To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
An Efficient Intrusion Detection System for Attacks Detection in MQTT Protocol Using Machine Learning |
|
Author |
Tahani Gazdar |
| Citation |
Vol. 22 No. 11 pp. 791-798 |
|
Abstract |
IoT has gained a lot of attention during the last decade. Although the applications provided in an IoT environment are supposed to simplify and automate our everyday tasks, their deployment is facing many security challenges. The high number of connected smart devices which are enabled with sensing and communication capabilities, in addition to the inherent connectivity to the Internet and the high amount of collected data might cause different attacks to emanate. Intrusion detection systems are one of the important defense techniques against attacks. In IoT, lightweight, accurate, and efficient IDSs are required. Many studies show that ML has the potential to accurately detect attacks given adequate data. Although distinct efforts have been made in designing intrusion detection systems for IoT, IDS based on Machine learning models are still in the early stages. More important, few IDS consider detecting attacks specific to IoT protocols like MQTT, CoAP, DDS, AMQP, etc. Almost all studies suppose that IoT applications are based only on HTTP traffic. Apart from that, almost all intrusion detection models designed for IoT are not trained on IoT datasets, old datasets are used instead. These old datasets do not contain records about new attacks or IoT-specific attacks. In this paper, we propose a novel intrusion detection system to detect attacks in protocol MQTT using Machine Learning. Before training the models, many techniques are used to pre-process the data and select the most important features from the dataset. The proposed algorithms are evaluated on a dataset named MQTT-IoT-IDS2020 that contains MQTT-specific attacks. Obtained results show that Decision tree and Logistic regression models outperform KNN and Naive Bayes models in terms of accuracy and precision. |
|
Keywords |
IoT, Intrusion detection system, Machine Learning, MQTT, Feature |
|
URL |
