无线音频传感网络中的压缩采样

在无线音频传感网络中，降低带宽一直是研究人员关心的问题。新近出现的压缩采样技术，以远低于奈奎斯特频率的速度采样，精确地恢复出原始信号。压缩采样，针对稀疏信号，根据测量矩阵进行随机采样，应用最优化算法进行信号重构，恢复出原始信号。以随机抽取的局部DCT矩阵作为测量矩阵，采用CoSaMP作为恢复算法，实现了音频传感网络中的压缩采样。实验结果表明，在采样速率降低4倍的条件下，音频信号能够清晰恢复，信号处理速度完全满足实时性要求。

Compressive sampling in wireless acoustic sensor networks

The network bandwidth is utilized more effectivly with compressive sampling in wireless acoustic sensor network. The compressive sampling is a new theory that combines sampling and compression into a single process. It captures sparse signals at a rate significantly below the Nyquist rate, and recovers exactly the original signals using convex optimal algorithms. This compressive sampling is implemented in a wireless acoustic sensor network,where the measurement matrix is randomly decimated from the DCT matrix, and the restore algorithm is CoSaMP. Through experiments, we find that the acoustic sig- nal can be clearly restored at the one fourth of normal sampling rate and the back-end computer can process the acoustic signal at real-time speed.