Improving Safety of an Automotive AES-GCM Core and its Impact on Side-Channel Protection

• Autor(a) principal: Dany José Correia Teixeira
• Data de Publicação: 2020
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: https://hdl.handle.net/10216/132868
• topic: Engenharia electrotécnica, electrónica e informática; Electrical engineering, Electronic engineering, Information engineering

The increase in electronic components and the corresponding increment in the data flow among electronic systems in automotive applications made security one of the main concerns in this sector. The use of IP cores that implement the Advanced Encryption Standard (AES) was seen as a solution to this problem, preventing improper access to vehicle data, through its encryption. The AES algorithm does not currently have any effective vulnerability, but the same does not happen with its implementations, which are subject to side-channel attacks, where information that results from the operation of these implementations is exploited in an attempt to discover the encrypted data. The application of IP cores in the automotive sector requires that the implementations comply with the ISO-26262 standard in order to ensure that their operation does not compromise the vehicle's safety. This compliment implies changes in the core architecture that can influence the characteristics of operation that are normally exploited in attacks. Thus, the development of safety and security components in the automotive sector, which are still considered as independent processes, may be related because safety improvements may cause changes in the system's vulnerability to attacks that can compromise its security. This work aims to develop an architecture capable of reaching the metrics for the highest level of safety certification (ASIL-D), based on an existing architecture, and compare the two architectures in terms of vulnerability to side-channel attacks that exploit their dynamic power consumption. The results show that for the ASIL-D architecture, the identification of points of interest and relevant data on the power consumption traces is more evident, which suggests greater effectiveness of the attacks performed in this architecture.