Class imbalance data handling with optimal deep learning-based intrusion detection in IoT environment

The Internet of Things (IoT) has performed a paradigm shift in the method devices and systems interact, allowing seamless connectivity and communication. But, the enhancing interconnectedness and difficulty of IoT platforms also establish novel security problems, making intrusion detection a vital feature of IoT system security. Intrusion detection systems (IDS) role a vital play in detecting and monitoring unauthorized actions or malicious behavior in IoT networks. Typical IDS frequently depends on handcrafted rules or signatures that cannot not efficiently capture the difficult and developing nature of recent cyber threats. Deep learning (DL) methods automatically study and extract high-level representations in raw data, allowing more accurate and adaptive IDs. Most problems in emerging effectual IDSs are the presence of class imbalances in the database, but the count of normal instances far outweighs the count of intrusion samples. This paper focuses on the design of class imbalance data handling with optimal deep learning-based intrusion detection (CIDH-ODLID) techniques in IoT environments. The purpose of this study is to develop a CIDH-ODLID technique for the identification of intrusions in the IoT platform. For the class imbalance data handling process, the SMOTE approach is used in this work. In addition, the CIDH-ODLID technique employs an echo state network (ESN) approach for intrusion detection and classification. Finally, the snake optimization algorithm (SOA) was carried out for an optimum hyperparameter selection of the ESN approach. The performance validation of the CIDH-ODLID approach was performed on the benchmark database. To prove that the previously provided model performed better, a thorough simulation was run. The researchers provided a thorough comparative analysis that showed the proposed approach was better than other current procedures. It had an accuracy of 99.56%, precision of 97.50%, recall of 98.42%, and an F-score of 97.95%. The experimental outcomes exhibit the promising performance of the CIDH-ODLID technique over other models.