Real-Time Scheduling Approach of Reconfigurable Embedded Systems Based On Neural Networks with Minimization of Power Consumption

摘要

This paper deals with the real-time scheduling of embedded systems based on the neural networks with low power consumption optimization. Indeed, while most embedded systems are designed for real-time applications, they suffer from resource constraints and energy consumption. Many techniques have been proposed for real-time task scheduling to reduce energy consumption. A combination of Dynamic Voltage Scaling (DVS) and feedback scheduling can be used to scale dynamically the frequency by adjusting the operating voltage, and improve the run-time reliability of embedded systems. We present in this paper a novel hybrid contribution that handles real-time scheduling of embedded systems and low power consumption based on the combination of DVS and Neural Feedback Scheduling NFS with the priority-energy earliest-deadline-first (PEDF) algorithm.