Font Size: a A A

Research On Key Techniques Of Query Processing Over Wireless Sensor Networks

Posted on:2013-01-15Degree:DoctorType:Dissertation
Country:ChinaCandidate:Y L LiuFull Text:PDF
GTID:1268330422980480Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
WSNs (Wireless Sensor networks) have a wide range of applications in military defense,medical services, traffic control and etc. WSN is a kind of distributed embedded system and thesensor nodes have very limited computing power, storage capacity, communication and energy.Therefore, the development of sensor network applications is extremely difficult. Unlike theother networks, wireless sensor networks are data-centric. The main purpose of using WSNs bythe users is to collect the events or data generated by the network. WSN database managementsystem provides users with a simple query interface which shields the complexity of queryprocessing. It greatly simplifies the development of sensor network applications and has becomean important middleware for WSN. Query processing is the core technology of WSN databasemanagement system.(1) Spatial window query is used to obtain the sensory data of all the nodes in the user’s areaof interest. The energy consumption of spatial window query processing algorithms depends onthree parameters: number of query message forwarding, number of sensory data returned to Sink,the average number of sensory data packets sent to the Sink by a node. The existing algorithmsonly optimize one of the parameters, while ignoring the other two parameters, resulting in a largeamout of energy. We propose a spatial window query processing algorithm which makes anholistic optimization of the above three parameters. It schedules some but not all nodes in thequery region to distribute query messages to ensure that all nodes within the query region receivea query message, which reduces the number of query message forwarding. It selects somerepresentative nodes in the query region to return their sensory data back to sink. The sensorydata of the other nodes in the query region is estimated by siink under the premise of ensuringthe accuracy of the query results. In addition, the sensory data in the query region is returned toSink directly by geographic routing protocol, so that the number of forwarding while sending asensory data o the Sink is reduced. The theory and simulation results show that our algorithm issuperior to existing algorithms in terms of energy consumption.(2) In order to reduce the energy consumption of irregular spatio-temporal query processingand improve the query success rate, a tree-based algorithm is proposed to processspatio-temporal data collection queries with irregular query regions. It organizes sensor nodes inquery regions as a tree. The nodes in the tree send local data to their parents until reaching theroot of the tree. An itinerary-based algorithm to process spatio-temporal data aggregation querieswith irregular query regions is also proposed here. It collects the data of nodes in the queryregion and aggregates them along an itinerary to generate the final query result. Both of themdivide the complex and irregular query region into some simple convex polygons in order to reduce the computational complexity of determining whether the nodes are in the query regionand ensure that only the nodes in query regions send the sensed data, thus reducing the energyconsumption. The experimental results show that the proposed algorithms outperform theexistion spatio-temporal query processing algorithms for irregular region query.(3) Sketch can estimate the equal join size of data stream with high precision and histogramcan calculate the distribution of data stream accurately. An efficient complex data aggregationquery processing algorithm for data stream is proposed based on sketch and histogramtechniques. The algorithm can provide approximate answers to a certain kind of complexaggregate queries. The theory and experimental results show that the algorithm has highprecision and small space complexity.(4) Wireless sensor networks are mainly used for monitoring the status of an area or theevents happened. When an event occurs in the monitored area, users can obtain the results of knearest neighbors query to analyze the causes and forecast the development trend of the event.The current state-of-the-art k nearest neighbor query processing algorithms have fairy highenergy consumption and low query success rate. A robust data collection protocol called ROC isproposed in this paper. It divides the query region into several ring sectors. Each ring sector has acluster head node which collects the sensory data in it, calculates the partial query result andsends it to the cluster head node in the next ring sector. It takes advantage of geographic routingprotocol to bypass the region which has no nodes, which ensures that query processing is notinterrupted. We also propose a class of k nearest neighbor query processing algorithms calledROC-KNN based on ROC protocol. They only access the nodes which may contain query resultsin order to reduce the energy consumption. When a cluster head node fails during queryprocessing, any node within the sector which the cluster head node locates in can replace it tocontinue query processing, which improves the query success rate. The experimental resultsshow that ROC-KNN outperforms the existing itinerary-based algorithms in terms of energyconsumption and query success rate.
Keywords/Search Tags:data management, wireless sensor network, query processing, data aggregation query, k nearest neighbor query, spatial window query, energy efficiency
PDF Full Text Request
Related items