![]() ![]() ![]() |
![]() |
|
|
![]() ![]() ![]() ![]() ![]() |
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Return to Papers Advances in computer graphics, authoring tools and the explosive growth of the Internet has increased the use of multimedia presentations. This article presents a new retrieval scheduling technique to support the display of multimedia presentations in a multi-user environment. A multimedia presentation consists of a collection of objects with temporal constraints that define when the objects are rendered. A scheduling algorithm must determine when objects are retrieved from disk to satisfy the temporal constraints of the presentation. The time elapsed between the arrival of a request and the onset of its display (latency) depends upon the resources (CPU, disk, and memory) available to the system. The resources available depend upon those consumed by other presentations already being displayed. Therefore, the latency must be computed when the new request for a presentation arrives and that latency must include its computation time. Prior scheduling techniques applicable to arbitrary resource requirements have quadratic time complexity. Unlike priorwork, our scheduling algorithm has linear time complexity. We compare the performance of our scheduling technique with one that exhaustively searches for the earliest time to schedule a presentation. Our simulation results show that our technique significantly reduces the latency of a presentation as compared with the exhaustive search. ![]() DiSC'02 © 2003 Association for Computing Machinery |