Efficient Privacy-Preserving Aggregation Scheme for Data Sets

Many applications depend on privacy-preserving data aggregation schemes to preserve users' privacy. The main idea is that no entity should be able to access users' individual data to preserve privacy, but the aggregated data should be known for the application functionality. In these schemes, each user should encrypt a message and send it to an aggregator to compute and send the ciphertext of the aggregated messages to the decryptor without learning the individual messages. The decryptor should decrypt the ciphertext to obtain the aggregated message. However, the existing schemes are designed to aggregate one type/size of data and it is inefficient to modify them to aggregate messages that have data sets of different data types and sizes. In this paper, we propose an efficient privacy-preserving aggregation scheme for data sets. Unlike the existing schemes that do multibit number addition, the proposed scheme aggregates individual bits. Moreover, comparing to the existing schemes, our scheme has two new features. First, in some applications (such as those that need reporting location information), the aggregator can verify the encrypted messages to detect data pollution attacks without accessing the messages to preserve privacy. Second, our scheme has two types of decryptions; called full and partial. In full decryption, the decryptor can decrypt the whole data set, while in partial decryption, the decryptor can enable some entities to decrypt some data in the set. Our analysis demonstrates that the proposed scheme is secure and can preserve users' privacy. Extensive experimental results demonstrate that our scheme is more efficient than the existing schemes.