Study On Skyline Query Processing Algorithms In Wireless Sensor Networks

With the rapid development of communication techniques, embedded computingtechniques and increasing maturity of sensor techniques, micro-sensors with perceptionability, computing ability and communication ability are developed. Wireless SensorNetworks(WSNs) are composed of large numbers of micro-sensor nodes due to be usedwidely in various fields has received considerable attention. A skyline query can get theoutline of the dataset, and the dominating relationship of tuples in the process of the skylinequery processing is a partial order relationship. Skyline queries mainly are used in the area ofmulti-objective decision analysis, data mining, and other fields. Skyline queries have beenextensive research in a traditional database, but due to the characteristics of such a distributedenvironment for wireless sensor networks, the existing skyline algorithms cannot be applieddirectly to it.Whereas most previous studies have resolved some problems of skyline queries in acentralized environment. For skyline queries in distributed sensor network environment, withthe increase of data dimensionality, there is less likely to the dominating relationship amongdata points and skyline query results extends largely, which is unable to provide interestinginformation. To find the more important and more meaningful skyline points in highdimensional data space, a definition of the k-dominant skyline was introduced, which canreduce the number of skyline tuples by relaxing the definition of the dominance. However,the existing k-dominant skyline algorithms need to be improved on time efficiency, spacecomplexity and progressive output. Based on the above reasons, a new k-dominant skylinequery algorithm is proposed, namely, KSkySensor, which is an effective method for skylinequeries in WSNs. It can collect effectively the more meaningful skyline results from theWSNs. First, sensing data are preprocessed by calculating the domination power of tuples,and then by sorting sensing data by domination power and the sum of all attributes values ofeach tuple; Second, a cluster-based architecture is designed in the KSkySensor to collect allsensor readings; Finally, a pruning method is then proposed to progressively sift out theskyline results from the sensor network. The KSkySensor avoids the need of collecting datafrom all sensors in the network, which is an extremely expensive action, when searching forthe skyline results. Extensive experiments show that the cluster-based KSkySensor methodcan performance skyline queries efficiently, minimizing the cost transmitted in WSNs,reducing the energy consumption of nodes, prolonging the lifetime of the WSNs.