Fog Alliances: A Decentralized Cloud Structure with Federated Machine Learning for Citiverses

Relevance. In recent years, Intelligent Transportation Systems (ITS) have played a pivotal role in urban management and the reduction of transportation risks, having become a key infrastructure technology for future urban network universes (Citiverse). A key challenge is enhancing system efficiency through accurate, real-time traffic prediction. Traditional centralized deep learning models suffer from network propagation delays and vulnerability of the central server in terms of both security and computational overload.

Objective. To develop a decentralized cloud framework based on dynamic fog computing and federated machine learning for traffic prediction in ITS, eliminating dependency on a central server and ensuring system fault tolerance through the proposed architecture.

Methods. The study employs literature analysis in the subject area, mathematical modeling, and computational experiments for performance evaluation. The proposed cloud framework integrates three technologies: Decentralized Federated Learning, Fog Computing, and Adaptive Graph Convolutional Recurrent Networks.

Results. The proposed framework operates effectively without a central server. Experiments on the real-world datasets PeMSD4 and PeMSD7(M) show that the model reduces communication overhead by approximately 48% compared to traditional FL methods. Convergence speed is significantly faster (a 17.8% reduction in the loss function during initial training rounds), while prediction accuracy remains at a competitive level compared to models relying on a central server. A novel decentralized system architecture is proposed that eliminates the central server while maintaining a balance among prediction accuracy, model efficiency, and network resource consumption.

Theoretical Significance. The study confirms the theoretical validity of integrating dynamic fog computing with decentralized federated learning. Implementing this approach using AGCRN at the fog node level enables accurate modeling of complex spatio-temporal dependencies, while eliminating the need for raw data transmission and central server involvement.

Practical Significance. Experimental results on real-world datasets confirm the feasibility of deploying the proposed solution in large-scale ITS within smart cities. The solution is particularly effective under conditions of limited network bandwidth, connectivity disruptions with the cloud, and overload of the central server.

1. Jiang W., Luo J. Graph neural network for traffic forecasting: A survey. Expert Systems with Applications. 2022;207:117921. DOI:10.1016/j.eswa.2022.117921. EDN:MOJUQL

2. Zanella A., Bui N., Castellani A., Vangelista L., Zorzi M. Internet of Things for Smart Cities. IEEE Internet of Things Journal. 2012;1(1):22–32. DOI:10.1109/JIOT.2014.2306328

3. Caterini A.L., Chang D.E. Recurrent Neural Networks. In: Deep Neural Networks in a Mathematical Framework. SpringerBriefs in Computer Science. Cham: Springer; 2018. p.59–79. DOI:10.1007/978-3-319-75304-1_5

4. Ateya A.A., Abd El-Latif A.A., Muthanna A., Volkov A., Koucheryavy A. Enabling Metaverse and Telepresence Services in 6G Networks. 2025. DOI:10.1201/9788770046749

5. Qi T., Chen L., Li G., Li Y., Wang C. FedAGCN: A traffic flow prediction framework based on federated learning and Asynchronous Graph Convolutional Network. Applied Soft Computing. 2023;138:110175. DOI:10.1016/j.asoc.2023.110175. EDN:XAIBCN

6. Yuan X., Chen J., Yang J., Zhang N., Yang T., Han T. FedSTN: Graph Representation Driven Federated Learning for Edge Computing Enabled Urban Traffic Flow Prediction. IEEE Transactions on Intelligent Transportation Systems. 2023;24(8):8738–8748. DOI:10.1109/TITS.2022.3157056. EDN:YKQIZO

7. Zhao L., Song Y., Zhang C., Liu Y., Wang P., Lin T. T-GCN: A Temporal Graph Convolutional Network for Traffic Prediction. IEEE Transactions on Intelligent Transportation Systems. 2020;21(9):3848–3858. DOI:10.1109/TITS.2019.2935152. EDN:EDVHGA

8. Zhang C. Zhang S., Yu J.J.Q., Yu S. FASTGNN: A Topological Information Protected Federated Learning Approach for Traffic Speed Forecasting. IEEE Transactions on Industrial Informatics. 2021;17(12):8464–8474. DOI:10.1109/TII.2021.3055283

9. Feng Z., Zha H., Xu C., He Y., Lin Y. FCGCN: Feature Correlation Graph Convolution Network for Few-Shot Individual Identification. IEEE Transactions on Consumer Electronics. 2024;70(1):2848–2860. DOI:10.1109/TCE.2023.3322224

10. McMahan B., Moore E., Ramage D., Hampson S., A. y Arcas B. Communication-Efficient Learning of Deep Networks from Decentralized Data. Proceedings of the 20th International Conference on Artificial Intelligence and Statistics, PMLR, vol.54. 2017. p.1273–1282. URL: https://proceedings.mlr.press/v54/mcmahan17a.html [Accessed 29.09.2025]

