Detection of False Data Injection Attack using Machine Learning approach

The "False Data Injection" (FDI) attack is one of the significant security risks that the deep neural Network is susceptible to. The purpose of the FDI attacks is to deceive industrial platforms by faking sensor readings. considered a few relevant systematic reviews that have been previously published. Recent systematic reviews may include both older and more recent works on the topic. Therefore, I restricted myself to recently published works. Specifically, we analyzed data from 2016-2021 for this work. Attacks using FDI have effectively beaten out traditional threat detection strategies. In this paper, we provide an innovative auto-encoder-based technique for FDI attack detection (AEs). use of the temporal and spatial correlation of sensor data, which may be used to spot fake data. Additionally, the fabricated data are denoised using AEs. Performance testing demonstrates that our method is effective in finding FDI attacks. Additionally, it performs much better than a similar technique based on a support vector machine. The ability of the denoising AE data cleaning method to recover clean data from damaged (attacked) data is also shown to be quite strong.