Research On Mixing Sequence Optimization And Blind Calibration Method For Modulated Wideband Converter

As users' demands for broadband wireless communications continue to increase,limited radio spectrum resources become more and more crowded.Spectrum sensing technology can detect the occupancy of each frequency band,and improve spectrum utilization through cognitive reuse of idle frequency bands.In the spectrum sensing of wideband signal,spectrum of the signal is always sparse at different moments,and the traditional Nyquist sampling method will have redundancy due to its high sampling rate and large data amount.The emergence of compressed sensing theory makes subsampling spectrum sensing possible.Modulated Wideband Converter(MWC)based on this theory has got lots of attention because of its simple structure and good performance.However,in the actual realization of MWC,the existence of non-ideal factors in hardware makes the actual model of the system different from the theoretical model.If the actual MWC system is not calibrated,the reconstruction of signal's spectrum support set may fail.Based on MWC,this paper studies the blind calibration method of non-ideal MWC system under the condition of amplitude and phase error in each channel,which only needs to obtain the actual samples of the system to complete system calibration and signal reconstruction at the same time.The mixing sequence of MWC is then optimized to improve the effect of proposed blind calibration algorithm at low signal-to-noise ratio.The main work is as follows.1.The basic principle of MWC and non-ideal factors of this system are analyzed.This paper studies the working principle of MWC for compressive sampling of sparse multiband signals,and establishes a mathematical model of ideal MWC.The sources of amplitude and phase errors existing in each channel of actual MWC are explored,and the effect of these errors on reconstruction ability is illustrated by simulation experiments.2.In view of independent amplitude and phase errors between channels existing in non-ideal MWC system,the blind calibration algorithm is designed based on the bilinear underdetermined equation.This paper establishes the actual mathematical model of the non-ideal MWC system.The calibration problem is transformed into a bilinear inverse problem,and a blind calibration algorithm based on gradient descent and symmetric projection is designed accordingly.Simulation experiments verify the effectiveness of our algorithm,reveal its robustness under large errors,and explore the influence of system parameters on the effect of blind calibration.3.In view of performance degradation of the designed blind calibration algorithm under low signal-to-noise ratio,the MWC system is optimized based on sparse mixing sequence.In this paper,a sparse mixing sequence is designed to improve the system's own anti-noise ability.The paper analyzes and verifies the advantages of the improved ideal system in the presence of noise in reconstruction performance,and reveals that when the signal-to-noise ratio of input signal is low and the system has unknown amplitude and phase errors,the improved MWC system has a better blind calibration effect than before the improvement.4.The performance of the blind calibration algorithm is verified on the hardware platform.This paper shows the comparison of support set recovery when different system parameters are set for MWC or different mixing sequences are used in MWC,verifies that the results of simulation experiments are still valid in MWC hardware system with channel errors.The results of hardware experiments further illustrate the rationality and effectiveness of the proposed calibration method and mixing sequence optimization method.