Towards Resilient IoT Security: An Analysis and Classification of Attacks in MQTT-based Networks
Anshika Sharma, Himanshi Babbar · 2024
The term Internet of Things (IoT) refers to a system of actual physical objects, or things, that are integrated with software, firmware, sensors, and other technologies to communicate and share data with other internet-connected devices and systems. Since IoT security is even more comprehensive than IoT, many different approaches come under its purview. This paper has examined various attack scenarios against MQTT, a protocol utilized in IoT systems. The paper utilizes Machine Learning (ML) methods including Random Forest (RF), Adaptive Gradient Boosting (AdaBoost), Support Vector Machines (SVM), and Extreme Gradient Boosting (XGBoost). Machine learning has the potential to mitigate cyber threats and enhance security architecture through the identification of trends, the generation of real-time cybercrime maps, and the execution of thorough security testing. The meticulously selected data on IoT dangers served as a strong basis for evaluating and instructing the model. The assessment of the scalability and effectiveness of different algorithms has shown that proactive monitoring for IoT threats is achievable. XGBoost is the most accurate approach, with a 99% accuracy rate. The SVM, AdaBoost, and RF methods have detection rates of 85%, 89%, and 95% respectively.