基于改进m序列的压缩采样观测矩阵设计
Design of Compressed Sampling Observation Matrix Based on Improved m Sequence

为了降低对嵌入式系统数据采集的硬件要求，提出了一种适用于模拟信号采集系统的压缩采样方法，为压缩感知观测矩阵提供了一种低功耗硬件实现方法。介绍了观测矩阵设计的相关要求，并提出了带压缩采样矩阵的高斯随机观测矩阵及基于改进的m序列的压缩采样矩阵硬件实现方法。考虑系统功耗及外围电路复杂度问题，提出利用MSP430微处理器完成压缩采样系统设计，并利用滚珠丝杠动态测试中的振动信号完成对系统的验证。实验结果表明，该系统能以低于亚乃奎斯特采样频率的采样率对振动信号进行压缩采样，能较为精确地重构出原始信号。
A compression-sampling method applied to the analog signal is proposed for the embedded system to decrease its power consumption. The proposed method reduces the requirements on the hardware when compressing the observation matrix. First, rip criterion are set for the observation matrix. Then, the compressive sampling matrix is constructed based on improved M-sequence. MSP430 microprocessor is chosen concerning the system power consumption and the complexity of the peripheral circuit. Finally, the system is verified by detecting the vibration signal in the ball screw dynamic test. The experimental results show that the vibration signal can be collected at the frequency lower than sub-Naikui sampling frequency and the original signal can be reconstructed accurately.