The trustworthiness assurance project addresses the challenge of certifying ”smart cyber-physical systems” – S-CPS, i.e. systems that are endowed with new capabilities through edge-computing, AI algorithms, sensors and collaboration, to act in more open environments. S-CPS promise to enable increased levels of autonomy, flexibility and improved performance in a variety applications, from industrial manufacturing and emergency scenarios (e.g. fire fighting) to road traffic. S-CPS, which are critical (to society, personal safety, security, etc.) will generally have to be certified to be deemed to be “ready” for deployment. Certification typically involves 3rd party assessment of a product or system and requires the developer to provide an assurance case, as a structured body of evidence that provides a compelling, comprehensible, and valid argument to support the claim that a system satisfies particular dependability properties (e.g., safety, security and reliability), in a given environment.
Existing certification and trustworthiness assurance approaches are however challenged by S-CPS. In the trustworthiness assurance project we are therefore seeking to enhance the understanding of how S-CPS should be developed, verified and validated, to provide a basis for certification, as well as investigating what the actual certification should entail to ensure that a particular S-CPS is ready to deploy with low enough residual risk.
In the project we will develop a framework, methods and test procedures for trustworthiness assurance to address the requirements and complexity of S-CPS. Our goal is to consider the multi-faceted dimensions of trustworthiness including classical dependability properties as well as privacy/confidentiality and transparency. We will pursue systems research related to trustworthiness assessment of the trustworthiness of the digital infrastructures and their integration with devices to form applications. Examples of research directions include how to combine different verification and validation approaches (from simulation to real world testing), and how to deal with continuously updated products for which the assurance also needs to be updated (”trustworthy DevOps”). Another research direction includes run-time monitoring and adaptation (architecture patterns). Tight collaborations with the Industrial Edge Applications and the Edge-AI track are expected.
Contact: Martin Törngren, KTH