Methodologies for Trustworthiness
The empirical determination of system properties, for example by testing software, is very limited in its informative value. Trustworthiness must already be considered in the system specification and in the system design process. This requires new methods that can be used to incorporate the desired level of trustworthiness into the system specification. These methods must be implementable using procedures that can be automated and executed by machines.
The work on these methods aims to design systems that are trustworthy from the outset ("by design") in the long term. To achieve this, BI relies on formal methods, which involve rigorous mathematical considerations and proofs, rather than empirical procedures such as measurements or tests.
As a general principle, the BI pursues a holistic approach that spans multiple phases of the lifecycle of these systems - from standardization, design and implementation through to deployment and runtime.
Our current focus is on methods for the design and implementation of trustworthy systems. Specifically, these are methods for the formal and automatable specification of systems, programming languages and compilers that implement these specifications, and methods for the formal verification of the results generated by the compiler. These topics are dealt with by the groups Trustworthy Data Processing and Verified System Design Automation.
