Location calibration and DOA estimation using randomly distributed acoustic sensor network

In this paper, we have conducted experiments on location calibration and DOA estimation using randomly distributed acoustic sensor network. We are concerned with the situation where sensor array locations are not exactly known due to the operational restrictions. Examples of such sensor arrays can be a number of cooperative hydrophones dropped from an aircraft or laid by surface vehicles. With an initial sensor location obtained by GPS reading, we adopt a maximum likelihood approach to calibrate the sensor location with a higher resolution which can be achieved by utilization of the intersensor distances calculated by active ranging between sensors. The DOA estimation is then compared using conventional beamforming, MVDR, and MUSIC methods based on the sensor locations before and after calibration.