Use of machine learning for securing IoT
Loading...
Date
Author(s)
Citation for Previous Publication
Link to Related Item
Abstract
Description
IoT comprises of cyber-physical devices connected through the internet. These devices carry sensitive data and support critical services. Due to the sensitivity of user data shared across IoT, there is a risk of attacks such as spoofing, eavesdropping, and denial of service (DoS). These attacks can be detected and prevented by cybersecurity systems that use machine learning techniques for malicious pattern analysis. The primary objective of the research is to develop an improved machine learning model that classifies network traffic as either malicious or benign. For this purpose, the IoT-23 dataset is used, which includes twenty-three scenarios of network traffic, out of which twenty are from malwareinfected IoT devices, and three are from benign IoT devices. The dataset has several feature columns that are transformed in a way to feed into the model using feature engineering techniques. The model is constructed using a random classifier by choosing parameters to increase the accuracy of classifying network traffic. This accuracy can be improved by constructing an ensemble model that combines random forest classifier with other classifiers such as K-Nearest Neighbor and Gaussian Naïve Bayes using hard voting. This research provides a model so that malicious traffic can be detected with more accuracy using machine learning algorithms. The results of the model can be evaluated using confusion matrix.
Item Type
http://purl.org/coar/resource_type/c_1843
Alternative
Other License Text / Link
Language
en