The DOI matches its Crossref record; 50 of the 50 references it cites were found in a registry or an index; none of the 6 sources its claims cite is checked for support yet.
Carbon-Aware, Energy-Efficient, and SLA-Compliant Virtual Machine Placement in Cloud Data Centers Using Deep Q-Networks and Agglomerative Clustering
Open accessAbstract
The fast expansion of cloud computing has raised carbon emissions and energy usage in cloud data centers, so creative solutions for sustainable resource management are more necessary. This work presents a new algorithm—Carbon-Aware, Energy-Efficient, and SLA-Compliant Virtual Machine Placement using Deep Q-Networks (DQNs) and Agglomerative Clustering (CARBON-DQN)—that intelligibly balances environmental sustainability, service level agreement (SLA), and energy efficiency. The method combines a deep reinforcement learning model that learns optimum placement methods over time, carbon-aware data center profiling, and the hierarchical clustering of virtual machines (VMs) depending on resource constraints. Extensive simulations show that CARBON-DQN beats conventional and state-of-the-art algorithms like GRVMP, NSGA-II, RLVMP, GMPR, and MORLVMP very dramatically. Among many virtual machine configurations—including micro, small, high-CPU, and extra-large instances—it delivers the lowest carbon emissions, lowered SLA violations, and lowest energy usage. Driven by real-time input, the adaptive decision-making capacity of the algorithm allows it to dynamically react to changing data center circumstances and workloads. These findings highlight how well CARBON-DQN is a sustainable and intelligent virtual machine deployment system for cloud systems. To improve scalability, environmental effect, and practical applicability even further, future work will investigate the integration of renewable energy forecasts, dynamic pricing models, and deployment across multi-cloud and edge computing environments.
Overview
CoChat Summary
TL;DR
The paper proposes CARBON-DQN, a hybrid VM placement approach combining Deep Q-Networks with Agglomerative Clustering and carbon-aware data center profiling to balance carbon emissions, energy efficiency, and SLA compliance. Simulations in a 4000-machine data center report that CARBON-DQN outperforms RLVMP, GRVMP, NSGA-II, MORLVMP, and GMPR on carbon-aware scores, energy consumption, and SLA fulfillment.
Key findings
- CARBON-DQN consistently achieved the lowest carbon-aware scores across Amazon EC2 VM types—micro, small, extra-large, and high-CPU Medium—with the gap widening for larger, resource-intensive VMs.
- It also achieved the lowest energy consumption ratings across the same VM types, with the energy-efficiency gap widening as VM resource demands increased.
- CARBON-DQN sustained SLA fulfillment rates over 90% across all EC2 VM types, including extra-large and high-CPU Medium instances, while traditional algorithms such as GRVMP and NSGA-II struggled as VM resource requirements increased.
- The simulation used a data center with 4000 heterogeneous physical devices and experiments with 256, 512, 1024, 2048, and 4192 VMs, as stated.
Methods
- The proposed CARBON-DQN framework has five phases: carbon-aware data center profiling using a Carbon Efficiency Score (CES) from carbon intensity and energy performance; Agglomerative Clustering of VMs by normalized CPU, memory, storage, bandwidth, and SLA sensitivity; DQN state-space construction; DQN-based placement with a multi-objective reward combining energy saved, SLA compliance, added carbon intensity, and migration cost; and adaptive feedback/optimization.
- Simulations were run in Python 3.13.1 on a laptop with 2.13 GHz maximum frequency, 16 GB RAM, and 64-bit Windows 11, using Amazon EC2 VM request dimensions and an ε-greedy exploration strategy starting at ε = 0.1 and decaying during training.
- The proposed method was compared against RLVMP, GRVMP, NSGA-II, MORLVMP, and GMPR on carbon awareness, energy efficiency, and SLA compliance.
Limitations
- Stated: the approach lacks thorough temporal carbon intensity modeling and relies on static or averaged carbon intensity measurements, ignoring real-time grid carbon fluctuations.
- Stated: it does not integrate renewable energy forecasting, lacks native Kubernetes or container integration, and may face significant real-time inference and retraining overhead in high-throughput or latency-sensitive cloud environments.
