Constraint-Based Schedulability Analysis in Multiprocessor Real-Time Systems

Abstract

As the demand for more functions and capabilities in the system increases, the application of multiprocessors has brought advantages in many ways. Many systems now have multiprocessors, and safety-critical systems with real-time properties are no exception. In these systems where the satisfaction of real-time properties is directly linked to the safety of life, the predictability of the behavior is very important, and the behavior of the system can be predicted using the schedulability analysis. In this paper, we propose the schedulability analysis of a real-time system in a homogeneous multiprocessor environment through constraint solving approach. First, the constraints that must be satisfied in order for the system to function properly were derived. These include the constraints of the task behavior, the scheduling behavior, and the operating conditions of a homogeneous multiprocessor environment. Once all the constraints were identified, they were encoded in the form of first-order logic expressions. The encoded constraints are then entered into a constraint solver along with a set of tasks. Finally, the solver provides a schedulable answer if the set of tasks satisfies all the constraints.

A Survey of Techniques for Reducing Interference in Real-Time Applications on Multicore Platforms

“…Zhao et al [180] and Lee et al [181] proposed simplified schedulability models considering only tasks with uncertain worst-case execution time, respectively. Lee and Choi [194] generalized the problem by presenting a constraint-based schedulability analysis in real-time multiprocessor systems, considering a global approach. Power was included in the schedulability analysis in Lee et al [195] by considering transiently powered processors in the schedule.…”