SavedAdd your journal

Journals / Erciyes Üniversitesi Fen Bilimleri Enstitüsü Fen Bilimleri Dergisi / 2026 / Cilt: 42 Sayı: 3

Energy Consumption Prediction for 5G Base Stations: A Comparative Machine Learning Study under Hardware Heterogeneity

5G Baz İstasyonları için Enerji Tüketimi Tahmini: Donanım Heterojenliği Altında Karşılaştırmalı Makine Öğrenimi Çalışması

Ramazan Enisoğlu(KIRIKKALE ÜNİVERSİTESİ, KIRIKKALE MESLEK YÜKSEKOKULU)0009-0009-5686-6798
Burak Alp İnan(HACETTEPE UNIVERSITY)0009-0003-0462-9607
Pages
—
Publication Date
2026-10-21
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

The study also documents a sparsity-induced instability in linear models and the hardware distribution shift between training and evaluation data.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Background/Aim: Energy consumption of 5G base stations has become a central operational concern, and energy-saving mechanisms such as carrier shutdown and sleep modes require accurate prediction of station-level consumption.
📊
Key Findings & Data:Methods: Seven models – from a trivial baseline and linear models to tree ensembles and a multilayer perceptron – were evaluated on the ITU 5G Base Station Energy Consumption Modelling dataset (98,084 hourly observations from 923 base stations with heterogeneous hardware) using station-grouped five-fold cross-validation, station-clustered bootstrap intervals, repeated station partitions and out-of-time tests.
💡
Conclusion & Impact:The study also documents a sparsity-induced instability in linear models and the hardware distribution shift between training and evaluation data.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Enerji Tüketimi · Makine Öğrenmesi · Baz İstasyonu

Abstract

Background/Aim: Energy consumption of 5G base stations has become a central operational concern, and energy-saving mechanisms such as carrier shutdown and sleep modes require accurate prediction of station-level consumption. This study compares prediction models under conditions reflecting deployment to unseen base stations. Methods: Seven models – from a trivial baseline and linear models to tree ensembles and a multilayer perceptron – were evaluated on the ITU 5G Base Station Energy Consumption Modelling dataset (98,084 hourly observations from 923 base stations with heterogeneous hardware) using station-grouped five-fold cross-validation, station-clustered bootstrap intervals, repeated station partitions and out-of-time tests. An analytical load-dependent power model and four published data-driven approaches were re-implemented under the identical protocol. Results: LightGBM attained the best accuracy among the seven models (R² = 0.910, WMAPE = 10.22%), matched in R² by the multilayer perceptron at roughly twenty times the training cost. LightGBM remained within 0.15 percentage points of the most accurate re-implemented technique, whereas the analytical model was clearly less accurate; lower published errors (WMAPE of about 4–5%) arise only when test stations were seen during training. Consumption was governed predominantly by static hardware attributes and their interaction with traffic load. Accuracy was stable across the observed days, but tree ensembles underestimated consumption when load exceeded the training range. Conclusion: Gradient boosting offers the most favourable accuracy–cost trade-off for unseen base stations, while seasonal use requires periodic recalibration. The study also documents a sparsity-induced instability in linear models and the hardware distribution shift between training and evaluation data.

Özet

5G baz istasyonlarının enerji tüketimi temel bir işletim sorunu hâline gelmiştir ve taşıyıcı kapatma ile uyku modları gibi enerji tasarrufu mekanizmaları, istasyon düzeyinde tüketimin doğru tahmin edilmesini gerektirmektedir. Bu çalışmada, basit bir temel modelden doğrusal modellere, ağaç tabanlı topluluk yöntemlerinden çok katmanlı algılayıcıya kadar yedi model, donanım açısından heterojen 923 baz istasyonundan elde edilen 98.084 saatlik gözlemden oluşan ITU 5G Baz İstasyonu Enerji Tüketimi Modelleme veri kümesi üzerinde karşılaştırılmıştır. Değerlendirmede istasyon gruplandırmalı beş katlı çapraz doğrulama, istasyon kümeli bootstrap güven aralıkları, farklı istasyon bölmeleri ve zaman dışı testler kullanılmış; analitik yük bağımlı bir güç modeli ile literatürdeki dört veri odaklı yaklaşım aynı protokol altında yeniden uygulanmıştır. LightGBM yedi model arasında en iyi doğruluğa ulaşmış (R² = 0.910, WMAPE = %10.22), çok katmanlı algılayıcı ise R² açısından aynı seviyeyi yaklaşık yirmi kat daha yüksek eğitim maliyetiyle yakalamıştır. LightGBM, yeniden uygulanan en doğru tekniğe 0.15 yüzde puan mesafede kalmış, analitik model ise belirgin biçimde daha az doğru bulunmuştur; literatürdeki daha düşük hatalar (WMAPE yaklaşık %4–5) yalnızca test istasyonları eğitimde görüldüğünde ortaya çıkmaktadır. Enerji tüketimi ağırlıklı olarak sabit donanım özellikleri ve bunların trafik yüküyle etkileşimi tarafından belirlenmektedir. Doğruluk gözlenen günler boyunca kararlı kalmış, ancak yük eğitim aralığını aştığında ağaç tabanlı modeller tüketimi düşük tahmin etmiştir. Sonuç olarak gradient boosting, görülmemiş baz istasyonları için en uygun doğruluk–maliyet dengesini sunmakta, mevsimler boyunca kullanım ise periyodik yeniden kalibrasyon gerektirmektedir. Çalışma ayrıca doğrusal modellerde seyreklik kaynaklı kararsızlığı ve eğitim ile değerlendirme verileri arasındaki donanım dağılımı değişimini ortaya koymaktadır.

