Article Sidebar
Main Article Content
Abstract
Real-time scheduling is a fundamental requirement in embedded systems, where system correctness depends not only on logical accuracy but also on meeting strict temporal constraints. Selecting an appropriate scheduling algorithm is critical to ensuring predictability, responsiveness, and efficient resource management. This paper presents a comprehensive review and practical evaluation of four classical real-time scheduling algorithms: Rate Monotonic Scheduling (RMS), Deadline Monotonic Scheduling (DMS), Earliest Deadline First (EDF), and Least Laxity First (LLF). A custom simulation framework was developed in Python to evaluate these algorithms under various workload scenarios (low to overload) and diverse task-set sizes (5, 10, and 20 tasks). Performance was assessed using metrics including deadline miss ratio, average response time, and scheduling overhead (quantified by context switches). Results demonstrate that while all algorithms perform reliably under normal conditions, significant divergence occurs under overload. RMS consistently achieved the lowest response times and overhead, whereas dynamic-priority algorithms like LLF suffered from performance degradation due to high scheduling complexity. These findings offer practical guidance for selecting scheduling strategies in resource-constrained embedded environments.