Sistema de detecção de intrusão em sistemas veiculares: estratégia de desenvolvimento para prevenção de ataques cibernéticos e detecção de anomalias

Abstract

Anomaly detection algorithms play a crucial role in ensuring the security and stability of various systems. These algorithms employ statistical learning techniques and machine learning models to identify unusual patterns and activities within the network to help detect attacks. As a result, technological advances around Artificial Intelligence (AI) have brought new possibilities and challenges, especially in the automotive sector with the development of Autonomous Driving Vehicles (VCA). In the meantime, the objective of this study is to analyze the vulnerabilities of these systems, focusing on detecting anomalies that could compromise their security and efficiency. The justification for conducting this research is given by the growing importance of VCA and the need for rigorous validation of their systems. The methodology adopted consisted of a systematic literature review following the PRISMA method, through which twenty-seven studies were selected for analysis. The results pointed to a critical need for advanced algorithms capable of efficiently detecting and mitigating anomalies. Thus, an anomaly detection system using Decision Tree and Artificial Intelligence Network was applied to the KDD99 database, evaluating its accuracy. It is concluded that security in VCA is an expanding field of research, with anomaly detection being a crucial element for the advancement of technology.