Research On Construction And Optimization Of Measurement Matrix Based On Compressive Sensing Of Wireless Sensor Networks

The core idea of compressed sensing theory is to compress data at the time of signal acquisition,and reconstruct the original signal with the least observation value.The performance of measurement matrix is the key factor.Wireless sensor network(WSN)is a self-organized distributed sensor network in wireless communication.The limited energy and computing power of the node is the biggest problem that affects the network lifetime.Therefore,this paper discusses the research status of compressed sensing in wireless sensor networks.At the same time,based on the research of existing measurement matrix,this paper makes an exploratory and innovative research.On the basis of this,we deeply study and analyze the construction methods and optimization methods of common measurement matrices,and summarize their advantages and disadvantages,and provides a solid theoretical basis for the research of this paper.In this paper,we first introduce the basic theory and concept of compressed sensing,and analyze the two guiding conditions that should be satisfied by the measurement matrix,that is RIP(Principle Isometry Restricted)criteria and relevance discriminant theory.In the aspect of measurement matrix construction,in order to solve the problem of redundancy and the transmission energy consumption in the process of data acquisition in WSNs,this paper deeply analyzes the linear measurement process of the signal,studies the construction method and performance of the mainstream deterministic measurement matrix,and proposes a method of constructing measurement matrix based on diagonal linear representation.This method combines the diagonal matrix and orthogonal basis of linear representation principle,constructed by linear structured method,and the process of construction is simple,fast,highly sparse,no redundancy,which is suitable for the nodes with limited hardware resources.The simulation results show that this method has higher success rate under the premise of precision signal reconstruction,sensor nodes can be observed by compressing the measurement data less,thus greatly reducing network traffic.In the aspect of measurement matrix optimization,this paper aims at the problem that the traditional measurement matrix method cannot meet the requirement of data acquisition in WSNs.In this paper,we study the characteristics of the wavelet transform basis coefficient distribution,combine the convergence characteristics of the gradient reduction algorithm,and optimize from two aspects of signal acquisition and measurement matrix.Then,a gradient descent hybrid algorithm based on discrete wavelets transform is proposed in this paper.This method can reduce the complexity of the data space,improve the convergence rate of the algorithm,and enhance the non-correlation between the measurement matrix and sparse matrix.The experimental comparison and analysis indicate that the method has fast convergence speed,the data reconstruction success rate is significantly higher than that of traditional measurement methods,reduces the measurement matrix design and implementation difficulty,improves the ability to remove the noise,and it is suitable for application in the low sampling rate network environment.