Stream-based real world information integration framework

Abstract

For real world oriented applications to easily use sensor data obtained frommultiple wireless sensor networks, a data management infrastructure is mandatory. The infrastructure design should be based on the philosophy of a novel framework beyond the relational data management for two reasons: First is the freshness of data. To keep sensor data fresh, an infrastructure should process data efficiently; this means conventional time consuming transaction processing methodology is inappropriate. Second is the diversity of functions. The primary purpose of sensor data applications is to detect events; unfortunately, relational operators contribute little toward this purpose. This chapter presents a framework that directly supports efficient processing and a variety of advanced functions. Stream processing is the key concept of the framework. Regarding the efficiency requirement, we present a multiple query optimization technique for query processing over data streams; we also present an efficient data archiving technique. To meet the functions requirement, we present several techniques on event detection, which include complex event processing, probabilistic reasoning, and continuous media integration. © 2010 Springer-Verlag Berlin Heidelberg.