A Method for Anomaly Traffic Detection in Power Wireless Access Networks Based on an Improved Transformer Model

Le résumé fourni par la source

With the increasing intelligence of power systems, the security of power wireless access networks has become a critical issue. To address the challenges of anomaly traffic detection, this paper proposes an improved Transformer-based detection method. By introducing relative position encoding and dilated causal convolution, the model's ability to capture long-term dependencies and causal structures is enhanced. A dual timewindow mechanism and data normalization strategy are employed to improve input data quality. Experimental results show that the proposed method outperforms LSTM and standard Transformer models in accuracy and recall, with lower false alarm and missed detection rates, demonstrating its practical applicability.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.