Research On Wideband Spectrum Compressed Sensing Based On Noise

With the rapid development of mobile communication technology and the concept of the internet of everything appearing,the demand for spectrum resources has increased dramatically.However,the strategy of allocating spectrum resources in a fixed manner results in a low utilization rate of spectrum resources.So,cognitive radio technology use the strategy of dynamically allocating spectrum resources to improve the utilization rate of spectrum resources.Spectrum detection technology is the premise of dynamic allocation of spectrum resources.If the Nyquist sampling theorem is used to sample the signal in the high frequency band,there will be a problem that the sampling rate is too high.The compressive sensing technology can realize the sub-sampling of the sparse signal and to judge the spectrum occupancy state according to the allocation of spectrum.When considering signal noise,the compressive sampling process in compressed sensing will multiply the signal noise in the observation vector,which is called the phenomenon of noise folding.The phenomenon of noise folding will not only affect the accuracy of the sparsity of signal estimation,but also affect the signal reconstruction stage.Aiming at the above problems,the main work of this thesis is as follows:1.In order to reduce the influence of noise folding on sparsity estimation,In this thesis,by improving the existing sparsity estimation model,the probability distribution function of noise in the observation vector is deduced first,and the noise component in the observation sample is filtered out before the sparsity estimation.Then,the number of observations is calculated according to the estimated sparsity,and the adaptive threshold denoising algorithm is improved.The simulation results show that the proposed scheme can accurately estimate the signal sparsity with signal noise and reduce the number of observations under the premise of signal can be accurately reconstructed.And proposed scheme can reduce the time required for the adaptive threshold denoising algorithm.,thereby reducing the impact of noise folding on compressed sensing.2.In order to solve the problem that when constructing a weighted sampling matrix in the existing selective sampling algorithm,a lot of a priori information is required and it is difficult to obtain in actual situations,a new method for constructing weighted sampling matrix is proposed.Firstly,this method construct weighted sampling matrix according to pre-sampled samples and information in perceptual matrix.Secondly,since there is still a small amount of signal noise in the reconstructed signal,an adaptive selection of reconstruction times algorithm is proposed.This algorithm takes the difference between the 1-norms of the two reconstructed signals as the termination condition to further improve the accuracy of spectrum sensing.The simulation results show that the scheme can realize selective sampling of wideband signals,and can adaptively select the number of reconstructions according to different signal-to-noise ratios.On the premise that the signal can be accurately reconstructed,the time cost and accuracy are balanced.