- Evident from the text: the reported carbon and energy comparisons are largely qualitative/figural rather than accompanied by exact numerical values, and the evaluation is simulation-based using Amazon EC2-derived VM request dimensions rather than a production deployment.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 3 Not valid · 4 Not checked · 54 Own results and interpretations
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Among many virtual machine configurations-including micro, small, high-CPU, and extra-large instances-it delivers the lowest carbon emissions, lowered SLA violations, and lowest energy usage.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Among many virtual machine configurations-including micro, small, high-CPU, and extra-large instances-it delivers the lowest carbon emissions, lowered SLA violations, and lowest energy usage.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Extensive simulations show that CARBON-DQN beats conventional and state-of-the-art algorithms like GRVMP, NSGA-II, RLVMP, GMPR, and MORLVMP very dramatically.”
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Among many virtual machine configurations-including micro, small, high-CPU, and extra-large instances-it delivers the lowest carbon emissions, lowered SLA violations, and lowest energy usage.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The experimental results indicate that the proposed CARBON-DQN algorithm outperforms conventional VM placement methods such as RLVMP, GRVMP, NSGA-II, MORLVMP, and GMPR regarding carbon awareness, energy efficiency, and SLA compliance.”
hedgedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
26 more claims
Create a free account to see all 61 claims and every source verdict.
- It may state two separate claims as one. 19
- Probably not a substantial claim. 13
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ 3.1. Illustration of the Proposed Model
“This involves collecting real-time environmental data from different physical equipment in the data center, such as carbon intensity and energy efficiency.”
§ 4.3. Limitations
“The current method ignores real-time fluctuations in grid carbon emissions, usually related to the availability of intermittent renewable energy, depending only on static or averaged carbon intensity measurements.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 4.1. Simulation Environment
“Each experiment has four scenarios”
§ 4.1. Simulation Environment
“256, 512, 1024, 2048, and 4192 VMs”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 2. Related Work
“MORLVMP [44] focused on resource optimization and energy efficiency”
§ 2. Related Work
“| [24] | MORLVMP | Energy efficiency and resource utilization”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 3. The Proposed Model
“The model is broken down into five phases”
§ 3. The Proposed Model
“carbon-aware data center profiling, Agglomerative Clustering of VMs, state space construction for DQNs and adaptive feedback and optimization.”
Cited works that were retracted
Works this paper cites that were later retracted, from the retraction notices publishers deposit. Major when the notice came out before this paper was published.
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
“Jeevitha, 2021, A novel scheduling approach to improve the energy efficiency in cloud computing data centers, J. Ambient. Intell. Humaniz. Comput., 12, 6639, 10.1007/s12652-020-02283-6”
Cited work 10.1007/s12652-020-02283-6 · retracted 2022-05-23 · notice 10.1007/s12652-022-03946-2 · this paper published 2025-07-15
Citations
Cites
1 retracted
Jeevitha, J. K., & Athisha, G. (2020) · Journal of Ambient Intelligence and Humanized Computing · doi:10.1007/s12652-020-02283-6 · Cited by 15
RetractedEnergy-Efficient Virtual Machines Consolidation in Cloud Data Centers Using Reinforcement Learning
Farahnakian, F., Liljeberg, P., & Plosila, J. (2014) · 2014 22nd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing · doi:10.1109/pdp.2014.109 · Cited by 163