🔖

Cite This Article

Enisoğlu, R., İnan, B. A. (2026). Energy Consumption Prediction for 5G Base Stations: A Comparative Machine Learning Study under Hardware Heterogeneity. Erciyes Üniversitesi Fen Bilimleri Enstitüsü Fen Bilimleri Dergisi, Cilt: 42 Sayı: 3. https://doi.org/10.65520/erciyesfen.1968962

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (37)

  1. Fowdur, T. P., Doorgakant, B. (2023). A review of machine learning techniques for enhanced energy efficient 5G and 6G communications. Engineering Applications of Artificial Intelligence, 122, 106032.
  2. Piovesan, N., López-Pérez, D., De Domenico, A., Geng, X., Bao, H., Debbah, M. (2022). Machine learning and analytical power consumption models for 5G base stations. IEEE Communications Magazine, 60, 56–62.
  3. International Telecommunication Union (2023). AI/ML for 5G-Energy Consumption Modelling by ITU AI/ML in 5G Challenge. ITU AI for Good. Available: https://zindi.africa/competitions/aiml-for-5g-energy-consumption-modelling [Accessed: 8 October 2026].
  4. Chen, T., Wang, Y., Chen, H., Zhao, Z., Li, X., Piovesan, N., Zhu, G., Shi, Q. (2024). Modelling the 5G energy consumption using real-world data: Energy fingerprint is all you need. arXiv preprint arXiv:2406.16929.
  5. Thantharate, A., Tondwalkar, A. V., Beard, C., Kwasinski, A. (2022). ECO6G: Energy and cost analysis for network slicing deployment in beyond 5G networks. Sensors, 22, 8614.
  6. Makroz, I., Nafidi, A., Gutiérrez-Sánchez, R., Aourik, S. (2024). A machine learning approach for energy consumption in 5G networks. ITU Journal on Future and Evolving Technologies, 5(4), 447–457.
  7. Grinsztajn, L., Oyallon, E., Varoquaux, G. (2022). Why do tree-based models still outperform deep learning on typical tabular data? Advances in Neural Information Processing Systems, 35, 507–520.
  8. Breiman, L. (2001). Random forests. Machine Learning, 45, 5–32.
  9. Friedman, J. H. (2001). Greedy function approximation: a gradient boosting machine. Annals of Statistics, 29, 1189–1232.
  10. Chen, T., Guestrin, C. (2016). XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785–794.
  11. Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., Ye, Q., Liu, T.-Y. (2017). LightGBM: A highly efficient gradient boosting decision tree. Advances in Neural Information Processing Systems, 30.
  12. Arik, S. Ö., Pfister, T. (2021). TabNet: Attentive interpretable tabular learning. Proceedings of the AAAI Conference on Artificial Intelligence, 35, 6679–6687.
  13. Gorishniy, Y., Rubachev, I., Khrulkov, V., Babenko, A. (2021). Revisiting deep learning models for tabular data. Advances in Neural Information Processing Systems, 34, 18932–18943.
  14. Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., Duchesnay, E. (2011). Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 12, 2825–2830.
  15. Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., Desmaison, A., Köpf, A., Yang, E., DeVito, Z., Raison, M., Tejani, A., Chilamkurthy, S., Steiner, B., Fang, L., Bai, J., Chintala, S. (2019). PyTorch: An imperative style, high-performance deep learning library. Advances in Neural Information Processing Systems, 32, 8024–8035.
  16. Ioffe, S., Szegedy, C. (2015). Batch normalization: accelerating deep network training by reducing internal covariate shift. International Conference on Machine Learning (ICML), 448–456.
  17. Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., Salakhutdinov, R. (2014). Dropout: a simple way to prevent neural networks from overfitting. Journal of Machine Learning Research, 15, 1929–1958.
  18. Kingma, D. P., Ba, J. (2015). Adam: A method for stochastic optimization. International Conference on Learning Representations (ICLR).
