<![CDATA[The Internet of Things (IoT) has become an integral part of modern life, connecting smart devices to enhance efficiency and convenience. However, with the increased adoption of IoT, cybersecurity threats, particularly malware, have also risen. This research focuses on detecting malware attacks in IoT networks using machine learning algorithms, specifically Decision Tree and Random Forest. The dataset used is CICIoT2023, which includes various types of IoT network traffic such as BenignTraffic, Mirai-greeth_flood, Mirai-greip_flood, and Backdoor_Malware. In this study, both algorithms demonstrated exceptionally high accuracy on the training data, reaching 100%, and on the test data, achieving 99.94% accuracy for the Random Forest algorithm and 99.90% for the Decision Tree algorithm. Although the performance of both algorithms on the training data was almost identical, Random Forest showed better performance in detecting the Backdoor_Malware class compared to Decision Tree when using test data. Random Forest achieved a precision of 99%, recall of 64%, and F1-Score of 78%, while Decision Tree achieved a precision of 71%, recall of 72%, and F1-Score of 72%. Results from 10-fold cross-validation indicate that the models did not experience overfitting, suggesting reliable and well-generalized models. This research provides insights that the Random Forest algorithm is more effective in detecting malware attacks in IoT networks compared to Decision Tree, particularly in identifying the Backdoor_Malware class. These findings are expected to contribute to the development of more efficient and reliable malware detection systems for IoT networks.]]>
Copyrights © 2024
