Aprendizado Federado Hierárquico: Uma Perspectiva Analítica com Redes de Petri Estocásticas

• Israel Araújo UFPI
• Luís Guilherme Silva UFPI
• Francinaldo Barbosa UFPI
• Iure Fé UFPI
• Geraldo P. Rocha Filho UESB
• Francisco Airton Silva UFPI

DOI: https://doi.org/10.5753/sbrc.2026.19936

Resumo

Avaliar Aprendizado Federado Hierárquico (HFL) em ambientes densos é difícil devido a limitações de escala e à variabilidade de rede e recursos. Embora arquiteturas hierárquicas empreguem servidores intermediários para agregações parciais, analisar seu comportamento por simulação ou experimentação ainda é custoso para explorar múltiplas configurações. Este artigo propõe um modelo em Redes de Petri Estocásticas (SPN) para representar HFL com servidor central e servidores intermediários. O modelo estima métricas de taxa de conclusão de rodadas, tempo médio de rodada e probabilidade de descarte. Os resultados mostram que aumentar o número de servidores intermediários melhora a vazão, reduz o tempo médio de rodada e amplia a capacidade operacional sob cargas elevadas. Em um cenário avaliado, a expansão de dois para quatro servidores intermediários reduziu o tempo médio de rodada em mais de 60% e elevou a taxa de conclusão de rodadas em até 160%.

Referências

Araújo, I., Silva, L. G., Brito, C., Min, D., Lee, J.-W., Nguyen, T. A., Leão, E., and Silva, F. A. (2025). Dds-p: Stochastic models based performance of iot disaster detection systems across multiple geographic areas. ICT Express, 11(1):34–40.

Cho, S., Lim, S., and Lee, J. (2024). Drl-enabled hierarchical federated learning optimization for data heterogeneity management in multi-access edge computing. IEEE Access.

DP, M. K., NagaSireesha, J., Venkatesh, B., Siddiqua, A., et al. (2023). Hierarchical federated learning-based method for privacy-preserving in the healthcare environment. In 2023 International Conference on Evolutionary Algorithms and Soft Computing Techniques (EASCT), pages 1–5. IEEE.

Farajzadeh, A., Yadav, A., and Yanikomeroglu, H. (2025). Federated learning in ntn: Design, architecture, and challenges. IEEE Communications Magazine, 63(6):26–33.

Fé, I., Silva, L. G., Soares, A., Silva, F. A., Mei, A., Rego, P. A., Choi, E., Nguyen, T. A., Lee, J. W., and Min, D. (2024). Resilient and efficient microservices: Stochastic modeling and quantification of energy consumption and recovery times. In GLOBECOM 2024-2024 IEEE Global Communications Conference, pages 5307–5312. IEEE.

Kumari, A., Shukla, D., Datt, R., Patel, A., Kumar, P., Kumar, S., Katarmal, U., et al. (2024). S-evpf: A secure and decentralized ev prediction framework using federated learning. In 2024 Eighth International Conference on Parallel, Distributed and Grid Computing (PDGC), pages 658–663. IEEE.

Li, Z., Chen, Z., Wei, X., Gao, S., Ren, C., and Quek, T. Q. (2022). Hpfl-cn: Communication-efficient hierarchical personalized federated edge learning via complex network feature clustering. In 2022 19th Annual IEEE International Conference on Sensing, Communication, and Networking (SECON), pages 325–333. IEEE.

Liu, L., Zhang, J., Song, S., and Letaief, K. B. (2020). Client-edge-cloud hierarchical federated learning. In ICC 2020-2020 IEEE international conference on communications (ICC), pages 1–6. IEEE.

Manju, A., Kumar, C. P., Jegan, J., Jagadeeshan, D., and Nunna, S. K. (2024). Hierarchical federated learning with fog nodes: Enhancing efficiency in smart city networks. In 2024 OITS International Conference on Information Technology (OCIT), pages 749–753. IEEE.

Ooi, M. P.-L., Sohail, S., Huang, V. G., Hudson, N., Baughman, M., Rana, O., Hinze, A., Chard, K., Chard, R., Foster, I., et al. (2023). Measurement and applications: Exploring the challenges and opportunities of hierarchical federated learning in sensor applications. IEEE Instrumentation & Measurement Magazine, 26(9):21–31.

Sharma, P., Sharma, S. K., and Dani, D. (2025). Edge-assisted federated learning for anomaly detection in diverse iot network. International Journal of Information Technology, 17(5):3035–3045.

Silva, B., Matos, R., Callou, G., Figueiredo, J., Oliveira, D., Ferreira, J., Dantas, J., Lobo, A., Alves, V., and Maciel, P. (2015). Mercury: An integrated environment for performance and dependability evaluation of general systems. In 45th dependable systems and networks conference (DSN), pages 1–4.

Silva, L. G., Cardoso, I., Brito, C., Barbosa, V., Nogueira, B., Choi, E., Nguyen, T. A., Min, D., Lee, J. W., and Silva, F. A. (2023). Urban advanced mobility dependability: A model-based quantification on vehicular ad hoc networks with virtual machine migration. Sensors, 23(23):9485.

Siriwardhana, Y., Porambage, P., Liyanage, M., Marchal, S., and Ylianttila, M. (2024). Shield-secure aggregation against poisoning in hierarchical federated learning. IEEE Transactions on Dependable and Secure Computing.

Statista (2025). Number of internet of things (iot) devices connected worldwide from 2022 to 2023, with forecasts from 2024 to 2034. Accessed: 2025-02-04; Forecast shows increase from approximately 19.8 billion in 2025 to 40.6 billion by 2034.

Wen, J., Zhang, Z., Lan, Y., Cui, Z., Cai, J., and Zhang, W. (2023). A survey on federated learning: challenges and applications. International Journal of Machine Learning and Cybernetics, 14(2):513–535.

Zhang, H., Bosch, J., and Olsson, H. H. (2020). Federated learning systems: Architecture alternatives. In 2020 27th Asia-Pacific Software Engineering Conference (APSEC), pages 385–394.

Zhao, T., Li, F., and He, L. (2022). Drl-based joint resource allocation and device orchestration for hierarchical federated learning in noma-enabled industrial iot. IEEE Transactions on Industrial Informatics, 19(6):7468–7479.