Research Article
Trust-Aware Federated Learning with Differential Privacy for Secure AIoT in Critical Infrastructures
@ARTICLE{10.4108/eetiot.10656, author={Ramana Kadiyala and C. V. Lakshmi Narayana and S. China Ramu and Narsaiah Putta and Shyam Sunder Pabboju and B. Ramana Reddy}, title={Trust-Aware Federated Learning with Differential Privacy for Secure AIoT in Critical Infrastructures}, journal={EAI Endorsed Transactions on Internet of Things}, volume={11}, number={1}, publisher={EAI}, journal_a={IOT}, year={2025}, month={12}, keywords={federated learning, Differential Privacy, Homomorphic Encryption, Graph Neural Networks, Trust-Aware Aggregation, Critical Infrastructures, AIoT}, doi={10.4108/eetiot.10656} }- Ramana Kadiyala
C. V. Lakshmi Narayana
S. China Ramu
Narsaiah Putta
Shyam Sunder Pabboju
B. Ramana Reddy
Year: 2025
Trust-Aware Federated Learning with Differential Privacy for Secure AIoT in Critical Infrastructures
IOT
EAI
DOI: 10.4108/eetiot.10656
Abstract
Federated learning offers a scalable solution for distributed intelligence in Artificial Intelligence of Things (AIoT) systems, yet privacy leakage, adversarial attacks, and system heterogeneity remain persistent challenges in critical infrastructures such as smart cities, agriculture, and forestry. This paper proposes PriSec- FedGuardNet, a trust-aware federated learning framework that integrates differential privacy, homomorphic secure aggregation, and graph neural network–based trust evaluation to safeguard both data and model updates. The framework preserves sensitive information by perturbing gradients with calibrated noise, encrypts local updates for aggregation without decryption, and assigns trust scores to filter unreliable participants. Experimental validation on ToN-IoT, Bot-IoT, and real-world sensor datasets demonstrates that PriSec-FedGuardNet maintains above 97.3% relative utility under strict privacy budgets, improves anomaly detection F1-scores by up to 18% under poisoning attacks, and reduces device-level energy overheads to less than 12%. Domain-specific evaluations across Indian smart city, agricultural, and forestry deployments further highlight the adaptability and efficiency of the framework. By balancing privacy, security, and utility, PriSec-FedGuardNet establishes a robust paradigm for secure federated learning in AIoT-driven critical infrastructures.
Copyright © 2025 Kadiyala Ramana et al., licensed to EAI. This is an open access article distributed under the terms of the CC BY-NC-SA 4.0, which permits copying, redistributing, remixing, transformation, and building upon the material in any medium so long as the original work is properly cited.
