Information-Flow Control by Means of Security Wrappers for Active Object Languages with Futures

Abstract

In this thesis, we experiment with customizing programming languages to enforce privacy and security policies. We enforce privacy and security requirements at the level of a programming language when a program executes. We design a language and enrich it with the essential features to enforce the requirements. Moreover, we model our language with formal methods and prove that programs written in our language do not violate the desired policies. We design the language's syntax and semantics and formalize the operational semantics with mathematical logic, which enables us to reason about the language's properties.The structure of this thesis is described in the following. First, we introduce privacy and security policies that we want to enforce. We choose the GDPR (General Data Protection Regulation), which has strict requirements to protect the individual's privacy when processing personal data. For security, we give an overview of existing language-based techniques to preserve confidentiality and limit access to sensitive data. Second, we state the research questions that we want to handle in this thesis. Third, we introduce the tools and logic that we use for our research methods and modeling our languages. Finally, we present the research papers and relate the research questions to our contributions.The main contributions of this thesis are presented in three research papers. In the first paper, we introduce a programming language with provable guarantees that protects privacy and enforces the GDPR's requirements. The second paper gives an overview of a category of programming languages, called active object languages, that are used to develop distributed systems. In the third paper, we introduce a security mechanism to enforce security in active object languages. We discuss and prove that our language-based approaches are exact when it comes to enforcing policies and restrictions. Moreover, our approaches can be generalized to other languages.