Detection of Botnet Attacks on IoT Using AI
Ahmad Rasheed, Mohammad M. Alnabhan · 2024
The wide adoption of IoT devices has resulted in a rapid increase in cybersecurity risks. A major countermeasure to minimize associated risk is to implement an Intrusion Detection System (IDS); however, the efficiency and performance of conventional IDSs might not match the sophistication of existing DDoS attacks within the IoT environment. This paper investigates recent advancements in botnet attacks within the IoT environment and highlights the use of CatBoost, a gradient-boosting algorithm, over traditional methods like SVM and Logistic Regression, for its ability to handle complex feature interactions and categorical data effectively. Experimentation was conducted on a dataset comprising network traffic data, encapsulating a wide range of features from basic connection metrics to complex interaction patterns, aimed at distinguishing normal operations from potential intrusions. Cross-validation techniques were used to validate the correctness of the predictions and address any potential overfitting scenarios. The accuracies obtained were 92.98%, 93.27%, and 99.49% for Logistic Regression, SVM, and CatBoost, respectively, demonstrating CatBoost's superior performance and practical implications for securing IoT environments against botnet attacks.