Comprehensive study of schedulability tests and optimal design for rate-monotonic scheduling

Abstract

Schedulability analysis of the rate-monotonic (RM) scheduling algorithm is an important research field in real-time systems. New dimensions in RM have been explored, ranging from faster schedulability tests to the optimal design of real-time systems. Recently, researchers have extended the schedulability problem to a more generalized case: the rate-monotonic optimal-design problem (RM-ODP), wherein the execution time is limited to an interval rather than a sole point. The RM-ODP processes task-execution-time adjustment such that (i) the system is RM schedulable and (ii) certain system performances (e.g., processor utilization) are optimized. In this paper, we review RM scheduling and summarize three aspects of the related research: (1) evaluate existing relevant feasibility tests by categorizing them based on inexact or exact conditions; (2) comprehensively review RM-ODP, summarize associated advantages and disadvantages, and provide some recommendations on schedulability tests to achieve the optimal system design under RM-scheduling policies; and (3) propose a hybrid search method based on linear programming with respect to both depth-first search and breadth-first search using the dynamic pruning strategy.