Study Of Raypaths Separation In A Shallow-Water Waveguide Based On Higher-Order Cumulants

In shallow water, multiple raypaths propagation is caused by reflection and refraction of the surface and the bottom. Each raypath is the copy of the emitted signal, so these raypaths are fully correlated or coherent. To separate those raypaths, many direction finding algorithms have been developed. However, these algorithms have good performance under the assumption of white Gaussian noise. Actually, there always exists colored noise in real ocean environment. The model assumption of existing raypath separation algorithms do not match the actual ocean environment. Based on the above analysis, in this thesis, we propose a higher-order wideband active MUSIC algorithm, named 4-Smoothing-MUSICAL algorithm. The proposed algorithm not only can suppress the white Gaussian noise, but also can separate the raypaths interrupted by colored noise. Comparative study shows that 4-Smoothing-MUSICAL algorithm has the following advantages compared with the existing smoothing-MUSICAL algorithm based on second order cumulants: (1) it achieves a more accurate separation, especially when the raypaths arrive closely; (2) it can suppress both colored noise and white Gaussian noise, and has fewer artifacts; (3) it is not necessary to assume that the number of sensors is larger than that of sources. Although 4-Smoothing-MUSICAL algorithm can improve the ability of raypath identification, it has its own limitation which is very time consuming, so that it can not be used for real-time detection of raypath.In order to solve the above mentioned shortcoming, and make full use of the emitted signal's spectral information, we develop an active wideband multiple signal classification algorithm named Root-MUSICAL algorithm. The Root-MUSICAL algorithm examines the roots of the polynomial to separate different raypaths, so it doesn't need to waste time to search the time and direction of arrival which lead to the peaks of spectral function. Thus this algorithm consumes less time, and it can be used for real time detection. Results of simulations illustrate that it greatly reduces the identification time of raypaths compared with proposed Smoothing-MUSICAL algorithm. And the identification precision is rather good.