Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{182534,
        author = {Aniket Rishikant Tiwari and Shubhangi P. Tidake},
        title = {Toward Personalized Federated Learning under Data Heterogeneity and System Constraints},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {12},
        number = {2},
        pages = {2430-2439},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=182534},
        abstract = {Federated Learning (FL) has emerged as a promising paradigm for privacy-preserving distributed machine learning, enabling multiple clients to collaboratively train a model without sharing raw data. However, in practical deployments, FL systems face two critical challenges: data heterogeneity and system constraints. The former arises due to inherently non-identical and non-independent (non-IID) data distributions across clients, while the latter includes issues such as limited computational resources, unreliable connectivity, and variable participation rates among edge devices. These factors often degrade the performance of conventional FL methods, especially when personalization is essential for client-specific tasks.
This research focuses on evaluating the effectiveness of personalized federated learning (PFL) techniques under realistic conditions involving both statistical heterogeneity and system limitations. We conduct a comparative analysis of three widely studied strategies—FedAvg, FedPer, and pFedMe—using heterogeneous data partitions and simulated constraints such as client dropout and reduced local computation. Our experimental setup is built using open-source FL frameworks and benchmark datasets. The results reveal meaningful trade-offs across personalization accuracy, communication efficiency, and client robustness. This study provides actionable insights for deploying federated learning in real-world, resource-constrained environments and highlights the adaptation potential and limitations of existing PFL methods.},
        keywords = {Personalized Federated Learning, Data Heterogeneity, System Constraints, Non-IID Data, Client Dropout, Edge Intelligence},
        month = {July},
        }