Compression Sampling System of Vibration Signal Based on Sparse AR Model

Abstract

In this paper, we propose a compression sampling system of vibration signal based on sparse AR (Auto Regression) model. It exploits the Compression Sensing (CS) theory and the architecture of Simple Random Sampling (SRS) system. A basis matrix is constructed based on sparse AR model for reconstructing the received vibration signal. The basis matrix is named SAR basis in this paper, in which the atoms are all prior vibration signal components. The signal can be reconstructed by optimization algorithms with the simple random sampling measurements and the SAR basis. For the case of signal sampling, SRS method scales down the sampling frequency efiectively. Additionally, since the SAR basis is a self-adaptive basis, a desired high reconstruction quality at a low sampling rate can be obtained. From both simulations and experiments, the results show the efiectiveness of the compression sampling system proposed in the terms of reconstruction accuracy (SNR) and Compression Ratio (CR).