2.?Related WorkIn-network aggregation query processing methods used in sensor networks such as TAG [13] only send the aggregated results inside the sensor network so as to reduce the number of messages. It enables one to increase sensor network lifetime by reducing energy consumption. Some special aggregation quality control queries such as SUM, MIN and MAX, are more effective in saving energy since they only aggregate a single aggregation value instead of all possible data. Also, DCSs process queries in a sensor network, and only send query results to a server. However, skyline queries exclude values only when data are dominated by other data, so it is difficult to find complete query results without inspecting all the data. Therefore, to process skyline queries in sensor network, it is important to establish criteria to exclude unnecessary data from the results.
Several skyline query processing methods such as [8�C11] have been proposed. Most of them focus on designing filters to exclude as much unnecessary data as possible. Huang et al. [8] dealt with a constrained skyline query problem on MANETs by devising a single point filter-based evaluation algorithm that is easily extended to sensor networks. Xin et al. [10] proposed two filter-based algorithms. One is the single point filter-based algorithm TF and another is the grid filter-based algorithm GI. The TF algorithm chooses the point that dominates the maximum number of points as the filter, assuming that the data distribution density is given beforehand, while the GI algorithm exploits the grid partition of data space and generates a grid filter.
Liang et al. [9] proposed a new filter-based algorithm which consists of multiple rather than single points as the filter, whereby each sensor sends part of its Brefeldin_A local skyline points chosen by a greedy algorithm to its parent and the root broadcasts the received points as the global certificate obtained through in-network aggregation. The points that cannot pass through the certificate will be filtered out from transmission. Xin et al. [10] proposed a density function based skyline query processing algorithm. It assumes that the density function of data is known beforehand. However, in real applications it is hard to find out the density function beforehand. Chen et al. [11] proposed two algorithms for evaluating skyline queries are devised, which find the skyline points progressively. It partitions the dataset this research into disjoint subsets, followed by returning the skyline points through examining each subsequent subset progressively, using some found skyline points so far as a filter to filter out those unlikely skyline points in the currently processing subset from transmission.3.?Proposed Skyline Query Processing Method Based on DCS3.1.