基于压缩感知的新型声信号采集方法  

New method for acoustic signal collection based on compressed sampling

作　　者：余恺[1] 李元实[1,2] 王智[1] 鲍明[2] 蔡盛盛[1] 

机构地区：[1]浙江大学工业控制技术国家重点实验室,杭州310027 [2]中国科学院声学研究所,北京100190

出　　处：《仪器仪表学报》2012年第1期105-112,共8页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(60873223;90818010);工业控制技术国家重点实验室开放课题(ICT1010;ICT1103);中国科学院无线传感网与通信重点实验室开放课题(No.2011001)资助

摘　　要：介绍了一种基于压缩感知的声音信号采集方法,特别适合于无线传感器网络的声学监控应用,能够大大降低采样过程中对硬件采样速率和能量的需求,同时减轻了无线通信的负担,从而实现无线传感器的低成本化和实用化。利用声学信号在频域中的稀疏性质,在压缩感知相关理论的指导下设计了一种新型的随机压缩采样方法,实现了信号有用信息的高效获取;同时,利用加权L1-magic算法对信号进行重构,完成完整的信号采集过程。仿真及实验表明,该方法能够实现在1/10以下标准采样频率下的声信号采集。In this paper, we introduce a method for acoustic signal collection based on compressed sampling, which is particularly suitable for acoustic monitoring applications in wireless sensor networks. The method greatly reduces sampling rate and energy requirement and also reduces the burden for wireless communications, making it possible for wireless sensor networks to be low-priced and practical. Making use of the sparse properties of acoustic signal in frequency domain, we design and implement a new randomly compressed sampling method under the guidance of compressive sensing,which achieves the acquisition of the useful information in signal. fill complete signal collection. Simulation and experiments Using L1-magic algorithm to reconstruct signal, we finally fulshow that the method we proposed can achieve acoustic signal collection with a sampling frequency below 10 percentage of standard sampling frequency.

关 键 词：压缩感知 随机压缩采样 信号重构 声学信号采集 传感器网络 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]