Research On Analog To Information Conversion Technology Based On Random Demodulation

Conventional signal processing patter is Nyquist Sampling followed by datacompression so that the redundant part of sampled data can be removed. Obviously,this kind of sampling-compressing patter could result in much waste of resources.Analog to Information Conversion is a promising solution, which is based on somenew signal processing theories and aimed at convering the analog signal to usefulinformation rather than redundant data. Compressed Sensing(CS) is a such kind ofimportant theory which exploits the sparsity of signals to enable compressing andsampling signals simultaneously without any loss of information, at rates much lowerthan the Nyquist rate. But physical implementation of CS is a hard problem.Therefore, this master’s thesis focuses on exploiting Random Demodulationtechnology to accomplish the physical implementation of CS and realize theconversion from analog signal to compressed information. In this paper, the mainresearch contents and achievements are as follows:1、Research on the principle of Analog to Information Conversion technologybased on Random Demodulation. Both intuitive interpretation and mathematicaldemonstration of Compressed Sensing theory and Random Demodulation are givento indicate that Random Demodulation is the extention of Compressed Sensingtheory from descrete-time domain to continuous-time domain. Then, MATLABsimulation experiments are exploited to prove the feasibility of RandomDemodulation technology, as well as explore the impact of many factors, such assampling rate, sampling phase, parameters of filter, the period of m sequence, etc.2、 Design a Random Demodulation(RD) system according to theoreticalresearch and simulation results of Random Demodulation technology. This systemconsists of hardware part and software part. Hardware part of RD system integratessignal conditioning circuit board and various PXIe devices which are interconnectedvia advanced PXIe bus. Hardware part is used to complete following tasks, likesignal production, mixing, filtering, amplifying, sampling, data transmission andstorage. Software part is a a set of upper-monitor software developed with LabVIEWto offer a flexible control of these hardware modules and perform varied operations,such as displaying signal, running algorithms and producing report forms.3、 Construct the sensing matrix of Random Demodulation system. Threeconstruction methods of the sensing matrix are proposed. The first method is basedon theoretical calculation; the second one is based on stimulation of step-by-stepsinusoids and cosines; the third one is fast sensing matrix construction method based on stimulation of m sequence and FFT.4、Conduct experiments by using of this Random Demodulation(RD) system.Cogent experiments demonstrate that our random demodulation system can achievesub-Nyquist sampling for analog signal with max frequency less than50kHz at rateof4kS/s. In other words, the compression ration is as much as4%and the Signal toNoise Ratio(SNR) is usually more than15dB.