Generalizing Semi-Private Function Evaluation

Bachelor Thesis


Julian Bieber


The cryptographic primitive Secure Function Evaluation (SFE) enables the evaluation of a function public function f (x, y), under input parameters that are provided by different parties without them revealing their specific value. Restricting SFE further by requiring the function to be only known by one party is called Private Function Evaluation (PFE). Furthermore Semi-Private Function Evaluation (SPF-SFE) can help to increase the maximum function size which can be evaluated in a somewhat private manner. One way to achieve SPF-SFE is to split the function into blocks, as was shown in [PSS09]. However their implementation was crafted specifically for Yao’s garbled Circuit protocol. Therefore there are protocols, namely Goldreich- Micali-Wigderson (GMW), by which it could not be evaluated.We provide circuit designs for different Privately Programmable Blocks, which are constructed to work with the GMW protocol as well. Additionally we provide a toolchain to achieve SPF-SFE efficiently. Furthermore we implemented a tool used in the toolchain, which uses the aforementioned designs and can combine them with plain circuits and Universal Circuits (UCs) to build a single programmable circuit. A function designer can use the toolchain to balance out his needs for function privacy with performance.