Сообщение

Federated Learning as a Method for Optimizing Service Quality in Heterogeneous Telepresence Networks

 
 orcid Tatyana Lapteva, orcid Ammar Muthanna

The Bonch-Bruevich Saint Petersburg State University of Telecommunications,
St. Petersburg, 193232, Russian Federation

DOI 10.31854/2307-1303-2026-14-2-25-36

EDN LPRJVF

 Full text

XML JATS

Abstract

Telepresence is a key service in sixth-generation networks, placing stringent quality of service requirements on them. Current and future communications infrastructures are characterized by a high degree of heterogeneity. Ensuring a stable level of quality of service is therefore a complex task requiring an integrated approach and innovative solutions. Traditional management methods, such as static prioritization or adaptive algorithms, lack sufficient flexibility, while centralized machine learning approaches are associated with privacy risks and high network load. The aim of this paper is to systematize and comparatively analyze federated learning methods used to predict and optimize quality of service in heterogeneous telepresence networks. The essence of the approach proposed in this paper lies in the author's classification of federated learning methods based on two criteria: architectural organization and the method of adaptation to heterogeneity. Detailed descriptions are provided for each class, and strengths and weaknesses are highlighted. The comparative analysis is based on comparing approaches based on key metrics: forecast accuracy, convergence rate, communication delays, privacy, device computational load, heterogeneity resilience, implementation complexity, and security. The analysis was based on data from peer-reviewed sources, which allowed us to identify quantitative advantages (e.g., a 1.91--38.89% reduction in training rounds and a 9.52--40.00% reduction in communication delays for the hierarchical MultiFed architecture). Experimental data presented in the analyzed papers confirm that federated learning provides a balance between predictive accuracy, privacy, and scalability, with the gains being most noticeable for devices with heterogeneous data. The scientific novelty lies in the proposed classification of federated learning methods and the systematization of their advantages and limitations as applied to the specifics of telepresence networks, which enables an informed choice of architecture depending on the use case. The authors' practical significance lies in their recommendations for selecting a federated learning method for various conditions (regional heterogeneity, limited device resources, and stringent latency requirements), which can be used in the design and operation of communication networks.

Keywords

federated learning, distributed learning, telepresence, QoS, heterogeneous networks, 6G, latency optimization, QoS prediction

Reference for citation

Lapteva T., Muthanna A. Federated Learning as a Method for Optimizing Service Quality in Heterogeneous Telepresence Networks // Telecom IT. 2026. Vol. 14. Iss. 2. PP. 25‒36. (in Russian). DOI: 10.31854/2307-1303-2026-14-2-25-36. EDN: LPRJVF
References

1. Taghia J., Moradi F., Larsson H., Lan X., Orucu A., Ebrahimi M., et al. Congruent Learning for Self-Regulated Federated Learning in 6G // IEEE Transactions on Machine Learning in Communications and Networking. 2024. Vol. 2. PP. 29--149. DOI: 10.1109/TMLCN.2023.3347680. EDN: VARJVJ

2. Temiz F., Salehi S., Erol-Kantarci M. Edge Learning via Federated Split Decision Transformers for Metaverse Resource Allocation // arXiv.org. 2026. URL: https://ar5iv.labs.arxiv.org/html/2602.16174 [Accessed 30.08.2026]

3. 3GPP TS 22.261 version 18.6.1. 3rd Generation Partnership Project. Technical Specification Group Services and System Aspects. Service requirements for the 5G system. Stage 1 (Release 18).

4. Kucheryavy A. E., Davud A. A., Volkov A. N., Muthanna A. S. A., Paramonov A. I., Elagin V. S., et al. Sixth-generation communication networks: fractal architectures, holographic interactions, telepresence, network robots. St. Petersburg; Moscow; Minsk: Piter Publ.; 2024. 320 p. (in Russian)

5. Al-Quraan M., Mohjazi L., Bariah L., Centeno A., Zoha A., Arshad K., et al. Edge-Native Intelligence for 6G Communications Driven by Federated Learning: A Survey of Trends and Challenges // IEEE Transactions on Emerging Topics in Computational Intelligence. 2023. Vol. 7. Iss. 3. PP. 957--979. DOI:10.1109/TETCI. 2023.3251404

6. Xu J., Lin J., Li Y., Xu Z. MultiFed: A fast converging federated learning framework for services QoS prediction via cloud--edge collaboration mechanism // Knowledge-Based Systems. 2023. Vol. 268. P. 110463. DOI: 10.1016/j.knosys.2023. 110463. EDN: ORNKNM

7. Setayesh M. Machine learning-based algorithms design for network slicing, federated learning, and 360° video streaming in wireless systems. PhD Thesis. Vancouver: University of British Columbia, 2024. DOI: 10.14288/1.0441439

8. Bragato F., Giordani M., Zorzi M. Federated Reinforcement Learning to Optimize Teleoperated Driving Networks. 2024. DOI: 10.48550/arXiv. 2410.02312

9. Ossongo E., Esseghir M., Merghem-Boulahia L. A multi-agent federated reinforcement learning-based optimization of quality of service in various LoRa network slices // Computer Communications. 2024. Vol. 213. PP. 320--330. DOI: 10.1016/j. comcom.2023.11.015. EDN: OQGADB

10. Song X., Xu B., Zhang X., Wang S., Song T., Xing G., et al. Everyone-Centric Heterogeneous Multi-Server Computation Offloading in ITS with Pervasive AI // IEEE Network. 2023. Vol. 37. Iss. 2. PP. 62--68. DOI: 10.1109/MNET.2023. 10061628. EDN: QWGJYN

11. Li T., Sahu A. K., Zaheer M., Sanjabi M., Talwalkar A., Smith V. Federated Learning: Challenges, Methods, and Future Directions // IEEE Signal Processing Magazine. 2020. Vol. 37. Iss. 3. PP. 50--60. DOI: 10.1109/msp.2020.2975749. EDN: BFBNOW

 

cc-by Статья распространяется по лицензии Creative Commons Attribution 4.0 License.

cc0  Метаданные статьи распространяются по лицензии CC0 1.0 Universal

 
 
войти

Авторизация