Year: 2024
Type: dissertation
Abstract: Widespread and increased cyberattack against Internet of Things (IoT) are causing enormous range of problem for individual and organizations. The growing need for these services has a possibility to contain anomalies in the IoT data network has emerged as a key challenge. This research evaluates machine learning algorithms for detecting both traditional network intrusion and IoT network logs. Two datasets have been used IoT network log and Network Intrusion log dataset to classify various models on the performance metrics. To deal the issue of class imbalance and scalability nature in network and IoT dataset, Feature selection like Random Forest (RF), Correlation Coefficient and Cross-Validation & Regularization has achieved 99%, by improving this real-time processing combined with effective anomaly detection ensures threats are identified and mitigated quickly. Supervised learning models Decision Tree and KNN model has shown high accuracy of 99%. The findings show the capacity for modifying the machine learning technique to achieve high accuracy to identify labeled malicious in network traffic and IoT logs.
Author: Sriramulu Deenadayala Babu, Prasanth
Institution: National College of Ireland
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7001057618
