Cascade stacked autoencoder neural network for intrusion detection in CAN-FD vehicular network

In this work, an Intrusion Detection System (IDS) for Controller Area Network with Flexible Data Rate (CAN-FD) Vehicle Networks based on hybrid Deep Learning (DL) is proposed. Initially, the CAN-FD vehicular network simulation is carried out, and then, the authentication protocol is utilized to incr...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Knowledge-based systems Ročník 329; s. 114301
Hlavní autori: Devi, V. Anjana, Reddy, P.V. Bhaskar, Ponnada, Sreenu, Kumar, K. Suresh
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Elsevier B.V 04.11.2025
Predmet:
Cascade stacked autoencoder neural network
Deep neuro fuzzy network
Deep stacked autoencoder
Intrusion detection System
Z-score normalization
Intrusion detection System
Deep neuro fuzzy network
Deep stacked autoencoder
Z-score normalization
Cascade stacked autoencoder neural network
ISSN:0950-7051
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:In this work, an Intrusion Detection System (IDS) for Controller Area Network with Flexible Data Rate (CAN-FD) Vehicle Networks based on hybrid Deep Learning (DL) is proposed. Initially, the CAN-FD vehicular network simulation is carried out, and then, the authentication protocol is utilized to increase the existing security of in-vehicle applications that verify the authenticity of the participating entities. Later, inter-service communication and external communication are established. Finally, IDS is performed, and the proposed IDS in CAN-FD In-Vehicle Networks is developed in the following manner. Initially, the input data undergoes normalization using z-score normalization. After that, feature selection is performed by a hybrid similarity measure based on Tanimoto and Jeffreys similarity to select the relevant features in the input data. Finally, intrusion detection is performed based on a hybrid DL model named Cascade Stacked Autoencoder Neural Network (CSANN). The proposed CSANN is developed using a Deep Stacked Autoencoder (DSA) and Deep Neuro Fuzzy Network (DNFN). Additionally, the performance of the implemented technique is evaluated using metrics such as accuracy, True Positive Rate (TPR), and True Negative Rate (TNR). The method achieved a maximum accuracy of 0.921, a TPR of 0.935, and a TNR of 0.921.
ISSN:0950-7051
DOI:10.1016/j.knosys.2025.114301