Title

Low-Cost Approaches to Detect Masquerade and Replay Attacks on Automotive Controller Area Network

Date of Award

Fall 2016

Project Type

Thesis

Program or Major

Electrical and Computer Engineering

Degree Name

Master of Science

First Advisor

Qiaoyan Yu

Second Advisor

Thomas Miller

Third Advisor

Franklin Rudolph

Abstract

Unlike the early automobiles that were mechanical and isolated, the modern automobile is a much more electronic and connected system. Connectivity to external networks, such as the Internet, is advantageous. However, it has also opened various attack points into the car’s In-Vehicle Network (IVN). The primary IVN used for communication is the Controller Area Network (CAN). The CAN was built in 1986 only for reliable real-time communication at the time. However, the need for security on CAN has become a serious concern with the advancement in automotive technology. In previous work, various types of attacks such as masquerade, replay, denial-of-service, and starvation attacks have been demonstrated to attack a CAN network. This thesis concentrates on low-cost detection techniques against masquerade and replay attacks on CAN. Using the invariables of CAN, we developed methods that achieved latency reduction of up to 68% initially and up to 3 orders in the later stages. Since hardware cost is also a concern, we propose methods that require only 1.8% and 6.2% of additional ROM and RAM, respectively as compared to a baseline CAN configuration. Our methods were also evaluated on the basis of their speed of response after attack detection. We were able to achieve response times of as low as 40μs. Lastly, we evaluated our method on the basis of attack misdetection rate and achieved the rates of the order of as low as 10^−5.

This document is currently not available here.

Share

COinS