The Real-Time Implementation Of Direction-of-Arrival Estimation By Small-Size Array Based On Multiple DSP

The direction-of-arrival (DOA) estimation is one of the most important techniques of array signal processing. Traditionally the DOA estimation is implemented by beam-forming the acoustic signal received by the hydrophone array distributed in the water. Because of the Rayleigh constraint of the resolution of general beam-forming, once the length of the array is fixed, its resolution is fixed. If the low frequency signal is needed to be processed, a long enough array is needed, which will increase the cost and bring about inconvenience in experiment.In the paper, the small-size vector hydrophone array is studied to distinguished with multiple targets. Aiming at improving target resolution ability in array signal processing, several methods are proposed, such as frequency division, imagined-sensor and constant beam-width, among which dividing frequency can reduce the affection of signals in the different or close frequency segments, increasing the number of imagined-sensors can get rid of the grate lobes produced by high frequency and increase the array's size at the same time, while constant beam-width can form the same narrow beam-width by the same array to the signals of different frequencies .The signal is phase-shifting beam-formed by offline reconstruction which is based on self-adaptive notch filter.The signal processing methods of the small-size vector hydrophone array to distinguish multiple targets are implemented on DSP. The system includes 4 pieces of TMS320C6416 DSPs. The data transfers from SDRAM to 3 pieces of DSP and is processed on each piece in one frequency band, then the outputs of the processing are transferred to the last piece of DSP by McBSP and integrated on it.In the paper, lots of calculation has been realized, using the data measured from the experimentation on lake. All the algorithms and software are proved viable and credible.