A machine learning based framework for IoT devices identification using web traffic.

Area of Science:

• Computer Science
• Network Security
• Machine Learning

Background:

• The proliferation of Internet of Things (IoT) devices in smart homes and offices introduces significant privacy and data security risks.
• Identifying unknown IoT devices is crucial for network management to prevent vulnerabilities and data theft.
• Manual device identification becomes impractical as the number of connected devices grows, necessitating automated solutions.

Purpose of the Study:

• To develop and evaluate an automated machine learning model for accurate identification of diverse IoT devices.
• To enhance network security by effectively distinguishing legitimate devices from potential intruders.
• To address the challenges of managing and securing networks with a large number of connected IoT devices.

Main Methods:

• Proposed an accuracy boosting model (ABM) employing ensemble machine learning techniques.
• Utilized Random Forest (RF) and Extreme Gradient Boosting (XGB) as base learners with adaptive boosting.
• Implemented feature engineering and cross-validation for robust IoT device identification.

Main Results:

• The proposed ensemble model achieved a high accuracy of 91% in identifying IoT devices.
• Demonstrated strong performance with precision, recall, and F1 scores all at 93%.
• Experimental validation was conducted using the IoT Device Identification dataset from a public repository.

Conclusions:

• The accuracy boosting model effectively identifies various IoT devices, including lights, thermostats, and security cameras.
• The developed machine learning approach offers a scalable and automated solution for securing smart environments.
• The model's high performance metrics confirm its potential for practical application in network security.