False Data Injection Attacks Detection in Real Smart Grid with Edge Computing

False Data Injection Attacks (FDIAs) pose great challenges to the security and stability of smart grid due to their stealthily distort measurement data. Excessively low data update frequency further makes it difficult to detect stealthy FDIAs. By taking the advantages of edge computing, it is possible to collect data adjacent to the terminals. In this work, we develop the Power Metering Edge Computing System (PMECS) that can collect data intensively and obtain the complete data from all terminals. A FDIAs detection scheme is implemented based on deep learning to achieve accurate detection of stealthy FDIAs using the real time data. Firstly, we demonstrate a data collection and stealthy FDIAs process in a smart grid with edge computing. Then, a deep learning based FDIAs detection scheme in edge computing is demonstrated. Experiment results verify that the implemented scheme can be applied to the real smart grid edge computing scenario for FDIAs detection.