Computationally Inexpensive cPIR

Computational Private Information Retrieval(cPIR) schemes allow a user to retrieve data from a database privately while exchanging only a small amount of data between the database and the user, where the privacy is based on a cryptographic assumption. All previous cPIR protocols require the database to compute a modular multiplication for every bit of the database on each query, making them too computationally expensive to be practical. In this paper, we propose two protocols that achieve the O(n ) communication complexity of previous protocols, while significantly reducing the amount of computation that must be done per query.