ToN-IOT Set: Classification and Prediction for DDoS Attacks using AdaBoost and RUSBoost

IoT security is a major concern due to its vulnerability to DDOS attacks. Due to inconsistent and increasing network traffic, vague packets are generated, which affects IoT security. Different machine-learning techniques have been employed to reduce DDoS assaults. The research gaps associated with DDoS attacks relate to the old datasets which lack information about recent attacks. Moreover, the centralized defense model also lacks the approach to prevent these attacks. As a solution to these prominent issues, reliable and scalable techniques need to be identified. The ML techniques provide different platforms for attack mitigation using a high level of security. This work employs a realistic IoT dataset named ToN-IoT, which has exhaustive and diversified network data. In this paper, we have used AdaBoost and RUSBoost for classification and prediction purposes. Through this ensemble classifier learning, the confusion matrix has been generated, and the performance of both models has been compared. The accuracy achieved using the RUSBoost classifier is 99.3%. Furthermore, the Adaboost achieved a better accuracy of 99.7% with less prediction speed.