Research On Distributed Probabilistic Skyline

Recently, most work about skyline query focus on optimizing computing cost, while few papers discuss the problem of optimizing the communication cost in skyline query. Given a distributed system based on the architecture of client-server, clients make records of data with devices like sensor devices, and the server continuously maintains skyline of dynamic records transmitted by clients. Because frequent data transmissions consume a mass of energy and bandwidth, too much communication between clients and the server would waste enormous resources. Moreover, it constrains the efficiency of continuous skyline query. The server always has abundant resources for massive computation, so communication cost become the main factor that influences the distributed skyline algorithm.Besides, data uncertainty widely spreads in most real-world scenarios, which makes skyline query even more complex and difficult. In this paper, we study continuous distributed probabilistic skyline query over uncertain data. Firstly, we define a kind of special probabilistic threshold skyline over uncertain data. Secondly, we propose a filter approach where the sever constructs filters for each record and clients only transmit updates that exceed the corresponding filter. In this way, unnecessary data transmission is avoided. Thirdly, experiments on synthetic data set at the end show that our approach can effectively reduce communication cost in the process of distributed probabilistic skyline maintaining.