Abstract: Conventional reconstruction methodologies of compressed sensing for pacing ECG is suffered significantly from various noise through telemonitoring pacemaker by wireless communication networks. A novel method with the ridge regression regularization parameter K in the iteration of residual error is proposed to reduce the impact of noise to the recovery results. By ridge regression, it proves that superior K is negatively related to SNR, which provides a theoretical evidence to choose the suitable K and obtains the most accurate recovery signal. Comparative analysis is also introduced among the recovery algorithm based on the ridge regression, the Block Sparse Bayesian Learning and the orthogonal matching pursuit. The results show that the ridge regression method can not only maintain high recovery efficiency under the low SNR environment and increase the recovery accuracy of pacing ECG simultaneously.

Key words: compressed sensing, pacing ECG, recovery algorithm, ridge regression

摘要： 针对传统压缩感知重构算法在起搏心电信号远程监测过程中易受噪声干扰的问题，提出在利用正交匹配追踪进行残差更新的迭代过程中引入岭回归正则化参数K，降低噪声对重构结果的影响。利用岭迹法证明了最佳K值与信噪比呈负相关，为选取K值以获得更接近真实解的重构信号提供了理论依据。对基于岭回归的重构算法与分块稀疏贝叶斯学习算法、正交匹配追踪算法进行了对比分析，实验结果表明，在低信噪比环境下，引入了岭回归思想的算法在保留高重构效率的同时提高了重构精度。

关键词: 压缩感知, 起搏心电信号, 重构算法, 岭回归