Optimized Intrusion Detection in IoT Networks Using Ensemble Deep Learning and Ant Colony Optimization

Many innovative applications in home automation, industry, health and environmental monitoring are possible because of the Internet of Things (IoT). At the same time, the rise in devices can make cybersecurity attacks more likely. Intrusion Detection Systems (IDSs) play a vital role in protecting networks by identifying and responding to attacks. Because IoT networks are used more often, there are privacy concerns since IDS use big data sets for deep learning (DL) and machine learning (ML). It is important to use effective DL/ML-based IDS systems to spot and classify attacks on IoT networks. The paper suggests using Ensemble Deep Learning Models and Ant Colony Optimization for an improved Intrusion Detection System (OIDS-EDLMACO). The goal is to improve detection of attacks in IoT networks by applying advanced ensemble models. Min-max normalization is used to preprocess the data at the start. After that, the Chimp Optimization Algorithm (ChOA) is used to select the most important features. OIDS-EDLMACO uses an ensemble of Bi-LSTM, SNN and SAE models for classification. Also, Ant Colony Optimization (ACO) is applied to tune the hyperparameters of the ensemble classifier for better performance. The OIDS-EDLMACO model reached an average accuracy of 99.27%, precision of 98.17%, recall of 98.17%, F1-score of 98.17% and detection rate of 98.86%. The findings show that OIDS-EDLMACO is effective for detecting intrusions. Future research will concentrate on protecting data during training and making sure federated IoT systems can scale in real time.