Loading...

IoT Threat Mitigation: Machine Learning-Assisted Heavy Hitter Detection in P4-Enabled Networks

Hooshangi Naghani, M ; Sharif University of Technology | 2024

25 Viewed
  1. Type of Document: Article
  2. DOI: 10.1109/CPSAT64082.2024.10745449
  3. Publisher: 2024
  4. Abstract:
  5. The Internet of Things (IoT) is crucial in various sectors, making IoT networks prime targets for denial of service attacks. Detecting heavy hitters-primary sources of such attacks-is essential for network security. We address this in IoT gateway networks, where direct traffic measurement faces privacy and scalability issues. Existing methods like sampling and sketch-based approaches suffer from information loss and inflexibility. We propose a two-step scheme: a Convolutional Neural Network (CNN) selects relevant gateways, followed by a tomography-based approach to identify heavy hitters. Our data collection framework uses SNMP and P4 technologies for enhanced flexibility and accuracy. Extensive numerical evaluations demonstrate the proposed algorithm's effectiveness in accurately identifying heavy hitters, despite potential estimation errors in the tomography method. © 2024 IEEE
  6. Keywords:
  7. Convolutional Neural Network ; Heavy Hitters ; IoT ; Network Monitoring ; P4 ; SNMP ; Tomography ; Denial-of-service attack ; Gateways (computer networks) ; Image segmentation ; Internet of things ; Machine learning ; Network security ; Primary sources ; Threats mitigations ; Traffic measurements
  8. Source: 2024 5th CPSSI International Symposium on Cyber-Physical Systems (Applications and Theory), CPSAT 2024 ; 2024 ; 979-833152928-4 (ISBN)
  9. URL: https://ieeexplore.ieee.org/document/10745449