11. Giannakis G.B., Ling Q., Mateos G., Schizas I.D., Zhu H. Decentralized Learning for Wireless Communications and Networking. In: Glowinski R., Osher S.J., Yin W. (eds.) Splitting Methods in Communication, Imaging, Science, and Engineering. Scientific Computation. Cham: Springer; 2016. p.461–497. DOI:10.1007/978-3-319-41589-5_14

12. Bai L., Yao L., Li C., Wang X., Wang C. Adaptive Graph Convolutional Recurrent Network for Traffic Forecasting. Advances in Neural Information Processing Systems. 2020;33:17804–17815.

13. Liu Y., Yu J.J.Q., Kang J., Niyato D., Zhang S. Privacy-Preserving Traffic Flow Prediction: A Federated Learning Approach. IEEE Internet of Things Journal. 2020;7(8):7751–7763. DOI:10.1109/JIOT.2020.2991401. EDN:MKJJWJ

14. Yu B., Yin H., Zhu Z. Spatio-Temporal Graph Convolutional Networks: A Deep Learning Framework for Traffic Forecasting. arXiv:1709.04875v4. 2018. DOI:10.24963/ijcai.2018/505

15. Liu Y., Yu J.J.Q., Kang J., Niyato D., Zhang S. Privacy-Preserving Traffic Flow Prediction: A Federated Learning Approach. IEEE Internet of Things Journal. 2020;7(8):7751–7763. DOI:10.1109/JIOT.2020.299140

16. Zhang H., Jiang S., Xuan S. Decentralized federated learning based on blockchain: concepts, framework, and challenges. Computer Communications. 2024;216:140–150. DOI:10.1016/j.comcom.2023.12.042. EDN:NYTCWY

17. Lian X., Zhang C., Zhang H., Hsieh C.-J., Zhang W., Liu J. Can Decentralized Algorithms Outperform Centralized Algorithms? A Case Study for Decentralized Parallel Stochastic Gradient Descent. arXiv:1705.09056v5. 2017. DOI:10.48550/arXiv.1705.09056

18. Turazza F., Pietri M., Picone M., Mamei M. FedBGS: A Blockchain Approach to Segment Gossip Learning in Decentralized Systems. Proceedings of the 45th International Conference on Distributed Computing Systems Workshops, ICDCSW, 21–23 July 2025, Glasgow, United Kingdom. IEEE; 2025. p.760–770. DOI:10.1109/ICDCSW63273.2025.00136

19. Yu E., Dong D., Xu Y., Ouyang S., Liao X. CD-SGD: Distributed Stochastic Gradient Descent with Compression and Delay Compensation. arXiv:2106.10796v2. 2021. DOI:10.48550/arXiv.2106.10796

20. Stewart G., Al-Khassaweneh M. An Implementation of the HDBSCAN* Clustering Algorithm. Applied Sciences. 2022;12(5):2405. DOI:10.3390/app12052405

21. Chen C., Petty K., Skabardonis A. Freeway Performance Measurement System: Mining Loop Detector Data. Transportation Research Record: Journal of the Transportation Research Board. 2001;1748(1). DOI:10.3141/1748-12

22. Sharma M., Jaggi P. Reliable federated learning in a cloud-fog-IoT environment. The Journal of Supercomputing. 2023;79:15435–15458. DOI:10.1007/s11227-023-05252-w. EDN:EETJGC

23. Manju B., Kumar C.S.P., Jegan J., Jagadeeshan D., Nunna S.K. Hierarchical Federated Learning with Fog Nodes: En-hancing Efficiency in Smart City Networks. Proceedings of the International Conference on Information Technology, OCIT, 12–14 December 2024, Vijayawada, India. IEEE; 2024. p.749–753. DOI:10.1109/OCIT65031.2024.00135

24. Lähderanta T., Lovén L., Ruha L., Leppänen T., Launonen I., Riekki J., et al. Capacitated spatial clustering with multiple constraints and attributes. Engineering Applications of Artificial Intelligence. 2024;127(Part A):107182. DOI:10.1016/j.engappai.2023.107182. EDN:CXFPOY

25. Hu J., Liu K. Raft consensus mechanism and the applications. Journal of Physics: Conference Series. 2020;1544(1):012079. DOI:10.1088/1742-6596/1544/1/012079. EDN:IJJLUI

26. Performance evaluation of IEEE 802.11ah systems. Proceedings of the 24th Telecommunications Forum, TELFOR, 22–23 November 2016, Belgrade, Serbia. IEEE; 2016. DOI:10.1109/TELFOR.2016.7818748