An Improved Type System for a Privacy-aware Programming Language and its Practical Applications
Confidential data needs to be processed in many areas, for example when making policy decisions using goverment databases or when providing cloud-based services. Sharemind is a framework for developing privacy-preserving applications which allows data to be analysed without revealing individual values. Sharemind uses a technology called secure multi-party computation. Programs using the Sharemind framework are written in a programming language called SecreC. Sharemind and SecreC are designed to support multiple secure multi-party computation methods which we call protection domain kinds. Different protection domain kinds have different security guarantees and performance characteristics and the decision about which one to use depends on the problem at hand which means SecreC should support different protection domain kinds that solve the needs of different applications. The goal of this thesis is to make it easier to add protection domain kinds to the SecreC language by allowing the programmer to define the protection domain kind data types, arithmetic operations and type conversions in the SecreC language without changing the compiler. The author developed a formal type system for the proposed language extensions, implemented them in the SecreC language compiler, described practical applications, open problems and proposed solutions.
Graduation Thesis language
Graduation Thesis type
Master - Computer Science
Dan Bogdanov, Jaak Randmets