Adapting Security Policy at Runtime for Connected Autonomous Vehicles

Abstract

Models@run.time has appeared as a new paradigm for continuously monitoring software systems execution to enable self-adaptation. This adaptation happens whenever a change, a failure or a bug is introduced within the execution environment. It is done by a causal bidirectional connection between design and execution artifacts supported by feedback-loop information. We developed an approach to integrate a secure lightweight dynamic adaptation of access control policies when new security requirements are introduced at runtime. The proposed runtime adaptation is supported by a fast checking of design models at runtime against security constraints. A vehicle-2-x security case study is used to demonstrate the benefits of our approach.