Semantic communication-based detection of False Data Injection Attacks in 6G-enabled smart grids

The increasing integration of advanced communication technologies in smart grids, particularly in the context of emerging 6G networks, exposes power systems to sophisticated cyber–physical threats such as False Data Injection Attacks (FDIAs). These attacks can bypass conventional detection mechanisms by introducing subtle yet contextually inconsistent data manipulations. Most existing FDIA detection approaches rely on statistical residual analysis or purely data-driven learning models, which often fail to exploit domain knowledge inherent to power system operations. This paper proposes a semantic communication-based framework for FDIA detection in 6G-enabled smart grids. The proposed approach integrates ontology-driven semantic encoding with Long Short-Term Memory (LSTM) networks to jointly capture contextual semantics and temporal dependencies in smart meter data. By embedding power system domain knowledge into the communication and detection pipeline, the framework enables the identification of semantically inconsistent measurements that may appear statistically plausible. To validate the proposed method, a custom smart meter prototype was developed to generate a large-scale dataset consisting of both normal and FDIA-compromised power consumption profiles. Extensive experimental results demonstrate that the proposed framework achieves high detection accuracy and low inference latency, while maintaining robustness under noisy communication conditions. Comparative evaluations against representative deep learning-based baselines show consistent improvements in detection performance and reliability. These results indicate that the proposed semantic-aware detection framework is well-suited for real-time monitoring and cybersecurity enhancement of future 6G-enabled smart grid systems.