A Carbon-Aware Task Offloading Framework for Sustainable Fog Computing

Authors:

Abstract
Fog computing is paving the way for time-sensitive and energy-efficient IoT applications by placing computation closer to end devices. However, sustainability concerns due to its increasing energy demands and associated carbon emissions stem from the expansion of fog infrastructure. Despite numerous offloading strategies that focus on energy consumption and makespan, green computing and its broader environmental impact are often overlooked. To address this gap, we propose a carbon-aware multi-objective optimisation model that integrates three metaheuristic algorithms - GWO, CSA, and SSA - to balance makespan and energy consumption while minimising $C O_{2}$ emissions. In our proposed framework, node suitability is evaluated based on a customised energy model designed to prevent the overuse of non-renewable energy sources and maintain system responsiveness. Our framework advances foglevel scheduling by integrating a hybrid energy model, unlike prior cloud-focused carbon-aware approaches. The optimisation process simultaneously considers three objectives - energy, makespan, and carbon emissions - through a hybrid metaheuristic algorithm. Simulation results using the LEAF simulator indicate that our framework achieves an average improvement in energy consumption, makespan, and $C O_{2}$ emissions of approximately $18 \%, 23 \%$, and 20 %, respectively, compared to EEAIOT-EFC and MOHHOSSA. These results show the viability of integrating carbon awareness into fog-cloud task offloading for green computing.

This publication has 26 references indexed in Scilit: