|
Abstract : |
This paper examines the problem of scheduling concurrent requests on multimedia storage servers consisting of multiple resources, which is a central issue in processing and optimization of complex queries in multimedia database systems [Cha94]. It introduces a formal model for the demands imposed by multimedia requests on the server resources, describes a heuristic algorithm for scheduling the requests that is based on the formal model, and presents a formal result that bounds the performance of the schedule produced by the algorithm compared to the optimal schedule. Scheduling requests that involve continuous media data types, e.g., audio and video objects, is a challenging problem due to two distinguishing characteristics of multimedia environments: 1. Service requests to servers are delay-sensitive, having specific real-time constraints for successive media quanta. 2. Service requests to servers are typically multi-dimensional, in the sense that they alternate between multiple resources of the server during their execution. For example, consider a content-based video retrieval request issued from a remote client site, e.g., asking for sequences of digital video frames that show the Parthenon. With respect to point 1 above, the server must ensure that the video frames are delivered to the user at some prespecified, |
