Incremental Scheduling Using Genetic Algorithm

Safety-critical systems are integral to ensuring the reliability and safety of operations in sectors where failures can lead to severe losses. These systems are characterized by their complex and reactive nature, requiring stringent adherence to functional and timing constraints. In real-time applications like those in the automotive industry, guaranteed functionality and timeliness are paramount. Task scheduling within these systems is a crucial component in preventing catastrophic failures. This paper presents an incremental scheduling algorithm based on a genetic algorithm designed to enhance the adaptability of embedded systems and minimize system disruptions, which will contribute to cutting down the system's re-validation cost. Our proposed two-level scheduling algorithm aims to minimize the changes to existing schedules, thereby reducing the computational overhead and re-validation costs, ensuring system stability and reliability. The first level computes a base schedule for the initial application model, while the second level adapts to new tasks with minimal adjustments. This approach maintains integrity and real-time delivery, outperforming traditional dynamic scheduling methods. The results showed that the proposed scheduling method outperforms the traditional genetic algorithm in minimizing changes to the base schedule by up to 57% for different application sizes.