Enhanced Detection of Evil Twin Attacks in Public Wi-Fi Networks Using Machine Learning Algorithms

The widespread use of public Wi-Fi networks has significantly increased users' exposure to cyber threats, with Evil Twin attacks posing a particularly insidious risk. These attacks manipulate users into connecting to fraudulent Wi-Fi access points, enabling attackers to intercept sensitive data or execute further malicious actions. This research introduces a robust, machine learning-driven methodology for the real-time detection of Evil Twin attacks in public Wi-Fi environments. Utilizing the AWID2 dataset, the study applied advanced feature selection and preprocessing techniques, including data balancing, noise reduction, and dimensionality reduction, to ensure data quality and model relevance. The resulting dataset was refined from 153 to 24 essential features, enhancing model performance and effi-ciency. Multiple machine-learning classifiers, including Random Forest, K-Nearest Neighbors, and Naive Bayes, were evaluated, with the Random Forest algorithm achieving a notably high accuracy of 99.9186%. These findings validate the framework's efficacy in real-time Evil Twin detection, providing a practical and effective solution to strengthen public Wi-Fi security and minimize cybersecurity risks for users and organizations.