Energy and Carbon-Efficient Placement of Virtual Machines in Distributed Cloud Data Centers
Khosravi, A., Garg, S. K., & Buyya, R. (2013) · Lecture Notes in Computer Science · doi:10.1007/978-3-642-40047-6_33 · Cited by 156
A Survey of Virtual Machine Placement Techniques in a Cloud Data Center
Usmani, Z., & Singh, S. (2016) · Procedia Computer Science · doi:10.1016/j.procs.2016.02.093 · Cited by 142
Shaw, R., Howley, E., & Barrett, E. (2021) · Information Systems · doi:10.1016/j.is.2021.101722 · Cited by 114
Energy-Efficient Algorithms for Dynamic Virtual Machine Consolidation in Cloud Data Centers
Khoshkholghi, M. A., Derahman, M. N., Abdullah, A., Subramaniam, S. K., & Othman, М. (2017) · IEEE Access · doi:10.1109/access.2017.2711043 · Cited by 109
Ghasemi, A., & Haghighat, A. T. (2020) · Computing · doi:10.1007/s00607-020-00813-w · Cited by 93
A global-energy-aware virtual machine placement strategy for cloud data centers
Feng, H., Deng, Y., & Li, J. (2021) · Journal of Systems Architecture · doi:10.1016/j.sysarc.2021.102048 · Cited by 85
A learning‐based approach for virtual machine placement in cloud data centers
Ghobaei‐Arani, M., Rahmanian, A. A., Shamsi, M., & Kenari, A. R. (2018) · International Journal of Communication Systems · doi:10.1002/dac.3537 · Cited by 81
GRVMP: A Greedy Randomized Algorithm for Virtual Machine Placement in Cloud Data Centers
Azizi, S., Shojafar, M., Abawajy, J., & Buyya, R. (2021) · IEEE Systems Journal · doi:10.1109/jsyst.2020.3002721 · Cited by 80
Research works printed but not linked: 3
- Alsadie, 2022, Virtual Machine Placement Methods using Metaheuristic Algorithms in a Cloud Environment-A Comprehensive Review, Int. J. Comput. Sci. Netw. Secur., 22, 147
- Madhumala, 2020, A Reliable Frame Work for Virtual Machine Selection in Cloud Datacenter Using Particle Swarm Optimization, Int. J. Math. Comput. Sci., 16, 677
- Devarasetty, 2018, Multi objective ant colony optimization algorithm for resource allocation in cloud computing, Int. J. Innov. Technol. Explor. Eng., 8, 68
Cited by
Alshammari, A. (2026) · Sensors · doi:10.3390/s26061780 · Cited by 3
“Studies on carbon-aware cloud resource management and green analytics frameworks have demonstrated significant potential for reducing emissions in large-scale computing infrastructures [39,40].”
Devadas, R. M., & Sowmya, T. (2026) · Energy Informatics · doi:10.1186/s42162-026-00651-8 · Cited by 4
Hierarchical risk-aware RL for DVFS-enabled IoE scheduling
Javadpour, A., Ja’fari, F., & Taleb, T. (2026) · Computing · doi:10.1007/s00607-026-01732-y
Uncertainty-Aware Predictive Energy Management for Computer Networks
Niranjana Sree, G., Madhumitha, S., Mohandas, R., et al. (2026) · doi:10.1109/icesc70182.2026.11687602
“Unlike the reactive control scheme that reacts to the current traffic conditions, the proposed scheme predicts the potential traffic overload conditions and reacts to the conditions accordingly, similar to the SLA-aware optimization approaches proposed in [14] and [17].”
Yerra, P. V. K., & Cherukupalle, N. S. (2026) · International Conference on Electrical, Computer and Energy Technologies (ICECET) · doi:10.1109/icecet65726.2026.11633191
“Infrastructure placement adds a third lever [14].”
Bhat, S. A., & Premalatha, H. M. (2026) · International Conference on Sustainable Computing and Smart Systems (ICSCSS) · doi:10.1109/icscss69635.2026.11646091
“Alex et al. [18], investigated a deep Q-network-based strategy of virtual machine placement that includes energy efficiency, carbon emissions, and SLAs in its scope.”
Farha, S., Venkatesh, K., & Madhavan, J. (2026) · International Conference on Integrated Intelligence and Communication Systems (ICIICS) · doi:10.1109/iciics67880.2026.11483498
Lunis, A., Bhuta, M., & Reddy, M. (2026) · doi:10.1109/otcon69332.2026.11629782
Alex, M., & Ojo, S. O. (2026) · Computers · doi:10.3390/computers15060339 · Cited by 1
“These strategies facilitate systems in acquiring optimal policies via interaction with the environment, hence adjusting to fluctuating workloads and system states [33,34].”
