Research On High Resolution Array Processing Algorithms

Ocean acoustic tomography uses changes in the sound velocity of shallow water to invert changes in ocean temperature and even climate changes.In the shallow water environment,there are reflections and refractions in the acoustic signal propagation.The acoustic signal will propagate in a multi-path way,and each signal(raypath)is a duplicate of the original one.Ocean acoustic tomography requires the separation of these different raypaths.However,the acoustic signals produced by multipath effects are coherent to each other and result in great interference to be separated.Most of the previous signal separation algorithms assume that the noise is Gaussian white noise.Actually,shallow-water environment noise can be classified as colored noise rather than white noise.The previous algorithms will get much worse resolution and lower robustness when dealing with colored noise compared with white noise.Additionally,the algorithms are usually based on second-order statistics,cannot fully utilizing the statistical information of the received data,and have certain requirements on the number of sensors.The paper proposes a high-order cumulant raypath separation algorithm based on dual arrays.Compared with previous secondorder statistics algorithms,the proposed algorithm makes full use of the cumulants and it is transparent to both Gaussian white noise and colored noise.This characteristic has a good noise suppression result.Meanwhile,the fourth-order cumulants have the function of element expansion that will increase the number of sensors and extend the effective aperture.Under the conditions of simulation experiment,water tank experiment and marine experiment,the proposed algorithm can improve the signal separation resolution and robustness to noise compared with the ones based on second-order statistics.In recent years,compressive sensing has been successfully used in signal processing.It avoids the requirement that the sampling rate must be greater than or equal to twice the signal bandwidth under the Nyquist sampling theorem,which enables the accurate reconstruction of the original signals with much less data.Because of the advantage of compressive sensing,DOA(Direction of Arrival)estimation algorithms based on compressive sensing have been gradually proposed and some certain results have been achieved.However,most the algorithms only focus on the aspect of compressive sensing,and do not combine the characteristics of compressive sensing with traditional algorithms.Therefore,we propose in this thesis a subspace-based compressive sensing DOA estimation algorithm.Under the conditions of simulation experiment,water tank experiment and marine experiment,compared with traditional algorithms and some compressive sensing ones,the resolution and robustness to noise can be improved.