Embedded Files
Federated Learning for Resource Constrained Edge Networks
Federated Learning for Resource Constrained Edge Networks
Federated learning enables collaborative model training across diverse edge and IoT devices while preserving data privacy. However, real-world deployments face challenges such as device and data heterogeneity, limited bandwidth and energy, non-IID data distributions, and the need for secure coordination. Our research focuses on addressing these issues by developing intelligent device selection, dynamic resource allocation, and data-aware aggregation methods to improve learning efficiency and accuracy. Additionally, integrating privacy-preserving techniques and blockchain enhances security and reliability. These advances aim to create efficient, robust, and scalable federated learning frameworks suited for complex, resource-constrained edge and IoT environments.
Intelligent Device Selection in Federated Learning
Federated learning is increasingly deployed in wireless edge and IoT environments, where devices differ widely in computational capabilities, energy reserves, and data relevance. A central challenge in these settings is how to optimally select which devices participate in each training round to maximize learning quality while respecting constraints on latency, bandwidth, and energy consumption. Device suitability can be evaluated based on factors such as data volume, data quality, network conditions, and available resources, enabling only the most impactful devices to contribute to each round. Intelligent device selection can significantly improve global model performance, accelerate convergence, and enhance resource efficiency, paving the way for scalable and robust federated learning in practical edge network deployments.
Publications:
D. Kushwaha, S. Redhu, C. G. Brinton and R. M. Hegde, "Optimal Device Selection in Federated Learning for Resource-Constrained Edge Networks," in IEEE Internet of Things Journal, doi: 10.1109/JIOT.2023.3243082.
D. Kushwaha and R. M. Hegde. Optimized Device Selection and Resource Management Framework for Federated Learning. In 2025 National Conference on Communications (NCC), New Delhi, India, pp. 1-6.
D. Kushwaha, M. Narkhede, A. Limaye, N. Pol, and R. M. Hegde, "Device Selection for Resource-Efficient Edge Caching in a Federated Learning Framework" in 2025 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Hyderabad, India, pp. 1-5, doi: https://doi.org/10.1109/ICASSP49660.2025.10890488.
D. Kushwaha, M. Kalavadia, V. Hegde, and O. Pandey, "Energy-Efficient and Latency-Aware Blockchain-Enabled Federated Learning for Edge Networks," in IEEE Transactions on Circuits and Systems II: Express Briefs, doi: 10.1109/TCSII.2023.3322340.
D. Kushwaha and R. M. Hegde, "Optimal Device Selection and Resource Allocation in Federated Learning" in 2025 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Hyderabad, India, pp. 1-5, doi: https://doi.org/10.1109/ICASSP49660.2025.10888473.
Resource Allocation for Efficient Federated Learning
Federated learning in wireless edge and IoT environments faces significant challenges in resource allocation due to the heterogeneity of device capabilities, limited bandwidth, and energy constraints. These issues can be addressed by dynamically allocating communication resources among participating devices. Additionally, considering factors such as device energy, channel quality, and bandwidth availability ensures resource allocation for efficient learning and system scalability. By integrating adaptive bandwidth allocation and resource-aware device selection, the training latency and energy consumption can be minimized while maintaining high model accuracy.
Publications:
D. Kushwaha, S. Redhu and R. M. Hegde, "Low Latency Federated Learning over Wireless Edge Networks via Efficient Bandwidth Allocation," 2022 IEEE 8th World Forum on Internet of Things (WF-IoT), Yokohama, Japan, pp. 1-6, doi: 10.1109/WF-IoT54382.2022.10152237.
D. Kushwaha and R. M. Hegde. Optimized Device Selection and Resource Management Framework for Federated Learning. In 2025 National Conference on Communications (NCC), New Delhi, India, pp. 1-6.
D. Kushwaha and R. M. Hegde, "Optimal Device Selection and Resource Allocation in Federated Learning" in 2025 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Hyderabad, India, pp. 1-5, doi: https://doi.org/10.1109/ICASSP49660.2025.10888473.
Data Heterogeneity in Federated Learning
A key challenge arises in federated learning when local datasets are non-IID (Independent and identically distributed), leading to suboptimal global model performance if not properly addressed. A data-aware local model aggregation approach can ensure that clients with data distributions more representative of the global population have a stronger influence, resulting in improved model accuracy and fairness in heterogeneous federated learning environments.
Publications:
D. Kushwaha, A. Mehrotra, and R. M. Hegde, "Data Distribution-Aware Model Aggregation for non-IID Data in a Federated Learning Framework," in 2024 IEEE/IFIP Network Operations and Management Symposium (NOMS), Seoul, Korea, pp. 1-7, doi: 10.1109/NOMS59830.2024.10575887.
Page updated
Google Sites
Report abuse