1. ВИЯВЛЕННЯ ВТОРГНЕНЬ В МЕРЕЖУ ІНТЕРНЕТУ РЕЧЕЙ ШЛЯХОМ АНАЛІЗУ НАЯВНОСТІ АНОМАЛІЙ В МЕРЕЖЕВОМУ ПОТОЦІ ТРАФІКУ AAA ПРОТОКОЛІВ
- Author
-
Гальчинський, Л. Ю.
- Subjects
- *
CYBERTERRORISM , *INTERNET of things , *TRAFFIC flow , *MODERN civilization , *INTERNET security , *MACHINE learning - Abstract
Today, one of the most powerful trends in innovative change in the modern world is the Internet of Things IoT. And although this trend is essentially only at an early stage, it already plays a significant role in the modern world, in modern civilization. It is safe to predict that further expansion of its scope. However, the emergence of this phenomenon is also accompanied by the emergence of related problems. This is especially true of cyber threats, which can cause significant harm to both the Internet of Things itself and the people who use those networks. Accordingly, there is a need for means to combat these threats. This is a rather complex problem, which primarily requires the development of methods and technologies for detecting such threats. At the same time, there are no clear concepts of relevant solutions for cyber security of various IoT applications, especially where financial resources are limited. Therefore, research is currently relevant to find solutions that may be acceptable for relatively low-budget facilities. The aim of the work is to obtain a method for detecting intrusions into the IoT network by detecting anomalies in the network traffic flow of AAA protocols. This problem is complicated by the fact that the solution found must comply with basic security policies. The study conducted in this paper showed that such a solution is possible in principle by using the capabilities of the AAA RADIUS protocol and the use of machine learning. Machine learning methods were tested on the UNSW_NB15 dataset. For machine learning, pre-processing was performed to normalize and standardize the data. Computer experiments have tested the suitability of detecting a set of cyberattacks for different models of machine learning. It was shown that the formulated approach has the potential to be used in practical developments to detect intrusions in the IoT network. [ABSTRACT FROM AUTHOR]
- Published
- 2022
- Full Text
- View/download PDF