Hybrid-AI Blockchain Supported Protection Framework for Smart Grids

Abstract

In the digtal era, smart grid is a vital critical digital infrastructure of the nation that has concern for security against cyber threats in its power and communication infrastructures. Huge data gathering and bi-way information flows open up the potential for compromising data security and confidentiality. Cyber security requirements of data availability and data integrity are addressed in the proposed work by a deep learning security framework to mitigate False Data Injection Attacks (FDIA) that affects data integrity and Distributed Denial of Service (DDoS) attacks that are threats to data availability. The proposed Hybrid-AI Blockhain supported Protection Framework (HABPF) utilizes a hybrid of Recurrent Neural Networks (RNN) and LeNet5 based Convolutional Neural Networks (CNN) to protect the communication infrastructure of smart grid. The proposed system also leverages blockchain to store all grid data to add a layer of security against data tampering. The extensive performance and security evaluation of the HABPF framework through the IEEE 14 bus system reveals that the proposed framework is highly competent with 96% accuracy in detecting attacks compared to other related works.