Papers
Topics
Authors
Recent
Search
2000 character limit reached

AI/ML-based Load Prediction in IEEE 802.11 Enterprise Networks

Published 11 Oct 2023 in cs.NI, cs.AI, and eess.SP | (2310.07467v1)

Abstract: Enterprise Wi-Fi networks can greatly benefit from Artificial Intelligence and Machine Learning (AI/ML) thanks to their well-developed management and operation capabilities. At the same time, AI/ML-based traffic/load prediction is one of the most appealing data-driven solutions to improve the Wi-Fi experience, either through the enablement of autonomous operation or by boosting troubleshooting with forecasted network utilization. In this paper, we study the suitability and feasibility of adopting AI/ML-based load prediction in practical enterprise Wi-Fi networks. While leveraging AI/ML solutions can potentially contribute to optimizing Wi-Fi networks in terms of energy efficiency, performance, and reliability, their effective adoption is constrained to aspects like data availability and quality, computational capabilities, and energy consumption. Our results show that hardware-constrained AI/ML models can potentially predict network load with less than 20% average error and 3% 85th-percentile error, which constitutes a suitable input for proactively driving Wi-Fi network optimization.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (24)
  1. L. Galati Giordano, G. Geraci, M. Carrascosa, and B. Bellalta, “What will Wi-Fi 8 be? A primer on IEEE 802.11 bn ultra high reliability,” arXiv e-prints, pp. arXiv–2303, 2023.
  2. B. M. Khorsandi, M. Hoffmann, M. Uusitalo, et al., “Hexa-X deliverable D1. 3: Targets and requirements for 6G-initial E2E architecture,” Online: http://hexax. eu/deliverables, 2022.
  3. S. Troia, R. Alvizu, Y. Zhou, G. Maier, and A. Pattavina, “Deep learning-based traffic prediction for network optimization,” in 2018 20th International Conference on Transparent Optical Networks (ICTON), pp. 1–4, IEEE, 2018.
  4. IEEE 802.11 Working Group (WG), “11-22/0597r3: 802.11 May 2022 WG Motions,” 2022.
  5. Wi-Fi Alliance, “ Wi-Fi CERTIFIED Data Elements™ (Version 2.1),” 2022.
  6. D. A. Tedjopurnomo, Z. Bao, B. Zheng, F. M. Choudhury, and A. K. Qin, “A survey on modern deep neural network for traffic prediction: Trends, methods and challenges,” IEEE Transactions on Knowledge and Data Engineering, vol. 34, no. 4, pp. 1544–1561, 2020.
  7. M. Abbasi, A. Shahraki, and A. Taherkordi, “Deep learning for network traffic monitoring and analysis (NTMA): A survey,” Computer Communications, vol. 170, pp. 19–41, 2021.
  8. W. Chen, F. Lyu, F. Wu, P. Yang, and J. Ren, “Flag: Flexible, accurate, and long-time user load prediction in large-scale WiFi system using deep RNN,” IEEE Internet of Things Journal, vol. 8, no. 22, pp. 16510–16521, 2021.
  9. Springer Science & Business Media, 2012.
  10. C. A. Hernández Suarez, O. J. Salcedo Parra, and A. Escobar Díaz, “An ARIMA model for forecasting Wi-Fi data network traffic values,” Ingeniería e Investigación, vol. 29, no. 2, pp. 65–69, 2009.
  11. Y. Jin, N. Duffield, A. Gerber, P. Haffner, W.-L. Hsu, G. Jacobson, S. Sen, S. Venkataraman, and Z.-L. Zhang, “Characterizing data usage patterns in a large cellular network,” in Proceedings of the 2012 ACM SIGCOMM workshop on Cellular networks: operations, challenges, and future design, pp. 7–12, 2012.
  12. F. Wilhelmi, S. Barrachina-Muñoz, B. Bellalta, C. Cano, A. Jonsson, and V. Ram, “A flexible machine-learning-aware architecture for future WLANs,” IEEE Communications Magazine, vol. 58, no. 3, pp. 25–31, 2020.
  13. S. Szott, K. Kosek-Szott, P. Gawłowicz, J. T. Gómez, B. Bellalta, A. Zubow, and F. Dressler, “Wi-Fi meets ML: A survey on improving IEEE 802.11 performance with machine learning,” IEEE Communications Surveys & Tutorials, vol. 24, no. 3, pp. 1843–1893, 2022.
  14. H. Feng, Y. Shu, S. Wang, and M. Ma, “SVM-based models for predicting WLAN traffic,” in 2006 IEEE international conference on communications, vol. 2, pp. 597–602, IEEE, 2006.
  15. A. Thapaliya, J. Schnebly, and S. Sengupta, “Predicting congestion level in wireless networks using an integrated approach of supervised and unsupervised learning,” in 2018 9th IEEE Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON), pp. 977–982, IEEE, 2018.
  16. M. A. Khan, R. Hamila, N. A. Al-Emadi, S. Kiranyaz, and M. Gabbouj, “Real-time throughput prediction for cognitive Wi-Fi networks,” Journal of Network and Computer Applications, vol. 150, p. 102499, 2020.
  17. W. Jiang, “Cellular traffic prediction with machine learning: A survey,” Expert Systems with Applications, vol. 201, p. 117163, 2022.
  18. H. D. Trinh, L. Giupponi, and P. Dini, “Mobile traffic prediction from raw data using lstm networks,” in 2018 IEEE 29th annual international symposium on personal, indoor and mobile radio communications (PIMRC), pp. 1827–1832, IEEE, 2018.
  19. S. Albawi, T. A. Mohammed, and S. Al-Zawi, “Understanding of a convolutional neural network,” in 2017 international conference on engineering and technology (ICET), pp. 1–6, Ieee, 2017.
  20. D. Zhang, L. Liu, C. Xie, B. Yang, and Q. Liu, “Citywide cellular traffic prediction based on a hybrid spatiotemporal network,” Algorithms, vol. 13, no. 1, p. 20, 2020.
  21. J. Feng, X. Chen, R. Gao, M. Zeng, and Y. Li, “Deeptp: An end-to-end neural network for mobile cellular traffic prediction,” IEEE Network, vol. 32, no. 6, pp. 108–115, 2018.
  22. V. Perifanis, N. Pavlidis, S. F. Yilmaz, F. Wilhelmi, E. Guerra, M. Miozzo, P. S. Efraimidis, P. Dini, and R.-A. Koutsiamanis, “Towards Energy-Aware Federated Traffic Prediction for Cellular Networks,” in 1st International Symposium on Federated Learning Technologies and Applications (FLTA), 2023.
  23. S. A. Budennyy, V. D. Lazarev, N. N. Zakharenko, A. N. Korovin, O. Plosskaya, D. V. Dimitrov, V. Akhripkin, I. Pavlov, I. V. Oseledets, I. S. Barsola, et al., “Eco2ai: carbon emissions tracking of machine learning models as the first step towards sustainable AI,” in Doklady Mathematics, vol. 106, pp. S118–S128, Springer, 2022.
  24. Y. Nahshan, B. Chmiel, C. Baskin, E. Zheltonozhskii, R. Banner, A. M. Bronstein, and A. Mendelson, “Loss aware post-training quantization,” Machine Learning, vol. 110, no. 11-12, pp. 3245–3262, 2021.
Citations (1)
View on Semantic Scholar

