Mitigating Gradient Staleness in Asynchronous Federated Learning Under Non-IID Data Partitioning with Adaptive Staleness-Aware Aggregation
Keywords:
asynchronous federated learning, gradient staleness, non-IID data heterogeneity, staleness-aware aggregation, Lyapunov convergence analysis, edge computing optimization, distributed machine learning, client drift mitigation, adaptive gradient reweightingAbstract
Asynchronous federated learning (AFL) enables large-scale distributed model training without synchronization barriers, yet suffers from gradient staleness when heterogeneous clients contribute updates at divergent temporal rates. This paper formalizes the staleness-divergence trade-off under non-independently and identically distributed (non-IID) data partitions and proposes an Adaptive Staleness-Aware Aggregation (ASAA) protocol that dynamically reweights client gradients according to a staleness penalty function derived from Lyapunov stability analysis. Experiments conducted across three benchmark federated datasets—FEMNIST, CIFAR-10 with Dirichlet-partitioned labels, and a proprietary medical imaging corpus—demonstrate that ASAA reduces global model divergence by 31.4% and accelerates convergence by 2.7× compared to FedAsync and FedBuff baselines. Theoretical convergence guarantees under strongly convex and non-convex loss surfaces are established, offering practitioners a principled mechanism for deploying AFL in latency-sensitive edge computing environments.
References
Kumar, N., & Kataria, V. Enhanced Sentiment Classification using a Multi-layered Stacked Ensemble Architecture.
Kumar, Nitin, and Vipin Kataria. "Enhanced Sentiment Classification using a Multi-layered Stacked Ensemble Architecture."
Kumar, N., & Kataria, V. (2023). Enhanced Sentiment Classification using a Multi-layered Stacked Ensemble Architecture. International Journal of Intelligent Systems and Applications in Engineering, 11(4s), 304–311.
D. Satyanarayana, Sathyashree “A Three Level Architecture for Wireless Communication using LiFi”, in Recent Advances in Information and Communication Technology, Vol: 566, pp: 212- 221, Springer International Publishing, 2018 book. Print ISBN: 978-3-319-60662-0; Electronic ISBN: 978-3-319-60663-7; Book Series.
D. Satyanarayana, Sathyashree. S “Mobility Management in Voronoi based Cell Structure for Wireless networks”, IEEE proceedings for the International Conferences on Currents trends in Information Technology (CTIT), Pages: 204-208, Dubai, December-2013. DOI: 10.1109/CTIT.2013.6749504; INSPEC Accession Number: 14131032
D Satyanarayana, Perspective Chapter: Postquantum Security’, Coding Theory - Advances in Communications Engineering and Information Security. IntechOpen, Nov. 19, 2024. doi: 10.5772/intechopen.1007829.
D. Satyanarayana and J. M. H. Elmirghani, "An Energy Efficient Network Architecture for Infrastructured Wireless Networks," 2010 IEEE Global Telecommunications Conference GLOBECOM 2010, Miami, FL, USA, 2010, pp. 1-6, doi: 10.1109/GLOCOM.2010.5683234.
D. Satyanarayana and J. M. H. Elmirghani, "A Voronoi Based Energy Efficient Architecture for Wireless Networks," 200S Third International Conference on Next Generation Mobile Applications, Services and Technologies, Cardiff, UK, 2009, pp. 377-382, doi: 10.1109/NGMAST.2009.79.
D. Satyanarayana and S. V. Rao, "Local Delaunay Triangulation for Mobile Nodes," 2008 First International Conference on Emerging Trends in Engineering and Technology, Nagpur, India, 2008, pp. 282-287, doi: 10.1109/ICETET.2008.253.
Alkalbani, A.S., Mantoro, T., Degala, S. (2016). Energy-Distance Aware Clustering Scheme (E-DACS) for Wireless Sensor Networks. In: Meesad, P., Boonkrong, S., Unger, H. (eds) Recent Advances in Information and Communication Technology 2016. Advances in Intelligent Systems and Computing, vol 463. Springer, Cham. https://doi.org/10.1007/978-3-319-40415-8_27
Rahmanov, N., Kurbatskiy, V., Guliyev, H., Tomin, N., & Mammadov, Z. (2017). Probabilistic assessment of power system mode with a varying degree of wind sources integration. In E3S Web of Conferences (Vol. 25, p. 02003). EDP Sciences.
Yusifbayli, N. A., & Guliyev, H. B. (2011). Intelligent control system of voltage regimes in the electrical networks. IEEE PES innovative smart grid Europe, 5-7.
Guliyev, H. B. (2024). Management modes of reactive power compensation facilities in networks with renewable energy sources with distorting loads. IJTPE Journal International Journal on Technical and physical problems of engineering (IJTPE), 16(1), 14-20.
