Article Sidebar
Main Article Content
Abstract
Modern safety-critical real-time systems, such as those in the automotive and aerospace domains, increasingly demand the concurrent integration of hierarchical scheduling, shared hardware resource management, and dynamic resource provisioning on multicore platforms. However, state-of-the-art response time analysis (RTA) techniques address these interdependent dimensions in isolation, compounding analytical pessimism or inducing prohibitive computational overhead. This paper introduces a unified RTA framework that holistically harmonizes time-partitioned hierarchical scheduling, cross-partition resource contention, and adaptive resource elasticity. By extending the classical periodic resource model to accommodate dynamic CPU and memory scaling while preserving strict temporal isolation guarantees, we derive novel, tightly integrated response time bounds that simultaneously capture co-running partition interference and mutual exclusion resource conflicts. To mitigate computational complexity, we develop an optimized O(n²) iterative convergence algorithm featuring accelerated initialization and early termination mechanisms. Empirical evaluations using both synthetic workloads and realistic automotive Electronic Control Unit (ECU) benchmarks demonstrate that our framework delivers up to 45% tighter response time bounds while reducing computational analysis overhead by 60% to 89% compared to state-of-the-art baselines. Ultimately, this framework provides a mathematically rigorous verification tool enabling system engineers to safely exploit resource elasticity without compromising hard real-time schedulability guarantees.