The Design and Implementation of A Smartphone-Based Acoustic Array System for DOA Estimation

Microphone array for high-quality audio recording has been realized by dedicated devices. But lack of portability and high cost prevent such system being widely adopted. Today's commercial-off-the-shelf (COTS) smartphones are relatively more convenient for audio recording. However, in order to estimate the target's direction of arrival (DOA), several asynchronous receivers (smartphones) must be precisely time synchronized. Nevertheless, the additional computational cost and errors caused by realizing time synchronization with traditional methods may lower the accuracy of DOA estimation and even make the DOA estimation failure. In this paper, we design and implement the Smartphone-based and DOA-estimation acoustic array system called SDOAs, in which the effective and straightforward RBS-based and Post-Facto Synchronization (RBS-PFS) mechanism with sample-counting technique is proposed to figure out the problem of time asynchronization. The real SDOAs is implemented on a test-bed using several android smartphones. The experiment demonstrates that our acoustic array system SDOAs can achieve high accuracy DOA estimation with Multiple Signal Classification (MUSIC) algorithm and the RMSE is lower than 2.7 degrees when the range of degree is between 0 degrees and 45 degrees.