Green Clouds - Load Prediction and Optimization in Private Cloud Systems

About

The amount of energy needed to run the data centers in the world was an astonishing 196 TWh during 2020. If we assume a price of 1 SEK/kWh the energy cost for data centers 2020 was approximately 200 billion SEK (200 * 10⁹ SEK). This figure is expected to grow to between 321 TWh and 2000 TWh by 2030. In 2030, data centers are expected to account for approximately 8% of the global use of electric energy. This means that the environmental and economical impact of the electricity consumption of data centers will be extremely significant in the future.

In the 'Green Clouds' project, we develop load prediction and optimization models for private clouds. These models will make it possible to reduce the use of electricity while still making sure that the service level agreements are met, thus reducing the energy cost and the environmental footprint for the cloud.

The project will be carried out at the computer science at BTH together with two company partners, Cleura AB and Ericsson AB. The computer science environment as well as BTH will be vitalized and renewed by this project since this project bridges BTH's two strategic foci: applied IT and sustainability. Creating such bridges is a priority for BTH and for the research and education environment in computer science.

The approach used in this project is to do data-driven load predictions and optimizations in private clouds with applications from two application domains. There are three reasons for considering private application domain specific clouds: (i) predictions models for application domain specific, and thus more homogeneous, loads are expected to be more accurate, and thus more useful for our industrial partners, (ii) we have access to experts from industry that are familiar with loads from these two application domains, (iii) our industrial partners will provide real world data for building and validating prediction models for these two domains.

Funding

Partners

• Blekinge Institute of Technology (BTH)
• Cleura AB
• Ericsson AB

People

• Prof. Håkan Grahn, project leader
• Prof. Lars Lundberg
• Assoc. Prof. Anton Borg
• Dr. Ashir Javeed, post doc
• Tobias Gustafsson, project assistant

Publications

Peer-reviewed papers

• L. Lundberg, M. Boldt, A. Borg, and H. Grahn, "Bibliometric Mining of Research Trends in Machine Learning," AI, 5(1):208-236, January 2024. doi: https://doi.org/10.3390/ai5010012
• A. Javeed, A. Borg, H. Grahn, L. Lundberg, D. Patel, and S. Shirinbab, "Predicting CPU Load in Heterogeneous Cloud Applications," in: Zbakh, M., Essaaidi, M., Tadonki, C., Touhafi, A., Panda, D.K. (eds) Proceedings of the 7th International Conference on Cloud Computing and Artificial Intelligence: Technologies and Applications (CloudTech'25), Lecture Notes in Networks and Systems, Volume 1941, pp 355-363, 2026. doi: https://doi.org/10.1007/978-3-032-23844-3_35
• A. Javeed, A. Borg, H. Grahn, L. Lundberg, D. Patel, and S. Shirinbab, "Improving Cloud Efficiency: A Machine Learning-Based Stacking Model for CPU Utilization Prediction," 8th International Conference on Data Science and Machine Learning Applications (CDMA2025), pp. 120-125, February 2025. doi: 10.1109/CDMA61895.2025.00026

Non peer-reviewed papers

• A. Javeed, A. Borg, H. Grahn, L. Lundberg, D. Patel, and S. Shirinbab, "A Two-Stage Forecasting System for CPU Workload Prediction in Private Clouds," Computing Research Repository (CoRR), arXiv:2609.03457 [cs.LG], September 2026, https://arxiv.org/abs/2609.03457

MSc thesis projects

• Nils Persson Suorra, "Multivariate Resource Usage Forecasting and Temporal Accuracy in Private Cloud Systems," Master's thesis, June 2025. The thesis was done together with Cleura.
  https://urn.kb.se/resolve?urn=urn:nbn:se:bth-27969
• Simon Lundberg, "Load Prediction and Forecasting in Private Cloud Systems," Master's thesis, June 2025. The thesis was done together with Ericsson.
  https://urn.kb.se/resolve?urn=urn:nbn:se:bth-27934
• Elliot Marcusson, "Predicting Application Load in Private Cloud Environments - A Comparative Study of Machine Learning Models in Private Clouds," Master's thesis, October 2025. The thesis was done together with Ericsson.
  https://urn.kb.se/resolve?urn=urn:nbn:se:bth-28793
• Gustav Karlsson, "Procurement planning and offline job scheduling in private clouds," Master's thesis, Augusti 2026.
  https://urn.kb.se/resolve?urn=urn:nbn:se:bth-29655