Enhancing IoT Data Balance Based on Feature Selection and Dataset Integration | ||||
| Al-Azhar Bulletin of Science | ||||
| Volume 2024, Issue 1, January 2024 | ||||
| DOI: 10.58675/2636-3305.1687 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
| Khaled Abdelhakim; Assad Ahmed Gad-Elrab; Mohamed Sayed Farag; Shaban Ebrahim Abu-Youssef | ||||
| Department of Mathematics, Faculty of Science, Al-Azhar University, Cairo, Egypt. | ||||
| Abstract | ||||
| The Internet of Things (IoT) has significantly advanced since its creation, revolutionizing both business processes and social interactions by connecting devices and data. However, this progress introduces security challenges. To tackle these issues, this paper presents an Imbalance Reduction Model in IoT Data Sets Based on Feature Selection architecture. This model employs machine learning methods to classify IoT traffic, emphasizing flow and Transmission Control Protocol data from datasets like UNSW-NB15 and Bot-IoT. The introduced model is proficient at distinguishing between normal, Denial of service, and distributed Denial of service traffic, addressing problems like data imbalance and overfitting. It achieves classification accuracy between 98.38% and 100%, significantly improving IoT security by effectively identifying and countering malicious traffic. Utilizing machine learning shows resilience to emerging threats, underscoring its potential as a strong intrusion detection system for IoT settings. | ||||
| Keywords | ||||
| Bot-IoT and UNSW-NB15 datasets; Cyber Security; Data Science; Feature Selection; Internet of Things; Intrusion Detection System; Machine Learning | ||||
|
Statistics Article View: 17 |
||||
