Abstract: Techniques are described herein for privately aggregating distributed time-series data. A requestor provides a query sequence to users. Each user evaluates the query sequence on the user's time-series data to determine an answer sequence. Each user transforms its answer sequence to another domain, adds noise, and encrypts it for further processing by the requestor. The requestor combines these encrypted sequences in accordance with a homomorphic encryption technique to provide an encrypted summation sequence. The requestor provides the encrypted summation sequence to at least some of the users, who may in turn provide respective decryption shares to the requestor. The requestor combines the decryption shares in an effort to decrypt the encrypted summation sequence. Decrypting the encrypted summation sequence provides a summation of the encrypted sequences from the users, which may be transformed back to the original domain to estimate a summation of the answer sequences of the users.

Abstract: Techniques are described herein for privately aggregating distributed time-series data. A requestor provides a query sequence to users. Each user evaluates the query sequence on the user's time-series data to determine an answer sequence. Each user transforms its answer sequence to another domain, adds noise, and encrypts it for further processing by the requestor. The requestor combines these encrypted sequences in accordance with a homomorphic encryption technique to provide an encrypted summation sequence. The requestor provides the encrypted summation sequence to at least some of the users, who may in turn provide respective decryption shares to the requestor. The requestor combines the decryption shares in an effort to decrypt the encrypted summation sequence. Decrypting the encrypted summation sequence provides a summation of the encrypted sequences from the users, which may be transformed back to the original domain to estimate a summation of the answer sequences of the users.