Abstract:
Currently, there is a rapid adoption of Internet of Things (IoT) devices in the world.
However, the use of IoT poses different security implications that require reliable anomaly
detection systems. This thesis attempts to deal with the security issues in IoT based on
anomaly detection using honeypots and supervised learning. IoT-23 labeled dataset is generated by a network log tool called Zeek used in this research which includes honeypot network
traffic. Data is analyzed, cleaned and classified then processed by supervised learning algorithms to produce the required models for use in detection systems. The objective of this
work is to find the best Machine Learning (ML) model by examining different ML algorithms
that are successfully utilized. This thesis examined the following algorithms: Decision Tree
(DT), GaussianNB (NB), Random Forest (RF), Logistic Regression (LR), and SVC linear
(SL). The best results were achieved by the DT algorithm, with an accuracy 99.9%. Finally,
the proposed model was integrated with an Intruder Detection System (IDS) code to detect
malicious activities in an IoT environment.