Adapting Privacy-Preserving Computation to the Service Provider Model

There are many applications for secure multi-party computation (SMC), but practical adoption is still an issue. One reason is that the business model of the application does not match the system architecture of regular secure computation. An important business model is that of a single service provider dealing with many customers. Applications with this business model are e.g. auctions or benchmarking. This paper provides SMC in a system architecture for service providers. Furthermore we achieve an interesting performance improvement. Our SMC protocol has a significantly improved complexity if all parties behave semi-honest, but can still deal with a minority of malicious parties at the usual complexity.The solution also relates to distributed algorithmic mechanism design which proposes to build distributed algorithms that implement mechanisms that compute the result given rational players.