Self-citationAl-Shammaa, A. A., & Al-Hadrawi, D. H. (2026) · Sustainable Computing Informatics and Systems · doi:10.1016/j.suscom.2026.101398
Energy-Aware Load Balancing in Cloud Data Centres: A Carbon-Efficient Approach
Rajman, R., Kaur, K., & Bhuyan, A. S. (2026) · doi:10.1109/ccge67142.2026.11581781
AI-Based Carbon-Aware Scheduling for Predicting and Improving Cloud Carbon Footprint
Srishwal, D., Payasam, M. P. G., Ushasukhanya, S., & Malleswari, T. Y. J. N. (2026) · International Conference on Networking and Communications (ICNWC) · doi:10.1109/icnwc68145.2026.11518339
“A new carbon-aware and SLA-compliant VM placement approach using deep Q-networks is presented in [7].”
Simon, J., & Kapileswar, N. (2026) · doi:10.1109/iciccs67901.2026.11502729
“The article in [10] provides an excellent summary of the intersection of artificial intelligence and sustainability with particular emphasis on energy consumption and carbon footprint as the critical issues in the entire process of AI lifecycle.”
Afrianto, Y., Munadi, R., Setyorini, S., & Widyanto, T. (2026) · Array · doi:10.2139/ssrn.6439498
GUDIPATI, S. K., MISHRA, N. A., & ANKAM, G. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-26211-0_18
Hachmi, F., & Riahi, M. (2025) · Computing, Communications and IoT Applications (ComComAp) · doi:10.1109/comcomap68359.2025.11353194 · Cited by 1
“Also, (9) proposes CARBON-DQN, a novel virtual machine placement algorithm that balances sustainability, SLA compliance, and energy efficiency.”
Gopal, K., Vijaybabu, D., Deepa, M., Poonguzhali, M., Poornima, G., & Mohanraj, B. (2025) · International Conference on Emerging Research in Computational Science (ICERCS) · doi:10.1109/icercs65898.2025.11580884
“Cao et al. [2] conducted a general survey on architecture design, carbon measurement model and techniques of energy management in carbon-neutral data centers The study by Alex et al. [3] presented deep reinforcement learning to place a SLA-compliant virtual machine in the case of clustering and DQN.”
Reduction of energy consumption in cloud data centers with proper placement of virtual machines
Naji, H. R., & Esmaeili, R. (2025) · The Journal of Supercomputing · doi:10.1007/s11227-025-08072-2
Dubey, A., Sharma, S., & Kumar, N. (2025) · doi:10.1109/icosec67334.2025.11459486
“Similarly, VM placement research advances either energy efficiency or SLA compliance but rarely integrates carbon-awareness with deep reinforcement learning [9].”
C–SLO: Carbon–and SLO–Aware Global Load Balancing for Online ServicesPreprint
Raja, R., Sarddar, D., Naaz, S., & Das, S. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5788950
Metadata
- Authors
- Alex, M., Ojo, S. O., & Awuor, F. M.
- Affiliations
- Durban University of Technology (ZA)
- Computing Sciences, School of Information Science & Technology, Kisii University, P.O. Box 408, Kisii 40200, Kenya
- Corresponding author
- Alex, Maraga
- Venue
- Computers
- Publisher
- MDPI AG
- Published
- 2025-07-15
- Volume
- 14
- Issue
- 7
- Pages
- 280
- ISSN
- 2073-431X (ISSN-L)
- OpenAlex
- W4412428793
- Cited by
- 22
- References
- 47
- Citations by year
- 2026: 17 · 2025: 5
- FWCI
- 23.07
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Computers (publishedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Durban University of Technology
- Innovate for African Universities
- Topics
- Physical Sciences › Computer Science › Information Systems › Cloud Computing and Resource Management
- Physical Sciences › Computer Science › Hardware and Architecture › Energy Efficiency in Computing
- Physical Sciences › Computer Science › Computer Networks and Communications › Software-Defined Networks and 5G
- Keywords
- virtual machine placement, cloud data centers, CO2 emissions, energy efficiency, service level agreements, deep Q-network, agglomerative clustering, sustainable resource management, deep reinforcement learning, SLA violation, virtual machine clustering
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-06, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-06 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.