Replacing Oblivious Computation with Private Search for Context Sensitive Communications
MetadataShow full item record
Context aware applications provide users with an increasingly rich set of services. From services such as interactive maps to restaurant guides and social networking tools, the use of information including location, activity and time can greatly enhance the ways users interact with their surroundings. Unfortunately, the dissemination and use of such information also potentially exposes private information about the user themselves. In this paper, we present Themis, a framework for developing two-party applications capable of making decisions based on context sensitive information without revealing either participants' inputs. Themis uses private stream searching to replace the memory and computationally intensive oblivious computation associated with related techniques. We compare the security guarantees and performance profile of our approach against Fairplay and show not only as much as a 96% improvement in execution time, but also the ability to efficiently run applications with complex inputs on both desktop computers and mobile phones. In so doing, we demonstrate the ability to create efficient context-sensitive applications based on private searching.