Abstract

Real-time scheduling is the primary research for designing real-time systems whose correctness is determined by not only logical correctness but also timely execution. Real-time scheduling involves two fundamental issues: scheduling algorithm design and schedulability analysis development, which aim at developing a prioritization policy for real-time tasks and offering their timing guarantees at design time, respectively. Among the numerous scheduling algorithms and schedulability analysis for a multiprocessor platform, the contention-free (CF) policy and response-time analysis (RTA) have received considerable attention owing to their wide applicability and high analytical performance, respectively. Notwithstanding their effectiveness, it has been conjectured that it is not feasible to exploit the two techniques together. In this study, we propose a new schedulability analysis for the CF policy, referred to as pseudo-response time analysis (PRTA), which exploits a new notion of pseudo-response time effectively capturing the time instant at which the schedulability of a task is guaranteed under the CF policy. To demonstrate the effectiveness of PRTA, we apply PRTA to the existing earliest deadline first and rate monotonic scheduling algorithms employing the CF policy, and show that up to 46.4% and 18.3% schedulability performance improvement can be achieved, respectively, compared to those applying the existing schedulability analysis.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call