An intelligent anomaly detection system for IoT using a hybrid metaheuristic evolutionary strategy

This research focuses on metaheuristic and evolutionary-based hybrid strategies for Intrusion Detection Systems (IDS) on the Internet of Things (IoT). The IoT is susceptible to cyberattacks, and security is a major issue. The family of metaheuristic algorithms is evolutionary in nature. The Whale Optimization Algorithm (WOA) improves its performance in each run. The WOA also optimizes the hyperparameters of the random forest classifier, aiming to detect malicious nodes with high accuracy and efficiency. The proposed hybrid strategy outperforms existing IoT anomaly detection systems, improving security and reliability in applications like smart homes, healthcare, and industrial automation. The proposed system’s effectiveness is evaluated using the famous IoT network datasets. This hybrid strategy outperforms existing anomaly detection systems for IoT using evaluation matrices. The proposed model with evolutionary intelligence can also be tested and evaluated on unseen datasets. Comparative analysis of machine, deep, and hybrid learning algorithms for IDS proves the superiority of the proposed hybrid evolutionary algorithm. For normal and malicious classes, the value of precision is 0.96 and 0.99, respectively. Normal node recall and F1-score are 0.91 and 0.94. Recall and F1-score for malicious nodes reach 1.00.