A hybrid priority-aware genetic algorithm and opposition-based learning for scheduling IoT tasks in green fog computing

• Published in: Computer networks (Amsterdam, Netherlands : 1999) Vol. 267; p. 111349
• Main Authors: Salimi, Rezvan, Azizi, Sadoon, Dogani, Javad
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.07.2025
• Subjects: Cost; Energy; Genetic Algorithm (GA); Green fog computing; Internet of Things (IoT); Opposition-based learning (OBL); Task scheduling; Internet of Things (IoT); Energy; Green fog computing; Task scheduling; Opposition-based learning (OBL); Genetic Algorithm (GA); Cost
• ISSN: 1389-1286
• DOI: 10.1016/j.comnet.2025.111349

With the rapid growth of Internet of Things (IoT) devices, efficient task scheduling in fog computing systems has become crucial to ensure optimal resource utilization. In addition, the increasing demand for eco-friendly solutions has led to the emergence of green fog computing, which aims to leverage renewable energy sources to power fog nodes. Difficulties such as the diverse requirements of IoT tasks, the distributed and heterogeneous nature of fog nodes, and the fluctuations of renewable energy sources have made the task scheduling problem increasingly complex and pose significant challenges. To address these issues, in this paper, we first present a mixed-integer nonlinear programming (MINLP) model with the objective of minimizing the total system cost, which consists of brown energy consumption, deadline violation time, and monetary cost. To provide an effective and efficient solution for the model, we then propose PGA-OBL, a hybrid algorithm that combines the priority-aware genetic algorithm with an opposition-based learning strategy. The proposed algorithm is implemented in Python and evaluated through various experiments, comparing its performance with a standard genetic algorithm, a priority-aware semi-greedy approach, and a green energy-aware algorithm. The results confirm that PGA-OBL achieves significantly better convergence than the standard genetic algorithm. Additionally, it reduces the total system cost by approximately 6.2% to 20.8% compared to competing approaches.