Efficient Privacy-Preserving Outsourcing of Large-Scale Geometric Programming

Abstract

Nowadays industries are collecting a massive and exponentially growing amount of data that can potentially promote business innovations. However, it is challenging for resourcelimited clients to analyze their data in a cost-effective and timely way as the data volume keeps growing. With cloud computing, one feasible solution is to analyze the massive data by outsourcing them to the cloud. Nonetheless, clients’ data may contain private information that needs to be kept secret. In this paper, we design a secure, efficient, and verifiable outsourcing protocol specifically for geometric programming, which is one of the most fundamental problems in data analysis with many applications. In particular, a secure and efficient transformation scheme is used to encrypt the original geometric programming problem at the client side and protect its privacy before offloading it, and the gradient projection method is employed to solve the encrypted geometric programming problem in the cloud side. Experiments are conducted on both Amazon Elastic Compute Cloud (EC2) and a laptop to evaluate performance of the designed outsourcing protocol, and the results show the feasibility and efficiency of the protocol.