Wideband Receiver Algorithm Based On Compressed Sampling

The rapid development of wireless communication technology has caused a shortage of spectrum resources.How to optimize the allocation of limited spectrum resources is an urgent problem to be settled.Cognitive radio technology is considered as the key technology to solve this problem.Through detecting dynamic changes of the spectrum,it locates the spectrum holes,which can be used to sense the spectrum and establish radio environment map.In this way,those unauthorized users are able to have access to the frequency band more effectively,such that the spectrum efficiency is improved.However,since the sampling rate of existing high-precision ADC chips can't achieve the Nyquist sampling of wideband signals,it is difficult for cognitive radio technology to be applied to wideband spectrum sensing.In order to sovle this problem,a compressed sampling framework with multiple wideband receivers is introduced in this dissertation.It uses compressed samples to reconstruct the power spectrum for spectrum sensing and parameter estimation along with the channel gain,which can be used for source location.The main research contents of this thesis are as follows:1.Since the receiver is difficult to implement Nyquist sampling in the wideband spectrum sensing,we first study the compressed sensing theory,including the sparse representation of a signal,the design of the measurement matrix and the reconstruction algorithm.Furthermore,we also study the compressed sampling frameworks that work on analog domain.2.Secondly,with serious channel attenuation and noise interference,the spectrum reconstruction method and the single receiver cannot be effectively spectrum sensing.Therefore,we use the compressed sampling framework to establish a multi-receiver collaboration sensing model and reconstruct the power spectrum.Simulation results demonstrate the performance advantages of the proposed algorithm.3.The mixed power spectrum recontructed by the above cooperative perception model contains channel information,which results in limited spectrum sensing performance.To solve this problem,we use the Non-negative Matrix Factorization(NMF)to separate the mixed power spectrum into two parts: the power spectrum of the source signal,and the channel gain.Simulation results indicate that the separated source power spectrum can realize spectrum sensing even if SNR is as low as-10 dB.4.Finally,we estimate the carrier frequency and bandwidth as well as source location in order to build a radio environment map.On one hand,we estimate the carrier frequency and bandwidth based on the power spectrum obtained by NMF.On the other hand,we use the channel loss coefficient obtained by NMF to estimate the source coordinates.For the amplitude ambiguity of NMF,we use the ratio of distance and path loss to subtly offset the influence on positioning.Simulation results show that the absolute error of source location can can be as small as 1m provided that SNR is 0dB and the range of field is 50m×50 m.