Federated Digital Twins for Secure and Scalable Smart Infrastructure

Smart infrastructure systems increasingly rely on Digital Twin (DT) technology to enable real-time monitoring, predictive maintenance, and optimization. However, traditional centralized DT architectures face significant challenges in scalability, privacy preservation, and security when deployed across distributed infrastructure networks. This paper presents a novel federated digital twin framework that leverages federated learning principles to create secure, scalable, and privacy-preserving smart infrastructure systems. Our approach distributes computational workload across edge nodes while maintaining a global consensus model through Byzantine fault-tolerant aggregation. We demonstrate through extensive experiments on real-world smart grid and transportation datasets that our federated approach achieves 54.3 % reduction in latency and 38.7 % improvement in throughput compared to centralized architectures while ensuring differential privacy with <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\epsilon=1.2$</tex>. The proposed framework successfully handles up to 100 concurrent infrastructure nodes with 99.2% model accuracy, proving its effectiveness for largescale deployments.

Перевод пока недоступен