Publicação
Automated and intelligent hacking detection system
| Resumo: | The Controller Area Network (CAN) is the backbone of automotive networking, connecting many Electronic ControlUnits (ECUs) that control virtually every vehicle function from fuel injection to parking sensors. It possesses,however, no security functionality such as message encryption or authentication by default. Attackers can easily inject or modify packets in the network, causing vehicle malfunction and endangering the driver and passengers. There is an increasing number of ECUs in modern vehicles, primarily driven by the consumer’s expectation of more features and comfort in their vehicles as well as ever-stricter government regulations on efficiency and emissions. Combined with vehicle connectivity to the exterior via Bluetooth, Wi-Fi, or cellular, this raises the risk of attacks. Traditional networks, such as Internet Protocol (IP), typically have an Intrusion Detection System (IDS) analysing traffic and signalling when an attack occurs. The system here proposed is an adaptation of the traditional IDS into the CAN bus using a One Class Support Vector Machine (OCSVM) trained with live, attack-free traffic. The system is capable of reliably detecting a variety of attacks, both known and unknown, without needing to understand payload syntax, which is largely proprietary and vehicle/model dependent. This allows it to be installed in any vehicle in a plug-and-play fashion while maintaining a large degree of accuracy with very few false positives. |
|---|---|
| Autores principais: | Carvalho, Bruno Alves Martins |
| Assunto: | Intrusion detection system Machine learning Controller area network Connected vehicles Sistema de deteção de intrusão Aprendizagem máquina Controller area network Veículos conectados Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática |
| Ano: | 2022 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade do Minho |
| Idioma: | inglês |
| Origem: | RepositóriUM - Universidade do Minho |
| Resumo: | The Controller Area Network (CAN) is the backbone of automotive networking, connecting many Electronic ControlUnits (ECUs) that control virtually every vehicle function from fuel injection to parking sensors. It possesses,however, no security functionality such as message encryption or authentication by default. Attackers can easily inject or modify packets in the network, causing vehicle malfunction and endangering the driver and passengers. There is an increasing number of ECUs in modern vehicles, primarily driven by the consumer’s expectation of more features and comfort in their vehicles as well as ever-stricter government regulations on efficiency and emissions. Combined with vehicle connectivity to the exterior via Bluetooth, Wi-Fi, or cellular, this raises the risk of attacks. Traditional networks, such as Internet Protocol (IP), typically have an Intrusion Detection System (IDS) analysing traffic and signalling when an attack occurs. The system here proposed is an adaptation of the traditional IDS into the CAN bus using a One Class Support Vector Machine (OCSVM) trained with live, attack-free traffic. The system is capable of reliably detecting a variety of attacks, both known and unknown, without needing to understand payload syntax, which is largely proprietary and vehicle/model dependent. This allows it to be installed in any vehicle in a plug-and-play fashion while maintaining a large degree of accuracy with very few false positives. |
|---|
Atividades financiadas
Não foram encontrados projetos financiados.