Evaluation of a Framework for Dynamic Source Selection in Stream Processing

The volume of stream data delivered from different information sources is increasing. There are a variety of demands to utilize such stream data for applications. Stream processing systems can continuously process stream data according to user requests. In conventional frameworks,the user must explicitly specify information sources in advance, and the user cannot change information sources during query processing. However, there are many cases in which target information sources the user is most interested in change over time. We therefore need an additional framework to deal with changes of the target information sources in the same query. We propose dynamic source selection. Our proposal consists of two functions: The first is a new operator to switch information sources dynamically. The second is connection management to reduce connections to unnecessary information sources. We have implemented the proposed framework in a stream processing system and have evaluated the efficiency of our proposed method in stream data simulator.