AI techniques for automated penetration testing in MQTT networks: a literature investigation

The proliferation of IoT has made MQTT systems frequent targets of cyberattacks. While existing literature predominantly focuses on intrusion detection systems, this research aims to explore how machine learning algorithms can be leveraged to automate penetration testing in MQTT environments. A review of the literature reveals a significant gap in the field of automated testing, with few studies utilizing specific brokers and realistic datasets. This study contributes to the security of MQTT networks through: (a) a survey of widely used MQTT brokers, detailing their features, vulnerabilities, and mitigation measures; (b) an analysis of common attacks against brokers, including the identification of attack vectors, intrusion techniques, and defense mechanisms; (c) a mapping of machine learning models, such as Decision Trees (DT), Generative Adversarial Networks (GANs), and Boosting Algorithms, for detecting malicious activities and automating penetration testing in MQTT environments; and (d) the proposal of metrics to evaluate the effectiveness of the proposed models, considering both their ability to detect attacks and to successfully exploit vulnerabilities.