Scheduling Algorithms for Multiprogramming in a Hard-Real-Time Environment
Journal of the ACM (JACM)
Timing Anomalies in Dynamically Scheduled Microprocessors
RTSS '99 Proceedings of the 20th IEEE Real-Time Systems Symposium
The worst-case execution-time problem—overview of methods and survey of tools
ACM Transactions on Embedded Computing Systems (TECS)
| Hi-index | 0.00 |
Along with the growing diversity and complexity of real-time embedded systems, significance of real-time scheduling is increasing. Rate monotonic (RM) and earliest deadline first (EDF) are representative scheduling algorithms for periodic tasks. RM has a merit that tasks with high-priority (short-period) have small jitters and short response times. However, it is impossible that processor utilization reaches 100% while maintaining schedulability or that tasks are given importance independent of their periods. On the other hand, EDF can utilize processors by 100% with schedulability, while it cannot give tasks fixed priorities or importance, therefore, it is difficult to keep jitters or response times of particular tasks short. This paper, in EDF algorithm, proposes a method of shortening response times of a task that is important for the system by introducing predictive execution times and dividing execution of the important task into two instances based on the predictive execution times. The use of predictive execution times has an effect of making deadline of the first instance earlier and therefore the first instance would be executed earlier by EDF. In the evaluation by simulation, the proposed method reduced average response times of the most important task by up to 14.2%.