Non-uniform linear sonar array based DOA estimation

In most of available DOA estimation algorithms, such as MUSIC, Capon algorithm, and so on, the Nyquist spatial sampling theory must be followed. Otherwise, ambiguous azimuth angles will appear in DOA estimation results, which lead to inconvenience on the judgment of target azimuth angle. To improve the sonar´s resolution, a general method is to increase the sensor elements then expand its observation aperture. Accordingly, the cost of manufacturing and data processing are increased. To achieve high resolution with few sensor elements, a kind of non-uniform linear random sonar array is designed, which breaks the Nyquist sampling rate restriction, and then the MUSIC algorithm is used to estimate underwater targets azimuth angles. Simulation results show that, with the same aperture, the performance of non-uniform array is extremely improved over uniform array under high signal SNR. But under low signal SNR, it´s performance decreases also. Then a mean method is proposed to solve this problem.