The Evolution of In-Vehicle Intrusion Detection Systems through Deep Learning: A Systematic Study

Abstract

The security of in-vehicle networks is jeopardized by the advancement of sophisticated automotive electronics. Historically, intrusion detection systems have been employed to safeguard these networks. Nevertheless, they cannot recognize advanced dangers; hence, recent advancements in artificial intelligence provide more refined and efficient detection systems. This systematic study emphasizes the shift from traditional methods to deep learning, examining the latest deep learning-based intrusion detection systems for in-vehicle networks. We assess the efficacy of various deep learning-based intrusion detection systems in detecting and preventing cyberattacks, including denial-of-service and spoofing, by examining their applicability, performance metrics, advantages, and disadvantages. The future of DL in in-vehicle security is also examined in our assessment, which suggests possible lines of inquiry