Towards Verification of Self-Healing for Autonomous Vehicles

In the scientific community, "Organic Computing (OC)" offers a promising approach to designing and developing highly reliable and cost-efficient systems, one of the main challenges in developing autonomous vehicles. Thereby, OC proposes the implementation of diverse self*-properties to make the system act self-aware and autonomously throughout environment and requirement changes. These properties include, a.o., the self-configuration of the system, self-healing from erroneous and corrupted states, and self-protecting against security attacks carried out by the system independent from human commands. Regarding the insurance of safety, reliability and security, mainly self-healing is introducing new possibilities for complex, cost-intense systems. Self-healing systems can detect, diagnose and repair failures in a self-aware manner, increasing the robustness and operation time without affording maintenance performed by humans. The emerging benefit lies in enhancing the system's safety and security parameters while keeping costs and resources reasonable.

Current research focuses mainly on architectures and functional implementations of self*-properties. However, to our knowledge, modeling and verification approaches for the properties' impact still need to be developed. This is key for OC being accepted beyond the scientific scope. Therefore, this paper performs a literature study on selfhealing and defines its core concept in terms of a generic modeling. This modeling builds the basis for a quantitative verification of self-healing. To show its effectiveness it is implemented in an existing assessment tool, "ERIS", and an example application is presented.

Keywords: Organic computing, Verification, Self-healing, Self-protecting, Safety, Security.

Download PDF