Research On Kalman Filter-Based Query Optimization For Data Stream

Database techniques have been widely used, with the development of network, telegraphy and sensor technology in recent years. On the basis of traditional database , many new kinds of database management system (DBMS) have been developed. Data stream management system (DSMS) is one of the hot research topic, because it can handle a lot of practical problems, such as network attacking data stream, stock data stream and so on. Query of data stream has the characteristic of being instantaneous, continual and long time .Just because the model of data stream plays an important part in many applications, much research has been done on the technology of data stream query .In order to increase the speedy of query, we need to adjust the order of filter , and reduce input data . But in existing approaches, the order of filters is based on past data , without considering present data and future data.In this paper, we introduce kalman filter to predict future data. So according to the predicted data , adaptive filters in center node can be optimized to improve query efficiency. Also, we install kalman filter in the source node, strengthen the intelligence of both nodes, and reduce data to be s transferred. Meantime, we use message to trigger update, so that we can synchronize the data of source node with the center node, to keep the data consistent.