Two-stage IDS for IoT using layered machine- and deep-learning models
André van der Walt, Tahmid Quazi, Brett van Niekerk · Cyber-Physical Systems · 2022
The ever-growing integration of Internet-of-Things (IoT) devices into our daily lives provides us with a level of convenience never before seen. However, the effects of attacks on these devices can be devastating. The discrete, low-powered nature of IoT devices makes their security a difficult problem to solve. To provide a solution, this work proposes a two-stage Intrusion Detection Systems (IDS) using layered machine and deep learning models. The potential benefits of the system are examined and the results presented show a reduction of threat detection/identification time of 0.51 s on average and an increase of threat classification F1-Score by 0.05.