An Empirical Examination of Machine Learning-Based Attack Classification in Internet of Things Networks
Virendra Kumar Verma, K Guhan, Ch. Ramakrishna, Savita Verma, Amit Kumar Sharma, S. Praveena · 2024
The rapid expansion of Internet of Things (IoT) networks has increased their vulnerability to cyberattacks, necessitating the development of complex security mechanisms. This study conducts an empirical evaluation of many machine learning (ML) techniques used for threat categorization in IoT networks, such as Decision Trees, Support Vector Machines (SVM), Random Forests, and Neural Networks. Using the UNSW-NB15 dataset, the research mimics a variety of IoT traffic, including both positive and harmful behaviors like malware and distributed denial of service (DDoS) assaults. A significant amount of preprocessing was carried out to improve the model's performance, including feature selection and data normalization. The results showed that Random Forests maintained the highest overall accuracy while outperforming other models in terms of precision, recall, and F1 score. The fact that there are still a lot of false positives and false negatives in all models shows how difficult IoT security is. By highlighting the benefits and drawbacks of several machine learning models in practical Internet of Things scenarios and offering suggestions for enhancing attack detection systems and guiding the creation of next IoT security solutions, this study advances the industry.