Quantum Dwarf Mongoose Optimization With Ensemble Deep Learning Based Intrusion Detection in Cyber-Physical Systems

Cyber-physical systems (CPS) combine computational and physical elements to enable effective and intelligent control of several applications. However, the increasing connectivity and complexity of CPS introduce new security challenges, making intrusion detection a critical aspect for maintaining the integrity and reliability of these systems. The rise in artificial intelligence (AI) techniques assists in addressing security problems related to CPS environments. Therefore, this study proposes a Quantum Dwarf Mongoose Optimization with Ensemble Deep Learning Based Intrusion Detection (QDMO-EDLID) technique in the CPS environment. The presented QDMO-EDLID technique aims to recognize the presence of intrusions by the feature selection (FS) and ensemble learning process. For feature subset selection purposes, the QDMO-EDLID technique employs the QDMO algorithm. Moreover, an ensemble of Convolution Residual Networks (CRN), Deep Belief Networks (DBN), and Deep Autoencoder (DAE) models are applied for the intrusion classification process. The experimental outcome of the QDMO-EDLID technique was tested employing benchmark intrusion databases. The simulation results highlighted the improved efficiency of the QDMO-EDLID approach concerning different performance measures.