Research On Data De-redundancy Algorithm In Wireless Sensor Networks

Wireless sensor networks(WSNs)have been closely related to people's life and widely used in various fields of society.However,due to the defects of limited battery,storage capacity and computing power,the sensing data shows an uncontrollable trend of crazy growth,resulting in a series of problems such as intensified energy consumption,insufficient storage space and increased computing complexity.In order to process mass data effectively in WSNs,data de-redundancy technology provides the possibility to reduce the transmission of sensing data and save energy.In this thesis,the data de-redundancy algorithms in WSNs are analyzed and summarized in detail.According to the different algorithms used in the de-redundancy process,the de-redundancy schemes are classified into three categories: schemes based on statistics,schemes based on data compression and schemes based on artificial intelligence.The thesis also makes an in-depth comparative study on representative de-redundancy methods,summarizes the advantages and disadvantages of schemes in WSNs and the existing challenges,and explores the future research hotspots and trends.Considering that the time correlation data redundancy in WSNs,large range of data and the local characteristics of local most value error lead to lost greatly,and the failure problem of data smoothly when the similarity threshold,an adaptive step length of data de-redundancy method(TCDA)is proposed based on the maximum time threshold.In order to ensure the de-redundancy rate,TCDA sets the maximum time threshold to prevent the failure of data similarity threshold and ensure the timeliness of data.Meanwhile,TCDA adopts the adaptive step length mechanism to reduce the computational complexity and energy.From the perspective of de-redundancy of spatial correlated data,a multi-phase hierarchical clustering similarity de-redundancy method(MHCSD)is proposed based on spatial correlation,to overcome the shortage of judging node redundancy only according to the distances between nodes.In the first stage,the sink node adopts the improved k-means algorithm,and the similarity distances between nodes are weighted by Euclidean distance and Pearson distance to ensure the rationality of spatial location similarity of redundant nodes.Then,sensor nodes are classified into clusters according to spatial coordinates to form spatial node similarity clusters.In the second stage,in the similar cluster of spatial nodes,the cluster head uses the Gaussian mixture clustering algorithm to analyze the similarity of data nodes within the cluster based on the sensing data,and form the similar cluster of data nodes within the cluster,thus the proposed scheme ensures the similarity of redundant nodes.In the third stage,the sensing data generated by similar clusters of data nodes are further removed through random weighting.The performance of Dat,TCDA,MHCSD and HMDA is verified and analyzed on the Python3.6 platform.The experiment fully considers the performance of the data de-redundancy method in different scenarios in terms of de-redundancy rate,network energy consumption,and data error.With the dataset of 2 million temperature data of Intel Berkeley laboratory,experimental results show that TCDA saves 3% transmission energy consumption and 50% calculation energy consumption,compared with those of data transfer protocol method.HMDA further removes 70% of redundant data and saves 1.25% of transmission power compared to those of TCDA.HMDA saves 4.25% of transmission power compared to those of Dat.