Theoretical analysis of security warnings in vehicles and design challenges for the evaluation of security warnings in virtual environments

In this paper, we present an approach for designing security warnings in vehicles for software based security incidents. With this we pursue the goal of reducing safety relevant component failures, which can be caused by manipulated or malicious software. The basis of our work is a theoretical analysis of the correlation of manipulated software (including malware) in automotive systems with the safety relevant failures of system components. We describe the potential of a security warning, which can be presented in time ahead of a traditional safety warning: The latter would only indicate safety-relevant implications that potentially arise later as an implication of the preceding security incident. In this paper we suggest three exemplary icons for a combined security-safety warning. Combined warning means a warning not at the time of a safety-relevant failure but already in the detection of the security-violation (e.g. manipulated software in the vehicle). An essential precondition is a recognition algorithm for such malicious software, which has been examined in previous research like [3]. Based on theoretical analyses, we introduce an exemplary design for the testing of these warnings in a virtual environment, precisely, in a driving simulator. A couple of factors play a central role in such evaluations, such as: perception, reaction of the driver, interpretation of warnings and security awareness. The results can be interpreted in the context of the fundamental aim: the reduction of accidents by security alerts. They thus serve as a recommended course of action for implementation in future vehicles.