Energy-Aware Dynamic Task Offloading Using Deep Reinforcement Learning in Cloud–Edge–IoT Systems
Main Article Content
Abstract
The rapid expansion of Internet of Things applications has increased the demand for continuous processing of computation-intensive tasks generated by resource-constrained devices. Conventional cloud-centric processing can introduce substantial transmission delays and communication energy consumption, particularly when large volumes of data are generated close to the network edge. Cloud–edge–IoT architectures provide a distributed alternative by placing computational resources closer to end devices; however, deciding where and when individual tasks should be processed remains challenging because device workloads, wireless conditions, queue states, available computing resources, and battery levels continuously change. This research proposes an energy-aware dynamic task offloading framework based on deep reinforcement learning for cloud–edge–IoT environments. The proposed framework models task placement as a sequential decision problem in which an intelligent agent observes device energy status, task characteristics, communication conditions, queue information, and available edge and cloud resources before selecting an execution location. The possible decisions include local processing, edge execution, and cloud processing, enabling the system to adapt its behavior to current operating conditions. A composite reward mechanism balances energy consumption, task completion delay, deadline violations, and resource utilization. The framework also incorporates dynamic resource availability so that overloaded edge nodes can be avoided while underutilized resources can be exploited. Deep reinforcement learning enables the offloading policy to improve through repeated interaction with changing system states without requiring an explicit analytical representation of every network transition. The proposed approach is intended to reduce unnecessary energy expenditure while maintaining acceptable response times for delay-sensitive IoT applications. Its performance can be evaluated using heterogeneous workloads and varying network conditions against local execution, greedy offloading, conventional optimization, and established reinforcement-learning strategies. The resulting framework provides a flexible basis for intelligent task placement and resource management across interconnected IoT devices, edge servers, and cloud resources.
Article Details
Section
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
How to Cite
References
[1]. W. Shi, J. Cao, Q. Zhang, Y. Li, and L. Xu, “Edge Computing: Vision and Challenges,” IEEE Internet of Things Journal, vol. 3, no. 5, pp. 637–646, 2016. This foundational work discusses bringing computation and services closer to end devices to address latency and resource limitations.
[2]. Y. Mao, C. You, J. Zhang, K. Huang, and K. B. Letaief, “A Survey on Mobile Edge Computing: The Communication Perspective,” IEEE Communications Surveys & Tutorials, vol. 19, no. 4, pp. 2322–2358, 2017.
[3]. Y. Mao, J. Zhang, and K. B. Letaief, “Dynamic Computation Offloading for Mobile-Edge Computing With Energy Harvesting Devices,” IEEE Journal on Selected Areas in Communications, vol. 34, no. 12, pp. 3590–3604, 2016.
[4]. Y. Mao, J. Zhang, and K. B. Letaief, “A Lyapunov Optimization Approach for Green Mobile-Edge Computing Systems,” IEEE Transactions on Wireless Communications, vol. 15, no. 11, pp. 7529–7543, 2016.
[5]. T. Q. Dinh, J. Tang, Q. D. La, and T. Q. S. Quek, “Offloading in Mobile Edge Computing: Task Allocation and Computational Resource Optimization,” IEEE Transactions on Communications, vol. 65, no. 8, pp. 3571–3584, 2017.
[6]. S. Sardellitti, G. Scutari, and S. Barbarossa, “Joint Optimization of Radio and Computational Resources for Multicell Mobile-Edge Computing,” IEEE Transactions on Signal and Information Processing over Networks, vol. 1, no. 2, pp. 89–103, 2015.
[7]. L. Huang, S. Bi, and Y.-J. A. Zhang, “Deep Reinforcement Learning for Online Computation Offloading in Wireless Powered Mobile-Edge Computing Networks,” IEEE Transactions on Mobile Computing, vol. 19, no. 11, pp. 2581–2593, 2020.
[8]. Y. Zhang, T. Liu, Y. Zhu, and Y. Yang, “A Deep Reinforcement Learning Approach for Online Computation Offloading in Mobile Edge Computing,” Proceedings of the IEEE International Conference on Communications, 2020. The work investigates online offloading and jointly considers task latency and energy consumption.
[9]. R. Zhao, X. Wang, J. Xia, and L. Fan, “Deep Reinforcement Learning Based Mobile Edge Computing for Intelligent Internet of Things,” Physical Communication, vol. 43, Art. 101184, 2020. The study uses DQN-based decision-making for task offloading and considers latency and energy-related system costs.
[10]. M. Chen, T. Wang, S. Zhang, and A. Liu, “Deep Reinforcement Learning for Computation Offloading in Mobile Edge Computing Environment,” Computer Communications, vol. 175, pp. 1–12, 2021. The study models offloading as an MDP and employs DDPG for distributed resource provisioning.
[11]. Z. Chen and X. Wang, “Decentralized Computation Offloading for Multi-User Mobile Edge Computing: A Deep Reinforcement Learning Approach,” EURASIP Journal on Wireless Communications and Networking, 2020, Art. 188.
