D-ACO/GA - A bio-inspired strategy for feature selection in anomaly traffic detection in smart Internet of Things environments
Ant. Helder G. L. Júnior, Joaquim Celestino, Gustavo Augusto Lima de Campos · 2024
With the growth of the Internet of Things (IoT) and its adoption, organizations are confronted with security challenges. This highlighted the need for controls to reinforce information security. Such growth comes with a significant increase in cyber threats accompanied by delays in detection. Consequently, the use of artificial intelligence techniques, especially machine learning and bio-inspired algorithms, like the ant colony algorithm (ACO) and the Genetic algorithm (GA), have become essential for enhancing IoT security by identifying and countering anomalies. This paper proposes an intrusion detection strategy, D-ACO/GA, for IoT environments. The approach combines both ACO and GA with supervised machine learning classification techniques. Like this, ACO and GA are applied to data subset selection, targeting the identification of crucial characteristics for anomaly detection. Through this comprehensive strategy, we achieve a sensible accuracy of 0.9937% with the 14 features selected by ACO, and 0.9886% accuracy with the 12 features selected by GA.