Stream Data Query Based On Feedback Mechanism

Stream data query is a very important field of data stream processing. Stream data must be processed promptly to keep up with the fast data arriving rate. The data stream input rate sharp increase can make the query system to be overload, which affects the query response speed, the query quality seriously. The load shedding of the stream data query system becomes a research hot spot. The result stream data produced by query processing often has very important reference value, so reuse the result stream data in data stream query is important. How to fully make use of the query result data stream also becomes an important research problem.Now query model of data stream can't deal with the overload of system efficiently, and can't make full use of result stream data either. In view of this, we present a dada stream query system based on feedback mechanism model. We have made a deeply study of the model and load shedding of data stream query system. Our work is mainly reflected in the following aspects:Firstly, we construct a data stream query system model based on feedback mechanism. We can evaluate the data stream input, and shedding overload more efficiently through the characteristic parameters came from classifying the result data stream feedback tree. With the parameter produced by feedback tree, the query over the data stream can also be performed more accurately.Secondly, a completely load shedding router map (LSRM) construction method and an adaptive LSRM construction method are presented after analyzed the Aurora's LSRM construction.Thirdly, a sliding window maintaining scheme with multi-time granularity is designed, and with that we can query steam data more accurately.Lastly, feedback tree is used to produce the parameter of the model, the growing and prune processes of the feedback tree are designed.