Detecting DDoS Attacks using Machine Learning in IoT Environment

Abstract

The escalating occurrence of distributed denial of service (DDoS) attacks within Internet of Things (IoT) environments stands as a significant focal point in the realm of cybersecurity. Our primary goal is to devise sophisticated detection mechanisms that leverage the capabilities of Machine Learning (ML) and Deep Learning (DL) techniques, addressing the imperative need for robust defense systems in IoT networks. Our investigation utilizes the CIC2023 IoT Dataset, employing diverse ML algorithms, including Logistic Regression, K-Nearest Neighbors (KNN) and Deep Neural Network (DNN), to adeptly recognize and categorize DDoS attack patterns. Our findings underscore noteworthy enhancements in detection accuracy, resulting in reduced false-positive rates and heightened precision in identifying various types of DDoS attacks, such as TCP, SYN, SlowLoris, HTTP, and UDP floods. Across all models and algorithms, we achieved a precision of 0.9999, accompanied by equally high recall, accuracy, and F1 scores. What sets this project apart is its distinctive contribution to the comprehensive evaluation of these algorithms in the context of IoT, offering valuable insights into their effectiveness and establishing a benchmark for future research in this critical area.