Self-Adaptive Cloud Cost Management Through Reinforcement-Driven Intelligence
Abstract
Cloud computing has transformed the current IT infrastructure by offering the ability to scale and on-demand resources. Nevertheless, the dynamic workloads changes and sophisticated pricing schemes frequently result in the inefficient use of the resources and the excessive operating expenses. The conventional rule-based auto-scaling and heuristic optimization methods are not flexible and do not react well to real-time variations in the workload. The paper will suggest an independent cloud cost optimization framework based on Reinforcement Learning (RL) and dynamically balancing the resource provisioning to ensure compliance with the Service Level Agreement (SLA). The optimization problem is stated as a Markov Decision Process (MDP) where the scaling actions are carried out by the RL agent, who monitors the states of the system (CPU utilization, memory consumption, and request rates) and applies the scaling actions to minimize the cost without compromising the performance. A reward functionality is intended to achieve a balance between the cost-cutting and penalties in SLA violations to make smart trade-off decisions. The given method involves the use of a Deep Reinforcement Learning model to train the best scaling policies for continuous interaction with a simulated cloud environment. Experimental analysis has shown a great amount of cost saving over the traditional threshold-based and predictive scaling techniques without experiencing much of the response time and utilization efficiency being affected. Findings show that the RL-based framework is able to adjust to the variations in workload and minimize instances of over-provisioning and under-provisioning. The paper identifies the promise of self-directed, learning-oriented processes to improve cloud resource management and attain sustainable cost-efficiency in the current cloud infrastructures.
References
D. Sundar, “Reinforcement learning techniques for autonomous cloud optimization and adaptive resource management,” International Journal of Artificial Intelligence, Data Science, and Machine Learning, vol. 6, no. 3, pp. 134–145, Sep. 2025.
F. Cheng, Y. Huang, B. Tanpure, P. Sawalani, L. Cheng, and C. Liu, “Cost-aware job scheduling for cloud instances using deep reinforcement learning,” Cluster Computing, vol. 25, no. 1, pp. 619–631, Feb. 2022.
S. D. Alizadeh Javaheri, R. Ghaemi, and H. Monshizadeh Naeen, “An autonomous architecture based on a reinforcement deep neural network for resource allocation in cloud computing,” Computing, vol. 106, no. 2, pp. 371–403, Feb. 2024.
A. Mampage, S. Karunasekera, and R. Buyya, “A deep reinforcement learning based algorithm for time and cost optimized scaling of serverless applications,” Future Generation Computer Systems, vol. 173, p. 107873, Dec. 2025.
P. Enyiorji, “Designing a self-optimizing cloud-native autonomous finance system for SMEs using multi-agent reinforcement learning,” International Journal of Financial Management and Economics, vol. 8, no. 1, pp. 596–605, 2025.
A. Jayanetti, S. Halgamuge, and R. Buyya, “A deep reinforcement learning approach for cost optimized workflow scheduling in cloud computing environments,” in Proceedings of the Asia Pacific Conference on Computing Technologies, Communications and Networking, Jul. 2024, pp. 74–82.
G. Rjoub, J. Bentahar, O. Abdel Wahab, and A. Saleh Bataineh, “Deep and reinforcement learning for automated task scheduling in large-scale cloud computing systems,” Concurrency and Computation: Practice and Experience, vol. 33, no. 23, p. e5919, Dec. 2021.
E. Di Caro and M. Tortonesi, “Machine learning for performance optimization in resource-constrained environments,” in Proceedings of the 21st International Conference on Network and Service Management (CNSM). IEEE, Oct. 2025, pp. 1–4.
N. Paton, M. A. de Aragão, K. Lee, A. A. Fernandes, and R. Sakellariou, “Optimizing utility in cloud computing through autonomic workload execution,” Bulletin of the Technical Committee on Data Engineering, vol. 32, no. 1, pp. 51–58, 2009.
J. Zhang, L. Cheng, C. Liu, Z. Zhao, and Y. Mao, “Cost-aware scheduling systems for real-time workflows in cloud: An approach based on genetic algorithm and deep reinforcement learning,” Expert Systems with Applications, vol. 234, p. 120972, Dec. 2023.
V. Asghari, Y. Wang, A. J. Biglari, S. C. Hsu, and P. Tang, “Reinforcement learning in construction engineering and management: A review,” Journal of Construction Engineering and Management, vol. 148, no. 11, p. 03122009, Nov. 2022.
J. Sekar and L. L. Aquilanz, “Autonomous cloud management using AI: Techniques for self-healing and self-optimization,” Journal of Emerging Technologies and Innovative Research, vol. 11, pp. 571–580, Jan. 2023.
M. Cheng, J. Li, and S. Nazarian, “DRL-cloud: Deep reinforcement learning-based resource provisioning and task scheduling for cloud service providers,” in Proceedings of the Asia and South Pacific Design Automation Conference (ASP-DAC). IEEE, Jan. 2018, pp. 129–134.
V. Koc and V. Kamanuru, “Predictive and reinforcement learning-based framework for cloud resource optimization,” in Proceedings of the Systems and Information Engineering Design Symposium (SIEDS). IEEE, May 2025, pp. 128–133.
M. Penmetcha and B. C. Min, “A deep reinforcement learning-based dynamic computational offloading method for cloud robotics,” IEEE Access, vol. 9, pp. 60265–60279, Apr. 2021.
X. Yu, J. Mi, L. Tang, L. Long, and X. Qin, “Dynamic multi-objective task scheduling in cloud computing using reinforcement learning for energy and cost optimization,” Scientific Reports, Nov. 2025.
