Research On Technology Of Blind Separation Based On Compressen Sensing

Compressed sensing is a theory between mathematics and Information Science created in recent more than 10 years,which breaks through the sampling rate limit of traditional sampling mode,and brings milestone development for signal sampling technology.According to the sparsity characteristics of the original data,it keep the original structure of the signal through the linear projection method,and sampls data at a speed much lower than the Nyquist sampling frequency,finally we can reconstruct the original signal through simplifying the process of solving numerical optimization problem.In summary,we can reduce the costs of storage,processing and transmission of the massive data.Blind source signal separation(BSS),as a new hot research topic,has a reliable theoretical basis and has been developed in many practical applications.Blind source separation is the process of estimating or separating each source signal based on the observation signal from the transmission system without knowing the source and the aliasing system.When the number of observation signals is less than the number of source signals,which is underdetermined,the problem of blind source separation meets the requirements of compressed sensing theory model.Therefore,the blind source separation technology based on compressed sensing has a good application prospect.Aiming at the shortcomings of traditional compressed sensing algorithm,and combines the advantages of several traditional compressed sensing reconstruction algorithms,this paper proposes an improved algorithm.Meanwhile,puts forward a kind of a blind source separation algorithm based on compressed sensing according to the similarity between compressive sensing algorithm and blind separation algorithm.The main contents are as follows:This paper analyzes existing problems of the traditional compressed sensing recovery algorithm,puts forward an Adaptive Step-size Two-step Iterative Shrinkage algorithm.This algorithm adopts different iterative step-size in the iteration step,and the accuracy and speed of the iterative are appropriate by proper control.If the difference between the results of the two iterations is large,the algorithm increases the step-size and repeats the current step,then resets the step-size to do the next step.In addition,the iterative speed of the whole algorithm is accelerated by the two-step iteration.Through the analysis of the similarity between signal blind source separation model and the theory of compressed sensing model,then the signals are processed to remove minor differences between the two model.Then the author transforms the signal blind source separation problem into the reconstruction problem of the signal that can be solved by the compression sensing theory.Thus,the K clustering algorithm is used to estimate the mixed matrix;the original signal is reconstructed based on the estimated hybrid matrix combined with the compressed sensing theory,thus the separation of the signal is realized.