A Hybrid Ensemble Learning-Based Intrusion Detection System for the Internet of Things

The applications of the Internet of Things (IoT) have grown significantly both in scope and complexity. IoT devices are becoming an integral part of our daily lives. This significant growth in IoT adoption is accompanied by a substantial increase in the interest of malicious actors. IoT devices are a preferred target for malicious actors due to their inherent vulnerabilities and limited computational resources, which make them difficult to protect and secure. This study introduces a novel ensemble learning-based intrusion detection system (IDS) using network flow features. The goal of the proposed system is to achieve both simplicity and high detection accuracy. The novelty behind the system lies in using a new feature called 'history', extracted from flow information, combined with traditional features. The core classification engine includes bidirectional long short-term memory (BiLSTM) and multilayer perceptron (MLP) classifiers, with a decision tree (DT) classifier finalizing the decision-making process. The proposed system has been evaluated using a public IoT network dataset with an accomplished accuracy of 99.6%. The system has achieved results comparable to those of other systems that are more complex. The obtained results demonstrate the superior performance of the proposed ensemble learning-based system in comparison to conventional network-flow-based intrusion detection systems.