基于块稀疏贝叶斯学习的体域网心电压缩采样∗

摘要

In order to improve the real-time performance and decrease the power consumption of the body sensor network,this paper proposes an ECG compressed sampling method of body sensor network based on block sparse Bayesian learning. In the body area network framework,the proposed method,using compressive sampling theory, use binary random measurement matrix to compressive sample ECG on the sensor nodes. After measured value are transmitted to remote monitoring center,the block sparse Bayesian learning reconstructed algorithm and the discrete cosine transform matrix and are used to reconstruct the ECG signal. The experiment results show that the SNR which base on block sparse Bayesian learning reconstructed algorithm is 3 dB~21 dB higher than that of the other recon-structed algorithm when the compression rate of ECG is at 70%~90%. The method can effectively reduce the data sampling,the subsequent pressure of data storage and data transmission,and improve the real-time performance of body area network. The method also has the advantages of low power and easy to hardware implementation.%为有效提高体域网的实时性和降低体域网的功耗，提出一种基于块稀疏贝叶斯学习的体域网心电压缩采样方法。该方法在体域网框架下，利用压缩采样理论，在体域网的传感节点利用二进制随机观测矩阵对心电信号进行压缩采样，远程监护中心获得采样值之后，利用块稀疏贝叶斯学习重构算法和离散余弦稀疏变换矩阵对心电信号进行重构。实验结果表明，当心电信号压缩率在70%~90%时，基于块稀疏贝叶斯学习的重构算法要比其他重构算法的重构信噪比高出3 dB~21 dB。该方法能有效减少数据采样，减轻后续的数据存储、数据传输压力，提高体域网的实时性。同时该方法具有功耗低，易于硬件实现的优点。