紧致麦克风阵列压缩采样与DOA估计方法

Compressive sampling and DOA estimation method for miniature microphone array

针对紧致麦克风阵采样信号大量冗余的问题，提出了一种ΣΔAD压缩采样方法。该方法将软硬件相结合，在ΣΔAD转换器内部进行压缩采样。压缩采样中采用自适应过程，去除信号中的冗余分量，并将压缩后的信号进行稀疏编码。仿真结果表明，使用该方法对紧致麦克风阵接收信号进行压缩编码时，通过选取合适的稀疏化阈值，可使源数据的压缩比达到10％～30％。压缩采样后的信号可以用于DOA估计等应用。针对八元紧致麦克风圆阵和DSP实时系统的DOA估计实验结果表明：这种DOA估计方法在阵元间距低至2 cm时仍能正常工作；当阵列尺寸减小时，相比经典MU SIC算法和PHATG-C C算法，该方法定位精度更高，噪声鲁棒性更强。

Aiming at the problem of the redundanyc of the sampled data ina compressive micro-phone arar y, a new ΣΔAD （ analog-digital ） compressive sampling method is proposed.Combined with software and hardware, the input signals are compressively sampled within theΣΔAD convert-er.With the help of adaptive estimation procedure in sampling, the redundancy of the input signal is removed and the sparse output signal is ne coded.The simulation results demonstrate that the com-pression rate is 10%to 30%when the proper sparseness threshold is set for the miniature microphone array during the signal coding process by this method.The compressively sampled signal can be used in the applications such as direction of arrival （ DOA） estimation.The experimental results for an eight-component miniature microphone arrya and a real-time id gital signal proec ssings yset m demon-strate that when the aperture of the micropho nea rray is low to 2 cm, teh direction of the arrivalsi es-timaet d successfully by this DOA estimation method.When the aperture of the array dce reases, com-pared with the traditional multiple signal classification （ MUSIC） algorithm and the phase transform generalized cross-correlation （ PHAT-GCC） algorithm, the proposed DOA estimation method exhib-its higher positioninga ccuracy and noies robustness.