Summary

  • The paper demonstrates that AI/ML models can predict network load with an average error below 20%, even under hardware constraints.
  • It employs practical metrics such as MAE, MAPE, and 85th-percentile error while accounting for data availability and energy efficiency.
  • The findings support proactive network management through dynamic adjustments and early detection of performance bottlenecks.

AI/ML Load Prediction in Enterprise Networks

This paper explores the application of AI/ML techniques for load prediction in IEEE 802.11 enterprise Wi-Fi networks, addressing the potential for enhanced network management, autonomous operation, and proactive troubleshooting through forecasted network utilization. The authors investigate the feasibility and suitability of AI/ML-based load prediction, considering constraints related to data availability, computational resources, and energy consumption within practical enterprise Wi-Fi deployments. The findings suggest that even with hardware limitations, AI/ML models can achieve satisfactory load prediction accuracy, offering a valuable tool for optimizing Wi-Fi network operation.

Background and Motivation

The increasing complexity and demands on enterprise Wi-Fi networks necessitate intelligent solutions for resource management and performance optimization. AI/ML-based traffic and load prediction offers a data-driven approach to anticipate network demands, enabling proactive adjustments that improve user experience, energy efficiency, and overall network reliability. This paper addresses the practical challenges of implementing AI/ML solutions in enterprise Wi-Fi environments, specifically focusing on the trade-offs between prediction accuracy and the computational resources required for deployment.

Methodology and Model Selection

The research focuses on the utilization of hardware-constrained AI/ML models to ensure feasibility in resource-limited enterprise network devices. The paper considers factors such as data availability, data quality, computational capabilities of network devices, and energy consumption when selecting appropriate models. The performance metrics considered include \gls{mae}, \gls{mape}, and 85th-percentile error. The paper emphasizes the importance of these metrics for evaluating the suitability of load prediction models in practical Wi-Fi network optimization scenarios.

Key Findings and Results

The evaluation demonstrates that AI/ML models can predict network load with reasonable accuracy even under hardware constraints. Specifically, the results indicate that the models can achieve an average prediction error of less than 20% and an 85th-percentile error of 3%. These results suggest that AI/ML-based load prediction can provide a practical and reliable input for proactively optimizing Wi-Fi network configurations and resource allocation.

Implications and Future Directions

The successful implementation of AI/ML-based load prediction in enterprise Wi-Fi networks has several implications. It enables proactive network management, allowing for dynamic adjustments to optimize performance and energy efficiency. Furthermore, accurate load prediction can facilitate better troubleshooting by identifying potential bottlenecks and anomalies before they impact user experience. Future research directions include exploring more advanced AI/ML models, investigating the use of federated learning to improve data availability and model generalization, and developing adaptive algorithms that can dynamically adjust model complexity based on available resources and network conditions.

Conclusion

This paper makes a compelling case for the adoption of AI/ML-based load prediction in enterprise Wi-Fi networks. The results demonstrate the feasibility of achieving satisfactory prediction accuracy with hardware-constrained models, paving the way for proactive network optimization and improved user experience. The study contributes valuable insights into the practical considerations and trade-offs involved in deploying AI/ML solutions in real-world Wi-Fi environments, highlighting the potential for future advancements in this area.

Paper to Video (Beta)

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Sign Up to Generate

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Generate Now

Collections

Sign up for free to add this paper to one or more collections.

User Circle Single Streamline Icon: https://streamlinehq.com Sign Up