A soft aperiodic task scheduling algorithm in dynamic-priority systems
RTCSA '95 Proceedings of the 2nd International Workshop on Real-Time Computing Systems and Applications
| Hi-index | 0.00 |
A real-time system typically has a mixture of periodic and aperiodic workload. Requests in each periodic task have regular arrival times and hard deadlines. Aperiodic requests, on the other hand, have random arrival times and may have hard or soft deadlines. We called these aperiodic requests with hard deadlines sporadic requests and those with soft deadlines aperiodic requests. This thesis extends two common approaches to scheduling periodic requests to address the problems of jointly scheduling periodic requests with aperiodic requests or sporadic requests. Specifically, we present algorithms for scheduling aperiodic requests in dynamic-priority systems, sporadic requests in dynamic-priority systems, and aperiodic requests in fixed-priority systems. Our algorithms are based on the concept of slack , which is the amount of processor time available to schedule the aperiodic or sporadic requests without causing any periodic requests to miss their deadlines. Two features that distinguish our algorithms from previous slack stealing algorithms proposed in the literature are (1) the efficient computation of slack, and (2) the optimal usage (i.e., when and how much to use) of the available slack to schedule the aperiodic or sporadic requests. For aperiodic requests, our algorithms are optimal in that the aperiodic requests have the minimum possible response times. For sporadic requests, they guarantee the deadlines of as many sporadic requests as possible.