Energy-aware hybrid artificial bee colony and reinforcement learning for multi-objective task scheduling in green cloud–fog industrial IoT systems

Abstract

Efficient task scheduling in Cloud–Fog computing environments remains a critical challenge in Industrial Internet of Things (IIoT) systems due to the conflicting requirements of energy efficiency, latency reduction, and Quality of Service (QoS) assurance. This paper proposes an Energy-Aware Improved Artificial Bee Colony (EA-IABC) algorithm integrated with a Q-learning-based reinforcement learning (RL) module for multi-objective task scheduling in Green IIoT infrastructures. The proposed method enhances the exploration and exploitation capabilities of the standard Artificial Bee Colony (ABC) algorithm through adaptive scout behavior, inertia-weighted Lévy flight search, and constraint-aware optimization, while the RL component enables dynamic, state-dependent resource selection. The scheduling problem is formulated as a bi-objective optimization model that minimizes execution makespan and total energy consumption under QoS and resource constraints. Throughput and carbon footprint are considered as derived evaluation metrics to assess performance efficiency and environmental impact. Extensive simulations are conducted using an enhanced iFogSim-based environment with synthetic workloads and trace-driven benchmark workloads including NASA iPSC and HPC2N. Experimental results demonstrate that EA-IABC consistently outperforms baseline metaheuristic and learning-based methods, achieving significant improvements in energy efficiency and execution latency, while improving throughput as a derived performance indicator. Statistical analysis confirms the robustness and scalability of the proposed approach across heterogeneous and large-scale workload scenarios. The results indicate that hybridizing swarm intelligence with RL provides an effective direction for sustainable and adaptive batch-oriented scheduling in next-generation Cloud–Fog IIoT systems.