Energy Efficiency Analysis of Real-Time IoT Systems under DVFS, Offloading, and Task Load Variations

This paper presents an in-depth analysis of energy efficiency in real-time Internet of Things (IoT) systems by jointly considering dynamic voltage and frequency scaling (DVFS), task offloading strategies, and varying task set loads. Unlike prior studies that treat these techniques in isolation, we investigate how their interplay affects overall energy consumption under strict real-time constraints. We develop a comprehensive evaluation framework that explores diverse system configurations, including CPU frequency settings, offloading decisions to edge or cloud servers, and network bandwidth conditions. Through extensive simulations, we identify distinct patterns: DVFS alone provides modest benefits under light workloads, while offloading significantly reduces energy consumption as computational load increases. Furthermore, we show that edge offloading is generally preferable under light loads due to low latency, whereas cloud offloading becomes more effective under heavy loads because of superior computing capacity. Our analysis also reveals how CPU frequency distributions shift depending on the presence of offloading, and how subsystem-level energy profiles (CPU, memory, network) vary across execution strategies. These findings offer practical insights for designing energy-aware real-time IoT systems by guiding configuration choices across workload and infrastructure conditions.