  19. Hollmann, N., Müller, S., Eggensperger, K., Hutter, F. (2023). TabPFN: A transformer that solves small tabular classification problems in a second. International Conference on Learning Representations (ICLR).
  20. López-Pérez, D., De Domenico, A., Piovesan, N., Xinli, G., Bao, H., Qitao, S., Debbah, M. (2022). A survey on 5G radio access network energy efficiency: Massive MIMO, lean carrier design, sleep modes, and machine learning. IEEE Communications Surveys & Tutorials, 24(1), 653–697.
  21. Zimmermann, R. S. (2024). Hybrid-boosted model with an approach inspired by a mixture of experts for 5G energy consumption. ITU Journal on Future and Evolving Technologies, 5(4), 493–502.
  22. Kapoor, S., Narayanan, A. (2023). Leakage and the reproducibility crisis in machine-learning-based science. Patterns, 4(9), 100804.
  23. Auer, G., Giannini, V., Desset, C., Godor, I., Skillermark, P., Olsson, M., Imran, M. A., Sabella, D., Gonzalez, M. J., Blume, O., Fehske, A. (2011). How much energy is needed to run a wireless network? IEEE Wireless Communications, 18(5), 40–49.
  24. Debaillie, B., Desset, C., Louagie, F. (2015). A flexible and future-proof power model for cellular base stations. IEEE 81st Vehicular Technology Conference (VTC Spring), 1–7.
  25. Piovesan, N., López-Pérez, D., De Domenico, A., Geng, X., Bao, H. (2023). Power consumption modeling of 5G multi-carrier base stations: A machine learning approach. IEEE International Conference on Communications (ICC), 3633–3638.
  26. Lorincz, J., Garma, T., Petrović, G. (2012). Measurements and modelling of base station power consumption under real traffic loads. Sensors, 12(4), 4281–4310.
  27. Spagnuolo, A., Petraglia, A., Vetromile, C., Formosi, R., Lubritto, C. (2015). Monitoring and optimization of energy consumption of base transceiver stations. Energy, 81, 286–293.
  28. Prokhorenkova, L., Gusev, G., Vorobev, A., Dorogush, A. V., Gulin, A. (2018). CatBoost: unbiased boosting with categorical features. Advances in Neural Information Processing Systems, 31.
  29. Shwartz-Ziv, R., Armon, A. (2022). Tabular data: Deep learning is not all you need. Information Fusion, 81, 84–90.
  30. ITU AI/ML in 5G Challenge (2024). 5G base station energy consumption dataset (BSinfo, CLdata, ECdata, estimation and prediction files), archived snapshot. Software Heritage persistent identifier swh:1:cnt:c529efd4080f5b7302af1fe7ec2d343dcbfe6e89 (revision swh:1:rev:0e5e768d1a58f665bb3121ce4cb83080990d6a78). Available: https://archive.softwareheritage.org/browse/revision/0e5e768d1a58f665bb3121ce4cb83080990d6a78/
  31. Roberts, D. R., Bahn, V., Ciuti, S., Boyce, M. S., Elith, J., Guillera-Arroita, G., Hauenstein, S., Lahoz-Monfort, J. J., Schröder, B., Thuiller, W., Warton, D. I., Wintle, B. A., Hartig, F., Dormann, C. F. (2017). Cross-validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure. Ecography, 40(8), 913–929.
  32. Field, C. A., Welsh, A. H. (2007). Bootstrapping clustered data. Journal of the Royal Statistical Society Series B: Statistical Methodology, 69(3), 369–390.
  33. Demšar, J. (2006). Statistical comparisons of classifiers over multiple data sets. Journal of Machine Learning Research, 7, 1–30.
  34. Bergmeir, C., Benítez, J. M. (2012). On the use of cross-validation for time series predictor evaluation. Information Sciences, 191, 192–213.
  35. Xu, F., Lin, Y., Huang, J., Wu, D., Shi, H., Song, J., Li, Y. (2016). Big data driven mobile traffic understanding and forecasting: A time series approach. IEEE Transactions on Services Computing, 9(5), 796–805.
  36. Zhang, M., Fu, H., Li, Y., Chen, S. (2019). Understanding urban dynamics from massive mobile traffic data. IEEE Transactions on Big Data, 5(2), 266–278.
  37. Quiñonero-Candela, J., Sugiyama, M., Schwaighofer, A., Lawrence, N. D. (2009). Dataset Shift in Machine Learning. MIT Press, Cambridge, MA.