Verifiable Summation of Read-Once Formula Specified Data

Securely outsourcing computation to untrusted cloud servers has attracted a lot of attention in recent years. In this paper we consider datasets where every item is labeled by a number of attributes and study the problem of securely outsourcing the summation of a subset of the data items. We propose a verifiable summation on outsourced data (VSOD) model and construct two schemes in this model. Our VSOD schemes allow an input client to store a dataset on a cloud server and allow an output client to offload the summation of a subset of the data items to the cloud server, where the subset can be defined by a read-once formula and a number of indicator functions. This kind of computations can capture many real-world queries to datasets. Our schemes are secure such that no malicious server is able to persuade the output client to accept a wrong result. Our schemes are outsourceable such that the client's computation is substantially faster than the native computation of the summation. Compared with the existing solutions, the outsourceability of our schemes is not based on multiple datasets.