Augmenting cybersecurity through attention based stacked autoencoder with optimization algorithm for detection and...

The Internet of Things (IoT) network is a fast-growing technology, which is efficiently used in various applications. In an IoT network, the massive amount of connecting nodes is the existence of day-to-day communication challenges. The platform of IoT uses a cloud service as a backend for processing information and maintaining remote control. To manage the developing intricacy of cyberattacks, it is critical to have an effectual intrusion detection system (IDS), which can monitor computer sources and create data on suspicious or abnormal actions. The IoT network's security can progressively become a critical concern as IoT technology obtains extensive use. Protecting IoT systems with traditional IDS is challenging due to the vast variety and volume of IoT devices. Currently, Machine Learning (ML) and Deep Learning (DL) techniques are utilized to address the security threats in IoT networks. This manuscript proposes a Cybersecurity through an Attention-based Stacked Autoencoder with a Pelican Optimization Algorithm for the Detection and Mitigation of Attacks (CASAE-POADMA) methodology on an IoT-assisted network. The main purpose of the CASAE-POADMA methodology is to identify and mitigate the presence of cybersecurity attack behavior in the IoT-assisted network. At first, the presented CASAE-POADMA approach utilizes min-max normalization to scale input data into a uniform design. Besides, the greylag goose optimization (GGO) method is employed for the feature selection process. For the detection and mitigation of attack, the presented CASAE-POADMA approach employs the attention-based stacked autoencoder (ASAE) method. Eventually, the hyperparameter tuning of the ASAE method is executed by using pelican optimization algorithm (POA) method. The simulation validation of the CASAE-POADMA approach is verified under a benchmark database. The experimental validation of the CASAE-POADMA approach exhibited a superior accuracy value of 99.50% over existing techniques.