Assurance-Case Driven Framework to Support Cyber-Physical Systems

ABSTRACT

The safety and security of cyber physical systems (CPS) play an important role many safety-critical sectors. The software assurances to cyber physical systems depend on the correctness, resilience and integrity that can be achieved not only before the deployment time but also the run-time. However, there some serious challenges when developing safe and secure cyber physical systems. In this talk, we introduce a novel assurance case driven framework that emphasizes on the following two perspectives.

• The development process perspective. The framework employs assurance case templates that automatically builds an assurance case in the Goal Structuring Notation (GSN) based on the artifacts generated at different phases of software development lifecycle. We apply the Model-Driven Engineering (MDE) as the foundation to implement the automatic generation of assurance cases. Once an assurance case is generated, the framework has the following two features:
  • The Maintenance of assurance cases. Software evolution cannot be avoided in CPSs and how to efficiently maintain assurance cases during the software development lifecycle becomes an important. The framework attempts to establish the cause-effect relation between the assurance case elements and their corresponding artifacts to minimize the cost of the maintenance of assurance cases for CPSs. o The certification of assurance case. The evaluation of an assurance case is manually done by certifiers. To increase the automation in a certification process, the framework adopts some machine learning technique to study the previous certified assurance cases done by a certifier and then employed the Dempster-Shafter (D-S) theory to achieve automatic evaluation of a new assurance case as approximation of the certifier's evaluation without requiring an actual review.
• The run-time perspective. Many CPSs have the self-adaptive feature. To support the assurance case driven feature, the framework connects the key elements of assurance case elements to the implementation of a CPS based on the MAPE-K control loop. When an environment change is detected by monitors, the framework starts the calculation based on the current structure of an assurance case to decide how the current assurance case evolves. The result of calculation is sent back to the CPS that make a corresponding adjustment. The run-time calculation can support various V&V techniques. This is joint work collaborated with Mr. Steve Drager and Dr. Matt Anderson from the AFRL.

BIO

Wuwei Shen received his PhD from the Univ. of Michigan in computer science. He has been working on various techniques such as the Model-Driven Engineering (MDE) to increase automation in software development, implementation, and validation in various application domains. Dr Shen received the Senior Research Award from the National Research Council (NRC) Associateship Programs, the National Academies of Sciences, Engineering, and Medicine in 2017. He also received several fellowships from the Visiting Faculty Research Program (VFRP), the Air Force Research Laboratory's Information Directorate (AFRL/RI) from 2015 to 2018.