Home // SECURWARE 2023, The Seventeenth International Conference on Emerging Security Information, Systems and Technologies // View article

Heterogeneous Network Inspection in IoT Environment with FPGA based Pre-Filter and CPU based LightGBM

Authors:
Zhenguo Hu
Hirokazu Hasegawa
Yukiko Yamaguchi
Hajime Shimada

Keywords: Malicious Traffic Detection; Machine Learning; FPGA; LightGBM

Abstract:
With the development of modern society, IoT has entered many aspects of our daily lives. At the same time, cyber attacks in IoT environments are becoming increasingly rampant. We urgently need a method to effectively inspect and detect them such as the usage of malicious traffic detection technology. Malicious traffic detection is usually divided into two aspects: signature based method and machine learning based method. The former method usually relies on pre-defined signatures or rules and cannot effectively detect unknown threats such as zero-day attacks. Although the latter method can detect unknown attacks, most of them focus on offline traffic and cannot adapt to the current realtime IoT network environment. In this paper, we propose a heterogeneous malicious traffic detection system which combines both of them to achieve the realtime detection. In this design, we utilize the bloom array to execute pre-filter in an FPGA board, and implement a CPU based LightGBM classifier to identify the filtered traffic. We also implemented an experiment to evaluate the proposed system on both training stage and inference stage, which shows the system has the ability to identify malicious traffic in the IoT network environment.

Pages: 27 to 32

Copyright: Copyright (c) IARIA, 2023

Publication date: September 25, 2023

Published in: conference

ISSN: 2162-2116

ISBN: 978-1-68558-092-6

Location: Porto, Portugal

Dates: from September 25, 2023 to September 29, 2023