Issues in achieving temporal and semantic coherency in distributed real-time applications

Abstract

Modern information-intensive applications require timely access to temporally accurate data. This demands the collection, integration and processing of temporally-coherent data and its display in a “continuous” fashion. When utilizing several pieces of information, each must be fresh relative to the use of that information-and together, they must be relatively consistent. The techniques used must ensure temporal coherency among the data being integrated. When information from multiple media must be processed and presented, the quality of the final information must “make sense”. This leads to semantic coherency requirements. Techniques must support the processing of such data under strict timing concerns too. Timeliness is an important QoS (quality of service) parameter. The impedance mismatch between what users think they want and what the system components involved are and what they can deliver, plus the multiplicity of components that must be orchestrated together makes the problem of guaranteeing a specified QoS a challenging one. Specifically, we need (1) precise ways of specifying QoS requirements and (2) an integrated platform supporting hard real-time control tasks, multimedia data and a distributed data repository, built over a common set of computational resources. In this paper, we elaborate on these two needs.