Research On Analog Signal Acquisition System Based On Compression Sensing

Compressed sensing theory is proposed in 2006. So far, the research object of compressed sensing is mostly known, discrete sequence, and its hardware implementation are facing great difficulties. Therefore, analog signal acquisition system based on compressed sensing is investigated in the paper, and a new hardware implementation method of compressed sensing measurement matrix is provided.The traditional sampling theorem is firstly introduced briefly in the paper, and the realization process and the basic concept of compressed sensing theory are researched. In the third chapter, the existing observation matrix is introduced, and the analog signal of random sampling in the compression facing problem is studied. In view of this, the use of random sequence to achieve compression sampling thought and based on the "rotation matrix" and "the two observation" thought of the observation matrix design scheme are proposed. The matrix is given in MATLAB, and the coherence parameters of discrete Fourier transform based (DFT) and the restricted isometry constraint(RIC) are calculated. The observed performance matrix is simulation in the end, and the Gaussian matrix with compressive sampling matrix is selected as the observation matrix for analog signal acquisition system. The overall design diagram of analog signal acquisition system is introduced in the fourth chapter; the two different random sequence generation methods are proposed in the acquisition system and "m" sequence generator circuit is given. According to the design requirements of the system, the MSP430G2 MCU is selected as the system hardware; the MCU and its peripheral modules are introduced. At the end of this chapter, the software design flow chart of the two different random sequence generation system is proposed. The existing convex optimization algorithm, greedy algorithm and its improved algorithm for compressed sensing reconstruction algorithm are studied in the fifth chapter, the related parameters are analyzed, including the reconstruction time, residual error and other parameters. Subspace tracking algorithm (SP) is selected to rebuild compression sampling signal. Signal reconstruction process based on the SP algorithm is given.In the experiment section, the sound and vibration signal are selected to complete the verification of the system. The error with the traditional Nyquist sampling method and performance advantages and disadvantages of the acquisition system with different random sequence generating are compared and analyzed. At last, the error sources are given. Experimental results show that the system is able to complete the analog signal compressive sampling, and recover the original signal in the upper machine accurately. The sampling frequency is lower than the Nyquist theorem, when the sparsity of signal is estimated accurately, the residual error of the reconstruction is 0.0560.