Research Of WSN Spatio-temporal Correlation Data Compression Based On Compressed Sensing

As an important part of the Internet of Things,wireless sensor networks are used in many fields.The amount of data generated is increasing day by day,and the research on this type of sensing data compression has always been one of the most critical technologies in Io T applications.Due to the limitation of storage space,power consumption,and computing capacity of sensing nodes,how to meet application requirements under low energy consumption conditions to achieve data compression has become one of the urgent problems to be solved.Combining the characteristics of sensing nodes and base stations,this article improves the process of compressed sensing theory to process a type of sensing data with strong temporal and spatial correlation.The experimental results show that the improvement of this article can achieve better compression effects.The main contents of the are as follows:First of all,in the wireless sensor network,in view of the differences in monitoring data characteristics,the limited energy consumption of nodes,and the low efficiency of existing data compression methods,a data compression model based on dictionary learning and compressed sensing is proposed.The model implements sparse transformation through the base station,proposes an improved K-Singular Value Decomposition(K-SVD)dictionary learning algorithm for the initial dictionary,and uses the adaptive characteristics of the dictionary learning algorithm to train the sparse transformation base to meet the characteristics of large differences in monitoring data and reduce Compression complexity and node energy consumption.Compared with the improved gray model data compression algorithm of this research group and the commonly used discrete cosine transform based compressed sensing algorithm,the results show that the model in this paper has a significant improvement in data compression rate and recovery accuracy.Secondly,aiming at the problems of high computational complexity,inconvenient storage and hard hardware implementation caused by random measurement matrix in the compression process of compressed sensing,this paper designs a sparse measurement matrix based on regular low density parity check matrix for data compression of sensing nodes.The low density parity check matrix has less non-zero elements,which is easy to compress and save;the regular low density parity check matrix satisfies the statistical restricted isometric property,which ensures the accuracy of data recovery;in this way,when the measurement matrix and sensing data are multiplied for compression,the amount of calculation is greatly reduced.Theoretical analysis and comparative experimental results show that this method can reduce the amount of data transmission,improve the compression rate,save energy consumption and prolong the network life cycle.This paper reproduces all comparison algorithms on the public data set provided by Intel Berkeley Lab.The experimental results show that the data compression model based on dictionary learning and compressed sensing and the regular low density parity measurement matrix designed in this paper can improve the data compression rate,data recovery accuracy,and meet the practical application needs of strong spatial-temporal correlation.