Research On Skyline Query In Wireless Sensor Networks

Wireless sensor networks consist of large amounts of sensor nodes, which communicate witheach other by radio and construct self-organizing structure. They are deployed for monitoring itssensing areas or objects. Skyline query is used to filtering those data which are inferior to others.Based on its good performance in multi-object decision, data mining and visualization of databases,it attracts more and more researchers’ attention. Applying Skyline query into wireless sensornetworks can not only reduce the cost of application query, but also strengthen the self-cure abilityof the network, and has wide requirements and great significance. In this paper, we make wirelesssensor networks as our research background, and make a deep research of distributed Skylinequery algorithms in it.In distributed environment Skyline query algorithms have achieved some research findings,but they can’t satisfy wireless sensor network’s requirements. It is based on the followingconsiderations:(i) sensor data is horizontally stored, and algorithms based on vertical storagestructure can’t be used,(ii) sensor nodes’ computation and storage ability are constrained, and thedesign of algorithms should not be too complex,(iii) sensor nodes’ power is constrained, thecommunication cost of the nodes should be decreased as much as possible,(vi) sensor nodes’power can not be regenerated or renewed, power balance over the network should be achieved.Based on this, making full consideration of the properties of the network, we propose twodistributed Skyline query algorithms refer to wireless sensor networks.Firstly, fully mining the shortage of the construction of filters, we propose a light weightiteration filtering algorithm. By using sensor data’s dominate ability’s progressive principle, it canreturn query result which has the largest probability to satisfy users’ desire. Adopting mappingstrategy to construct the filter, it can improve the filtering effects, at the same time it can bringdown computation cost of the nodes, finally reduce communication cost of filtering stage. Basedon this, we further propose three optimizing strategy which can improve the algorithm’sperformance obviously. After that, considering the data-centric characteristic of wireless sensornetworks, we integrate distributed storage with query operation, and propose a space dividingstorage based distributed Skyline query algorithm, which can provide the most adaptive storagestructure for the query algorithm. First, it divides the sensor space into subspaces. Every cluster isin charge of a few subspaces which can not be dominated by each other. According to its attributes,every tuple can determine its storage node. The filtering is divided into two stages, node filteringand sensor tuple filtering, the former aims at reducing the number of nodes that need to be accessed,the latter aims at reducing the traffic of data.In the end, we use real and synthetic data to verify the algorithms’ performance. Large amounts of comparison experiments prove that the two algorithms can greatly reduce the numberof nodes that need to be visited and the network’s communication cost, progressively return thequery results, has great progressive property and scalability.