TDGCN-Based Mobile Multiuser Physical-Layer Authentication for EI-Enabled IIoT

TL;DR

This work tackles mobile multiuser physical-layer authentication in EI-enabled IIoT by introducing a Temporal Dynamic Graph Convolutional Network (TDGCN) that fuses Temporal Convolutional Networks for per-dimension temporal dynamics with Dynamic Graph Neural Networks to capture evolving cross-dimension dependencies. The architecture integrates Dynamic GINs and cascade node clustering pooling to efficiently aggregate information while maintaining manageable complexity. Empirical results show TDGCN markedly outperforms seven baselines, achieving near-ideal accuracy under clean CSI and substantial gains under realistic noise and mobility, thereby enhancing endogenous security for edge-enabled industrial networks. The approach enables robust, low-latency authentication for IIoT devices in dynamic environments, with practical implications for secure edge intelligence deployments.

Abstract

Physical-Layer Authentication (PLA) offers endogenous security, lightweight implementation, and high reliability, making it a promising complement to upper-layer security methods in Edge Intelligence (EI)-empowered Industrial Internet of Things (IIoT). However, state-of-the-art Channel State Information (CSI)-based PLA schemes face challenges in recognizing mobile multi-users due to the constantly shifting CSI distributions with user movements. To address this issue, we propose a Temporal Dynamic Graph Convolutional Network (TDGCN)-based PLA scheme, which employs Graph Neural Networks (GNNs) to capture the spatio-temporal dynamics induced by user movements. Firstly, we partition CSI fingerprints into multivariate time series and utilize dynamic GNNs to capture their associations. Secondly, Temporal Convolutional Networks (TCNs) handle temporal dependencies within each CSI fingerprint dimension. Additionally, Dynamic Graph Isomorphism Networks (GINs) and cascade node clustering pooling further enable efficient information aggregation and reduced computational complexity. Simulations demonstrate the proposed scheme's superior authentication accuracy compared to seven baseline schemes.

Figures (8)

• Figure 1: Proposed TDGCN-based PLA scheme.
• Figure 2: Dynamic Graph Transformation.
• Figure 3: Cascade node clustering pooling.
• Figure 4: Positions of Alices, Eves, IRSs and Bob, where the positions of Alices are (10, 82/84/86/88, 0)
• Figure 5: Authentication accuracy of the proposed TDGCN-based PLA scheme versus different iteration numbers under different SNRs.