Abstract:
In this paper, we address two critical yet conflicting aspects of real-time embedded systems: fault tolerance and energy
consumption. We propose an Optimizing Energy consumption and Fault-Tolerant Scheduling Algorithm (OE FTS), designed to achieve
simultaneous minimization of energy consumption and maximization of system reliability. OE FTS is based on a hybrid combination
of active and passive replication and the Dynamic Voltage and Frequency Scaling (DVFS) technique. Active and passive replication
are employed for tolerating multiple transient faults to improve reliability; therefore, Dynamic Voltage and Frequency Scaling (DVFS)
technique is used for minimizing power consumption. We classify real-time tasks into critical and non-critical categories. Critical
tasks are characterized by tight, stringent deadlines and limited execution time, while non-critical tasks have more flexible and less
strict deadlines. This classification is used to decide which type of redundancy to apply and how to exploit the available slack time
to lower the CPU frequency and voltage, thereby reducing energy consumption. Our technique is applied in a system designed to
manage and execute several dependent tasks scheduled on a set of homogeneous processors linked by a multi-point connection. Our
experimental findings confirm the effectiveness of the proposed approach which improves reliability while managing energy consumption.
