GLGOA-LSTM-BC: A Blockchain-Enabled Deep Learning Framework for Real-Time Cyberattack Detection in Drone Networks

Abstract

In the advancing domain of drone systems, cybersecurity is a critical issue owing to the rising threat of advanced cyberattacks. This study presents an innovative framework for drone cybersecurity that utilizes the integration of deep learning and blockchain technologies to efficiently detect and prevent malicious intrusions. The proposed architecture consists of four main stages: data normalization, feature selection utilizing the greylag goose optimization algorithm (GLGOA), long short-term memory (LSTM)-based cyberattack detection, and blockchain-based data validation. Initially, raw drone sensors and network data are standardized using normalization techniques to ensure consistency and minimize noise. GLGOA is utilized to extract the most pertinent features, thereby improving detection efficiency and reducing computational burden. The enhanced feature set is input into an LSTM model designed to capture temporal dependencies and classify potential cyber threats. Ultimately, blockchain integration guarantees the immutable recording of drone interactions and improves overall data security and reliability. Comprehensive experimental assessment illustrates the superiority of the proposed GLGOA-LSTM-BC model compared to traditional methods such as SVM, random forest, CNN, and GRU regarding -score. The proposed method demonstrates a 97.8% accuracy and a 97.5% f1-score, establishing it as a robust and reliable solution for real-time cyberattack detection in drone environments. The amalgamation of bio-inspired optimization, deep learning, and distributed ledger technologies facilitates the development of secure, intelligent, and autonomous drone systems within contemporary digital infrastructure.