GGNB: Graph-Based Gaussian Naive Bayes Intrusion Detection System for CAN Bus

• Published in: arXiv.org 2021-08
• Main Authors: Islam, Riadul, Devnath, Maloy K, Samad, Manar D, Syed Md Jaffrey Al Kadry
• Format: Article
• Language: English
• Subjects: Control equipment; Controller area network; Cybersecurity; Datasets; Denial of service attacks; Electronic control; Field programmable gate arrays; Highway safety; Information systems; Intrusion detection systems; Search algorithms; Spoofing; Support vector machines; Traffic safety
• ISSN: 2331-8422

The national highway traffic safety administration (NHTSA) identified cybersecurity of the automobile systems are more critical than the security of other information systems. Researchers already demonstrated remote attacks on critical vehicular electronic control units (ECUs) using controller area network (CAN). Besides, existing intrusion detection systems (IDSs) often propose to tackle a specific type of attack, which may leave a system vulnerable to numerous other types of attacks. A generalizable IDS that can identify a wide range of attacks within the shortest possible time has more practical value than attack-specific IDSs, which is not a trivial task to accomplish. In this paper we propose a novel {\textbf g}raph-based {\textbf G}aussian {\textbf n}aive {\textbf B}ayes (GGNB) intrusion detection algorithm by leveraging graph properties and PageRank-related features. The GGNB on the real rawCAN data set~\cite{Lee:2017} yields 99.61\%, 99.83\%, 96.79\%, and 96.20\% detection accuracy for denial of service (DoS), fuzzy, spoofing, replay, mixed attacks, respectively. Also, using OpelAstra data set~\cite{Guillaume:2019}, the proposed methodology has 100\%, 99.85\%, 99.92\%, 100\%, 99.92\%, 97.75\% and 99.57\% detection accuracy considering DoS, diagnostic, fuzzing CAN ID, fuzzing payload, replay, suspension, and mixed attacks, respectively. The GGNB-based methodology requires about \(239\times\) and \(135\times\) lower training and tests times, respectively, compared to the SVM classifier used in the same application. Using Xilinx Zybo Z7 field-programmable gate array (FPGA) board, the proposed GGNB requires \(5.7 \times\), \(5.9 \times\), \(5.1 \times\), and \(3.6 \times\) fewer slices, LUTs, flip-flops, and DSP units, respectively, than conventional NN architecture.