Optimized Siamese Humboldt Squid Attention Network Architecture for Cyber-Attack Detection in IoT 4.0

The rapid expansion of IoT within Industry 4.0 has heightened the risk of cyber-attacks, underscoring the need for advanced detection mechanisms. Existing cyber-attack detection systems in this context often struggle with reduced accuracy, high computational time, complexity, and elevated error rates. This study addresses these challenges by proposing an Optimized Siamese Humboldt Squid Attention Network (OSH-SqAtNet) architecture for cyber-attack detection in IoT Industry 4.0 environments. The input data, sourced from datasets like CSE-CIC-IDS2018, undergoes preprocessing via a Mixture of Conventional Fast Bilateral Filter (MC2FB) to enhance data quality. A Gold Rush Optimizer (GRO) is then employed for feature selection, isolating the most significant features to optimize classification performance. The preprocessed and optimized data are classified using the OSH-SqAtN et architecture for accurate cyber-attack detection. For secure data transmission, Rabin Fuzzy Cryptosystem assisted Elliptic Curve Cryptography (RFC-ECC) is utilized. Experimental simulations conducted on the Python platform reveal that the proposed method outperforms existing techniques, achieving higher accuracy (98%) and recall (0.993%). These results demonstrate the approach's superior efficacy and potential for advancing cybersecurity in IoT Industry 4.0.