Research On Uncertain Data Stream Database System

Recently, stream database which is a new and popular research field of database theory has been considered as the best method to handle unbound, rapid, time-varying sensor data. However, most of data stream research works focused on precise data. Sensor data streams, as the main application field, tend to be uncertain due to the all kinds of reasons. Therefore, the work of paper focuses on the sensor data stream with uncertain information, i.e. uncertain data stream. It aims to improve the useability of sensor uncertain data stream and gain more information from it.The main contributions of this paper are as follows. First, several uncertain data stream representative models have been proposed based on the anlysis of the sensor data characteristics and uncertain relations. Second, the method using probability density function to record moving object is proposed, because stream including values continuously changed cannot be handled by traditional stream database. That is, sliding window is used to get the stream including probability density function and probability and the function itself can be queried. Third, preprocess of noisy data in uncertain data stream has also been described. At the same time, the confidence of data stream tuples(or attributes) has been re-estimated accordingly. The last not the least, we introduce uncertain stream contiuous query language and the preliminary uncertain data stream database system.