Real-Time Database - Similarity Semantics and Resource Scheduling

While much recent work has focussed on the performance of transaction systems where individual transactions have deadlines, our research addresses the semantics of data usage in real-time applications and its integration with real-time resource management, in particular, the timeliness value of real-time data and the inherent path and not state-based constraints on concurrency control. Central to our research is the idea of similarity which is a reflexive, symmetric relation over the domain of a data object. By exploiting the similarity relation, we propose a class of efficient data-access policies for real-time data objects. We shall also discuss the design of a distributed real-time data-access interface. Our goal is to build a database facility which can support predictable real-time applications involving high-speed communication, information access, and multimedia.