Research On Multidimensional Top-k Query Processing In IoT Sensing Networks

Along with a increasing number of smart things being deployed in the real world,IoT Sensing Networks have being paid wide attention in many fields,especially in physical sensing and monitoring.These smart things actually act as sensor nodes to monitor and collect data of various objects.The multi-dimensional Top-k query aims to retrieve k data records with the highest score in sensing nodes,which is an essential ingredient.However,since the influence of hardware and environmental factors,some nodes powered by batteries,and the energy is limited and difficult to supplement.Besides,the data transmission is the main aspect of energy consumption.Therefore,in the query proessing of the IoT sensing network,how to reduce data transmission to reduce network energy consumption is one of the most important issues.Aiming at the problem of high energy consumption in the multi-dimensional Top-k query processing in IoT sensing networks,this thesis considers the query of the whole network region and the regional query of the network region.We propose the schemes of cache based multidimensional Top-k query processing in IoT sensing networks and the regional continuous multidimensional Top-k query processing in IoT sensing networks respectively,which can reduce energy consumption caused by redundant data transmission of sensing nodes under the premise of ensuring the correct results of the query.The main work of this thesis is as follows:(1)A continuous multi-dimensional Top-k processing scheme based on cache is proposed.At present,the query processing schemes that use the dominance relationship between multidimensional data points and the threshold for data filtering in existing research have the problem of high energy consumption.In response to this problem,this thesis first divides the network region into grid blocks,and constructs an index tree by merging the grid blocks to realize the energy efficiency of data transmission.Then,the data of the nodes in the popular grid block is cached to answer the query,and the cached data is used to calculate the threshold of the query to filter the data of the nodes in other grid blocks.In addition,the grid index is constructed when the data is gathered up to the head node along the index tree,and the dominance relationship is used to reduce the upload of false positive data.Finally,the effectiveness of reducing energy consumption is verified through comparative experiments.(2)A regional continuous multi-dimensional Top-k query processing scheme is proposed.Aiming at the regional continuous multi-dimensional Top-k query problem that has not been considered by existing works,this thesis first uses the constructed index tree to send the query to the relevant nodes.Then,the Skyline of nodes in the query region is collected,and a threshold estimation algorithm based on histogram is proposed.The estimated threshold is close to the true threshold of the query.Compared with the existing threshold selection methods,the estimated threshold is closer to the true threshold of the query,and the threshold can filter most invalid data records.In addition,using the nature of Skline,the nodes that do not need to receive the threshold are screened out.Finally,the energy consumption of the scheme under different parameters is compared,and the effectiveness of the proposed scheme is verified.(3)A continuous multi-dimensional Top-k query processing prototype system is designed and implemented.In response to the requirements of continuous multi-dimensional Top-k queries in the whole network region and regional continuous multi-dimensional Top-k queries,a sensing data processing module,a preference setting module,a query processing module and a result display module are designed to meet the needs of experiments for the proposed scheme according to the design concept of the modular program.The prototype system is programmed with Java language and can implement the proposed schemes.Finally,the function of the prototype system is displayed.