[12]. T. Alfakih, M. Hassan, A. Gumaei, et al., “Task Offloading and Resource Allocation for Mobile Edge Computing by Deep Reinforcement Learning Based on SARSA,” IEEE Access, vol. 8, pp. 54074–54084, 2020, doi: 10.1109/ACCESS.2020.2981434.
[13]. L. Ale, N. Zhang, X. Fang, X. Chen, S. Wu, and L. Li, “Delay-Aware and Energy-Efficient Computation Offloading in Mobile Edge Computing Using Deep Reinforcement Learning,” IEEE Internet of Things Journal, 2021. The work considers dynamic edge environments, multiple edge servers, task deadlines, and energy consumption.
[14]. L. Liao, Y. Lai, F. Yang, and W. Zeng, “Online Computation Offloading with Double Reinforcement Learning Algorithm in Mobile Edge Computing,” Journal of Parallel and Distributed Computing, vol. 171, pp. 28–39, 2023. The proposed DRLCO framework jointly addresses offloading, CPU frequency, and transmission-power decisions.
[15]. S. K. Nalluri, V. T. Bathini and V. K. B. Parasaram, "AI-Augmented Manufacturing Execution Systems for Precision Healthcare: A Scalable Framework for Smart Genomic and Proteomic Production," 2026 IEEE 2nd International Conference on Secure IoT, Assured and Trusted Computing (SATC), Houston, TX, USA, 2026, pp. 1-6, doi: 10.1109/SATC69565.2026.11542384.
[16]. P. Peng, W. Lin, W. Wu, H. Zhang, S. Peng, Q. Wu, and K. Li, “A Survey on Computation Offloading in Edge Systems: From the Perspective of Deep Reinforcement Learning Approaches,” Computer Science Review, vol. 53, Art. 100656, 2024.
[17]. A. Shakarami, M. Ghobaei-Arani, and A. Shahidinejad, “A Survey on the Computation Offloading Approaches in Mobile Edge Computing: A Machine Learning-Based Perspective,” Computer Networks, vol. 182, Art. 107496, 2020.
[18]. A. Shakarami, A. Shahidinejad, and M. Ghobaei-Arani, “A Review on the Computation Offloading Approaches in Mobile Edge Computing: A Game-Theoretic Perspective,” Software: Practice and Experience, vol. 50, pp. 1719–1759, 2020.
[19]. “A Survey on Computation Offloading in Mobile Edge Computing Networks,” Journal of Network and Computer Applications, 2022. The survey examines delay minimization, energy minimization, resource allocation, server selection, and RL-based approaches.
[20]. “A Survey on Computation Offloading Modeling for Edge Computing,” Journal of Network and Computer Applications, vol. 169, Art. 102781, 2020. The work reviews offloading decisions, task partitioning, resource allocation, MDPs, optimization, and learning-based models.
[21]. “Energy-Efficient Computation Offloading and Resource Allocation for Delay-Sensitive Mobile Edge Computing,” International Journal of Distributed Sensor Networks, 2018. The work jointly considers completion time, energy consumption, CPU frequency, transmission power, and offloading decisions.
[22]. Koneni, C., S. Lokula, and L. Panjiyar. "A Cloud-Native Approach to Statistical Anomaly Detection and Automated Data Quality Validation." International Journal of Data Science and IoT Management System 4.4 (2025): 533-543
[23]. “Resource Allocation and Computation Offloading with Data Security for Mobile Edge Computing,” Future Generation Computer Systems, 2019. The study formulates joint resource allocation and offloading while considering energy, time, and security requirements.
[24]. B. Huang et al., “Security and Cost-Aware Computation Offloading via Deep Reinforcement Learning in Mobile Edge Computing,” Wireless Communications and Mobile Computing, 2019, Art. 3816237. The work formulates secure offloading as an MDP while considering energy, delay, and task-loss-related costs.
[25]. “Deep Reinforcement Learning-Based Computation Offloading and Distributed Edge Service Caching for Mobile Edge Computing,” Computer Networks, vol. 250, Art. 110564, 2024. The study jointly considers latency, energy, resource allocation, task offloading, and edge service caching using LSTM-DDPG.
[26]. C. Koneni and S. Lokula, "Evidence-Quality Telemetry for Cloud Incident Response: Detecting Gaps, Drift, and Integrity Failures in Large-Scale Observability Pipelines," 2026 14th International Symposium on Digital Forensics and Security (ISDFS), Boston, MA, USA, 2026, pp. 1-6, doi: 10.1109/ISDFS69419.2026.11459097.
[27]. “Energy-Aware Dynamic Task Offloading Using Deep Reinforcement Transfer Learning in SDN-Enabled Edge Nodes,” Internet of Things, vol. 25, Art. 101118, 2024. The study combines dynamic partial offloading, transfer learning, trust evaluation, and energy-aware decision-making.
[28]. A. K. Chakravarthy, T. Rama Reddy, and N. Ramakrishnaiah, “EdgeAutoOffloadAI: A Deep Reinforcement Learning Framework for Adaptive Task Offloading and Resource Management in Mobile Edge Computing,” Discover Computing, vol. 29, Art. 136, 2026. The recent work combines DRL with Transformer-based state representation and considers latency, energy, resource utilization, and fairness in dynamic edge environments.