Research On Wireless Sensor Network Query Optimization Strategy

The steady development in relevant technologies of wireless sensor networks has made WSN the most fundamental technology of information acquisition in Internet of' Things.In WSN which is a typical data-centric application,the most important task is to monitor physical and environmental conditions.There are two stages in data acquisition which starts from user's query,query flooding and data return.Improving energy efficiency has become the hotspot of WSNs research due to the constraints of sensor nodes on computing,storage,power resource.To achieve the energy saving requirement two solutions aiming at improving energy and query efficiency are presented.The research contents are as follows:The query processing details are inseparable from the corresponding storage strategy.Most existing works on sensory data storage are focusing on storage strategies which are selecting the optimal storage location.It is a trade-off between the energy consumption of query flooding and data return.The approach query processing mechanism based on RCFile presented in this paper is focusing on the storage structure of sensory data and is different from the existing researches.RCFile is a fast and space-efficient data placement structure in MapReduce-based warehouse systems and it combines the advantages of both row-store and column-store.RCFile applies the concept of "first horizontally partition,then vertically partition".First,as row-store,RCFile guarantees that data in the same row are located in the same HDFS block,thus it has low cost of tuple reconstruction.Second,as column-store,RCFile can exploit a column-wise data compression and skip unnecessary column reads.The huge energy consumption difference on computing and communications in WSNs means that the appropriate compression processing for reducing data size is an useful solution for transporting energy saving.Meanwhile,as a typical application of multiple-query task,the efficiency of query processing in WSNs is also crucial.It is shown by comparing RCFile to other data placement structures,RCFile is more suitable for sensory data placement relative to other conventional storage structure.High distribution density of sensor nodes in WSNs leads to high data similarity between adjacent nodes.This feature is called the spatial correlation.If every sensor node responses to a query,the sensor nodes will consume much more energy by convey redundant data to the sink node.To reduce the amount of transmitted data,a dual-clustering based query algorithm exploiting spatial correlation of sensory data is presented.The dual-clustering based query algorithm includes three procedures:Data-based Clustering(DC)procedure,Location-based Clustering(LC)procedure and Representative Sensor node Selection(RSS)procedure.In the DC procedure,the WSN is partitioned to several clusters that are composed of the nodes with similar data by using ISODATA algorithm on sensory data.ISODATA is a centroid-based clustering algorithm using distance metric for data similarity.In the LC procedure,the result clusters in the last procedure are further divided into smaller clusters by density-based clustering algorithm DBSCAN on the location information of the nodes.The nodes in the clusters are spatial continuous and data-similar.In the RSS procedure,a representative sensor node is selected in each cluster by using error metric as the cost function.With experiment,the approach presented in this paper outperforms the comparing algorithm on clustering effect,error precision and energy consumption.