P2SDS: A Polynomial-Time Pattern-Guided Stable Dynamic Scheduling for Weakly Hard Control Task Systems

Article Type

Research Article

Publication Title

ACM Transactions on Embedded Computing Systems

Abstract

Real-time scheduling of control tasks in a weakly hard system, where the tasks can miss a few of their deadlines without impeding the system's performance, is a riveting research direction nowadays. While analyzing the schedulability of control tasks in a weakly hard setting, it is pivotal to take into account both the control stability and the desired performance of the underlying system. Though a scanty amount of research efforts are reported in the literature, focusing on the control-scheduling co-design aspects, most of them are completely offline in nature, and hence, not applicable for obtaining dynamic scheduling decisions at runtime. More specifically, a polynomial-time online scheduler, handling both stability and control performance, is almost absent in the literature. To bridge this design gap, in this work, we propose P2SDS, a novel online scheduling approach that preserves stability and enhances control performance by synthesizing optimal control execution patterns (CEPs) for scheduling control tasks, while running in polynomial time. The synthesized CEPs respect stability-induced weakly hard constraints endorsing optimal control performances of the underlying systems. A rigorous set of simulation-based experiments over 15 standard benchmarks from the automotive domain is carried out to establish the efficacy and real-time applicability of the proposed method. A comparative analysis of P2SDS with respect to the state-of-the-art approaches reports around 99% improvement in running time at maximum; 88%, 9.042%, and 87.5% improvements in stability, control performance, and schedulability ratio, respectively.

DOI

10.1145/3748329

Publication Date

9-13-2025

Link to Full Text

Share

COinS