Research On Technologies Of Radio Frequency Sampling System Based On Compressed Sensing

As the bridge between the analog signal and digital signal system, theperformance of the data sampling module has become a key factor restricting thedevelopment of digital signal processing systems. In many practical applications,the traditional Nyquist sampling system has been unable to meet the challenges ofhigh speed sampling. By diverting the pressure of data sampling system to thesubsequent digital signal processing, the compressed sensing theory provespossibility to sample the signal directly in a lower speed than the traditional Nyquistsampling system. So the development of the compressed sensing theory has greatsignificance in the field of digital signal processing.The traditional data compressing mathematical model has been introduced inthis paper. The advantages and disadvantages of the traditional data compression arealso detailed. The theoretical framework and the mathematical model of compressedsensing are elaborated. Then the three key problems, including the sparserepresentation of the signal, the design of the observation matrix to meet the RIPguidelines and the selection of signal reconstruction algorithm, are analysized inthis paper. The traditional Nyquist sampling system, including the parallelalternating sampling system and the channelized sampling system are alsoelaborated in this paper. The realization principle of AIC system solution based oncompressed sensing and the theoretical derivation of the observation matrix whichhas a equivalent form in the system is also detailed in this paper at last.Based on compressed sensing theory, the design of the RF signal samplingsystem is built, which including low-noise amplifier, balanced network, mixer, theanalog front-end of low-pass filter and the design of the digital module of the ADsampling and clock network. The simulation of OMP algorithm is accomplished.The algorithm is improved according to the characteristics of the processed signal.The amount of computation is decreased significantly without reducing the accuracyof the reconstructed signal, which is easier for FPGA implementation of thealgorithm. The design of the overall block diagram of the algorithm’s FPGAimplementation is accomplished. The design principle and the data simulation of thevarious modules including the module of maximum correlation, the module of the reconstruction dictionary, the projection module, the QR decomposition module andlinear equation solution module is detailed. The design of the global timing controlmodule is accomplished. Under the control of the global timing module, the FPGAimplementation of the improved OMP algorithm is also accomplished. Theexperimental results are also detailed in this paper.