Improving Prediction Accuracy in IoT Network Security Against DDoS Attacks by Combining RIPPER and KNN Algorithms

The Internet of Things (IoT) is a network of linked devices comprising interconnected devices that communicate through internet, integrating sensors, software and other technologies. As the number of connected devices grows, ensuring IoT security has become increasingly critical, with cyber threats and other related risks arising. A prevalent threat is the Distributed Denial of Service (DDoS) attack, which seeks to incapacitate server systems by overwhelming the network with excessive traffic. This study explores machine learning approach to improve intrusion detection systems (IDS) performance, such as accuracy in identifying and mitigating DDoS attacks. Specifically, it employs a hybrid approach combining the well-known RIPPER algorithm or Repeated Incremental Pruning to Produce Error Reduction and the K-Nearest Neighbors (KNN) algorithm as detection models. The study utilizes the Network Security Layer-Knowledge Discovery in Database (NSL-KDD) dataset, which contains 41 features for classifying DDoS attacks. To optimize the model's performance, the data underwent preprocessing to eliminate irrelevant and redundant features while retaining essential information. The classification results demonstrate a peak accuracy of 97.32% with k=6 and an average accuracy of 96.30% across 15 trials with varying k values from 1 to 15. These findings highlight the effectiveness of the RIPPER-KNN approach in improving the prediction accuracy of DDoS attack detection in IoT networks.