An efficient feature selection strategy based on bio-inspired algorithms for preventing cyber attacks in vehicular networks An efficient feature selection strategy based on bio-inspired

• Published in: International journal of information security Vol. 24; no. 5
• Main Authors: Pedro, Vinicius A., Missima, Gabriel N., Bouzon, Murillo F., Rodrigues, Paulo S.
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 03.09.2025
• Subjects: Coding and Information Theory; Communications Engineering; Computer Communication Networks; Computer Science; Cryptology; Management of Computing and Information Systems; Networks; Operating Systems; Regular Contribution; Feature selection; Bio-inspired Algorithms; Cybersecurity; Intrusion Detection System (IDS)
• ISSN: 1615-5262; 1615-5270
• DOI: 10.1007/s10207-025-01117-w

The escalation in the number of Internet users is paralleled by an increase in malicious attacks targeting both private and public networks. Notably, public networks endure the highest frequency of such malicious attempts. Automotive networks constitutes a quintessential example of public networks and are consequently among the most frequently targeted. Despite this, research focusing specifically on malicious attacks within these networks remains scarce. A prevalent issue encountered within these networks is the infiltration of malicious codes, referred to as malware, into communication protocols, complicating detection efforts. A core challenge in this context is the precise specification of the distinctive characteristics of malicious navigation. Existing studies predominantly identify the presence of intrusions without detailing constituent features. This task is particularly daunting as examining the interplay of numerous features is computationally prohibitive. Consequently, the work proposed herein advocates a solution utilizing novel optimization algorithms inspired by biological mechanisms, known as bio-inspired algorithms. Our findings show that the CAN Identifier (ID) is the most critical attribute for observation, exhibiting an accuracy of 95.39%, and its significance is enhanced when analyzed in conjunction with the data length code (DLC) attribute, resulting in an increased accuracy of 95.59%. To our knowledge, these contributions have not yet been explored in current literature.