernoulli matrix.Its property of asymptotical optimality is proven under a certain conditions and the column correlation is analyzed.Moreover,the specific parameter matrix examples are presented.Secondly,OOC-B matrix is applied to DOA estimation of CSA-MUSIC algorithm for underwater acoustic signal in order to study the performance acting as deterministic measurement matrix.Simulation results show that the estimation performance of OOC-B acted as a measurement matrix is better than that of OOC-H and random Gaussian matrix.And estimation bias is smaller in the lower SNR and fewer snapshot.Finally,OOC-B is applied to DOA estimation of spatial compressive reconstruction for underwater acoustic signal.Simulation results show that it has smaller estimation bias and higher successful recovery rate under the same SNR using OOC-B in DOA estimation.Aiming at the port/starboard ambiguity of acoustic pressure sensor linear array using MMUSIC algorithm,the DOA estimation algorithm based on CS vector hydrophone array is proposed.The structure of deterministic compressive sampling vector hydrophone array(DCV)is established by using deterministic measurement matrix OOC-B,which can reduce circuit cost.On the basis of this structure,the DCV-MMUSIC algorithm is proposed and simulation verification is carried out.Results show that the algorithm not only has advantages of lower computational burden and suppressed port/starboard ambiguity,but also can apply to DOA estimation for coherent underwater acoustic signal by using lower SNR,fewer snapshot,and fewer M.In addition,DOA estimation is done for actual merchant ship radiated noise signals using the algorithm.Simulation results show that the estimation bias of merchant ship radiated noises is slightly greater than that of simulated underwater acoustic signal in different SNR,snapshot,and M,but,it still has better estimation performance.