Research On Spatio-Temporal Correlation Based Clustering Algorithm And Data Collection Strategy In WSN

Wireless sensor network is composed of a large number of sensor nodes with perceptual,computational and communication capabilities.It performs environmental data acquisition,data processing and data transmission,presents the environmental information to people.With the increasing demand for various information,wireless sensor networks have been widely used in all fields.Energy conservation is one of the most critical issues because of the finite energy of sensor nodes.The dominant energy consuming part of sensor node is the communication module.Therefore,energy saving can be achieve by designing an efficient routing protocol or reducing the amount of data transmitted.Clustering is an effective energy-saving routing method.With data-aware clustering routing protocol,sensor nodes are organized into different clusters according to data similarity.Combined with the corresponding data collection strategy,it can effectively reduce the amount of data transmitted and communication energy consumption.In this paper,some researches on data-aware clustering algorithms and the corresponding data collection strategies are performed.This paper firstly introduces the background and significance of data-aware clustering and energy efficient data collection strategies in WSN,and then describes the network architecture,features,related technologies and key performance indicators of WSN.Several typical dataaware clustering algorithms and their corresponding data collection frameworks are introduced,whose advantages and shortcomings are discussed,and the core issues in current researches are summarized.To solve the problem that the similar nodes are more demanding which results in fewer redundant nodes being identified in the existing data-aware clustering algorithms,this paper analyzes the distribution characteristics of environmental data,proposes two complementary similarity measurement models and the corresponding data estimation methods.To solve the problem that the existing data-aware clustering algorithm sacrifices the energy efficiency of cluster structure to ensure the data consistency in clusters,a weighting based K-means clustering algorithm(WK-means)is proposed.K-means algorithm is introduced into the clustering process and many factors are taken into account,including similarity,residual energy and other factors that affect cluster structure's energy efficiency.A weighting function is defined for each member node to select cluster to join,and another weighting function is defined for the election ofcluster head.Considering the particularity of our cluster structure,we propose the corresponding data collection framework(SC-EEDC).In this framework,anchor node based data collection strategy is proposed to make full use of the similarity among sensor nodes in a cluster to remove spatial redundancy.Then this paper introduces a new sleep scheduling algorithm to adjust spatial sampling rate adaptively.And the length encoding compression is used to remove the temporal redundancy in SC-EEDC framework.And a cluster structure maintenance strategy is proposed to cope with the death of sensor node and the change of monitored object.Finally,the simulation experiments of the proposed algorithm and framework are carried out on Matlab platform.Here,similarity measurement is seen as a part of WK-means clustering algorithm and WK-means clustering algorithm is regarded as a part of SC-EEDC framework.According to the simulation results,each part of the SC-EEDC framework is analyzed in detail and some parameters are discussed.Moreover,our SC-EEDC framework is compared with other two typical energy efficient data collection framework EEDC and DSCCF in terms of network lifetime,energy consumption,data reduction percentage and accuracy.The simulation results show that the proposed SC-EEDC framework can effectively remove spatio-temporal redundancy,construct energy efficient cluster structure and significantly extend network's lifetime.