Study On Direction Of Arrival Estimation Technologies With A Long Linear Array Under Non-gaussian Background Noise

Underwater detection systems are facing great challenges due to the ever decreasing target noise level, which pressures the development of an increasing aperture linear array in underwater target detection system. However, in the actual ocean environment, the performance of the long linear array would degrade, and even fail because of the interference effect of normal waves in the underwater acoustic channel and the non-Gaussian background noise. Therefore, deep studies on DOA estimation methods of a long linear array in complex background noise and shallow water waveguide are solid foundations for its future application and the development of a high performance underwater target detection system.In this thesis, DOA estimation methods for long linear array working in the shallow water waveguide and complex background noise were explored, and the main work and innovations are as follows:1. The DOA estimation model for the long linear array is established based on the normal wave method. Relations between the horizontal coherence and factors such as the signal frequency and the inter-element space in shallow water are derived and analyzed. The phenomenon of the split or shift of the spatial spectrum in target directions is explained theoretically and through simulations. The results show that the horizontal coherence would fluctuate periodically, and the summation of many normal waves in shallow water would lead to the split or shift of the spatial spectrum in target directions.2. The subspace intersection methods and the improved methods are summarized. Simulations of comparisons between these methods are conducted under the scenarios, such as different SNR, snap shots and signal frequencies and mismatch conditions of the horizontal wave number, the mismatch of the number of normal waves, and the source number mismatch. The advantages and disadvantages of these methods and their applicability in the shallow water are also explored.3. The definition of cumulant and the principles of its property in restraining the Gaussian noise were introduced. The performance of higher-order cumulant DOA estimation algorithm was discussed by combining MUSIC algorithm with the subspace intersection algorithm that both based on plane wave model. Higher-order cumulant algorithm has greater normalized amplitude and a good ability to suppress the color noise. Finally, fourth-order cumulant subspace intersection algorithm based on shallow water waveguide was analyzed. In shallow water waveguide, the fourth-order cumulant subspace intersection algorithm can achieve accurate DOA estimation under Gaussian white noise, Gaussian distributed spatial colored noise, Rayleigh white noise and Rayleigh distributed spatial colored noise, respectively.