Machine Learning and Deep Learning Based Intrusion Detection for MitM Attack Mitigation in IoT Networks

The Internet of Things (IoT) connectssensors, smart appliances, mobile devices and industrial equipment so that they can exchange informationandrespond to events. This connectivity is useful, but it also creates security risks when devices have limitedcomputing resources and communicate through different protocols. A Man-in-the-Middle (MitM) attackisonesuch risk: an attacker places themselves between communicating devices and may observe oraltertheexchanged data. This paper presents a machine-learning- based workflow for identifying traffic associatedwithMitM attacks in IoT networks. The proposed workflow includes traffic preparation, preprocessing,featureselection, model training, classification and evaluation. Support Vector Machine (SVM), RandomForestandLong Short-Term Memory (LSTM) are considered because they offer different ways to classifynetworktraffic. The supplied synopsis reports results from the referenced 2025 study in which Random Forestachieved94% accuracy, LSTM 92%, and SVM 85.7% under that study’s experimental conditions. Thesefiguresare reported as prior-study results, not as results independently produced by this paper. Thediscussionalso identifies practical issues such as false alarms, limited device resources, dataset representativenessandthe need for testing in live IoT environments