Hybrid Sampling Approach to Enhance Intrusion Detection System in IoT Networks

Abstract

The growing adoption of Internet of Things (IoT) networks has introduced new challenges in cybersecurity and intensified the need for robust security mechanisms, particularly for detecting and mitigating network intrusions. Given the massive amount of data generated by IoT devices and the highly imbalanced nature of intrusion datasets, traditional detection methods often struggle with accuracy, especially in identifying minority class attacks. This paper presents a hybrid sampling approach designed to enhance Intrusion Detection Systems (IDS) for IoT environments by addressing data imbalance. The proposed method combines the Near Miss-1 undersampling technique with Synthetic Minority Over-sampling Technique (SMOTE) to create a more balanced dataset without significant information loss. Using the Bot-IoT dataset, we evaluate the proposed approach across various machine learning algorithms, assessing their performance using key metrics including accuracy, False Positive Rate (FPR), and False Negative Rate (FNR). The experimental results demonstrate that Random Forest (RF) and K-Nearest Neighbor (KNN) excel across all metrics. The performance of the proposed approach is compared with other state-of-the-art methods, focusing on binary classification tasks and the ability to detect attacks while minimizing FPR. This study provides valuable insights into the application of hybrid sampling techniques for improving the detection capabilities of IDS in IoT networks.