The Research And Design Of Analog-to-information Convertor Based On Compressed Sensing

Compressed sensing(CS),a new theory which can avoid the limit of traditional Nyquist sampling theory,points out that sparse signal could be sampled even if the sampling rate doesn't reach the conventional Nyquist frequency through reconstruction algorithm.With the increasing popularity of ultra bandwidth and high frequency technology,CS theory provides a new practical idea.Analog-to-information convertor(AIC)is an sampling and compression device which is based on the CS theory.So far the research of AIC stays at the theoretical and mathematical level.The lack of actual circuit research has prevented the application of AIC.In this paper,a multiphase random sampling FFT AIC is adopted to be designed and simulated.The function of AIC will be tested and the parameters will be gotten through the simulation.The main research and innovations of this paper are as follows:1)From the perspective of the design of circuit hardware,a multiphase random sampling FFT AIC which is based on random demodulation is proposed.As the AIC is consist of multiple branch ADC circuits,the performance of AIC is related to the number of branch circuits,which makes the AIC circuit could be easily expanded and improved.The whole system is apt to be carried out since the modular design of sub-circuit.The AIC function has been proved by the mathematics.The measurement matrix and the sensing matrix are derived by the mathematics and are going to be used in the simulation latter.2)The ideal model of multiphase random sampling FFT AIC is set up on MATLAB.The function of AIC is verified through a suitable algorithm.Through multiple experiments and the comparison with traditional random demodulation,the parameters of ideal AIC is obtained as the contrasting basis for actual AIC designed circuit.3)The design and simulation of multiphase random sampling FFT AIC is completed by craft UMC65 nm and software Cadence IC616 through transistor to layout.The simulation shows that the designed AIC is functional and the parameters are offered.In this paper,a ten branch circuits AIC is designed.It can work on the frequency of 2GHz.The signal length is 10.The AIC can manage signal with mostly 4 sparseness and least 32 MHz resolution and implement compressed sampling as long as reconstruction without distortion.In this paper,a new AIC model is carried out and designed as hardware circuit.Compare to the traditional random demodulation AIC,the multiphase random sampling FFT AIC has a more simple structure,higher resolution,higher stability and a better expansion.The practical value of AIC is proved by the simulation.The comparison between ideal AIC and actual designed AIC shows that the problems of the actual designed AIC and point out the improvement it could make.The research of the AIC design has provided parameter-basis for the implement of actual circuit design based on CS theory.