Low rate DoS attack detection in IoT - SDN using deep learning

• Main Authors: Ilango, Harun Surej, Ma, Maode, Su, Rong
• Other Authors: School of Electrical and Electronic Engineering
• Format: Conference or Workshop Item
• Language: English
• Published: 2023
• Subjects: Engineering; Engineering::Electrical and electronic engineering; Engineering::Electrical and electronic engineering::Wireless communication systems; Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence; Internet of Things; Software Defined Networking; Deep Learning; Low-Rate DoS Attacks; Network Security; CIC DoS 2017; CIC IDS 2017
• Online Access: https://hdl.handle.net/10356/167111
• Institution: Nanyang Technological University

The lack of standardization and the heterogeneous nature of IoT, exacerbated the issue of security and privacy. In recent literature, to improve security at the network level, the possibility of using SDN for IoT networks was explored. An LR DoS attack is an insidious DoS attack that hinders the availability of the network to its legitimate users. LR DoS attacks are difficult to detect and can be deadly to a network due to their hidden nature. Recently, the possibility of using ML or DL algorithms to detect LR DoS attacks have gained traction due to advancements in computing technology. The ML and DL algorithms that are currently available in the literature have a detection rate of 95 percent at best. In this work, a novel deep learning scheme called FFCNN is proposed to detect LR DoS attacks in a SDN environment. The CIC DoS 2017 and CIC IDS 2017 datasets provided by the Canadian Institute of Cybersecurity were used for the experimental analysis. The empirical analysis of the proposed algorithm shows that it outperforms the existing machine learning based algorithms. FFCNN promises a lower false alarm rate and better detection rate in the detection of LR DoS.