Adaptive Encryption Scheduling for Holographic Counterpart Communication in IoT Consumer Electronics

Xinyu He, Lingling Zhang, Jing Wang, Hong Zhang

Abstract: The convergence of Internet of Things (IoT) and Holographic Counterpart technologies in consumer electronics creates heterogeneous systems where data flows exhibit vastly different sensitivity levels. Traditional encryption approaches fail to address this heterogeneity. To address the fundamental challenge of adaptive encryption resource allocation in dynamic heterogeneous environments, this paper aims to maximize sensitivity-weighted security gain while ensuring strict resource constraint satisfaction. This paper proposes SA-GRL, a sensitivity-aware graph reinforcement learning framework for intelligent encryption resource scheduling in Holographic Counterpart-IoT consumer electronics systems. Our approach integrates data sensitivity into graph neural network representations through specialized embedding mechanisms, employs dual-stream policy networks for enhanced decision interpretability, and utilizes Lagrangian-based constrained reinforcement learning to guarantee resource constraint satisfaction. SA-GRL uniquely integrates sensitivity-aware graph neural networks with constrained reinforcement learning for IoT encryption scheduling. Extensive experiments on synthetic datasets demonstrate that SA-GRL achieves higher security gain compared to state-of-the-art methods while maintaining satisfactory resource utilization and perfect constraint satisfaction.

External IDs:doi:10.1109/tce.2025.3625227