Paper 2021/1490
Precio: Private Aggregate Measurement via Oblivious Shuffling
Abstract
We introduce Precio, a new secure aggregation method for computing layered histograms and sums over secret shared data in a client-server setting. Precio is motivated by ad conversion measurement scenarios, where online advertisers and ad networks want to measure the performance of ad campaigns without requiring privacy-invasive techniques, such as third-party cookies. Precio has linear (communication) complexity in the number of data points and guarantees differentially private outputs. We formally analyze its security and privacy and present a thorough performance evaluation. The protocol supports much larger domains than Prio. It supports much more flexible aggregates than the DPF-based solution and in some settings has up to four orders of magnitude better performance.
Metadata
- Available format(s)
- Category
- Cryptographic protocols
- Publication info
- Preprint.
- Keywords
- secure computingsecure aggregation
- Contact author(s)
-
betuldurak @ microsoft com
ckweng @ u northwestern edu
erikan @ microsoft com
kim laine @ microsoft com
melissac @ microsoft com - History
- 2023-08-15: last of 3 revisions
- 2021-11-15: received
- See all versions
- Short URL
- https://ia.cr/2021/1490
- License
-
CC BY
BibTeX
@misc{cryptoeprint:2021/1490, author = {F. Betül Durak and Chenkai Weng and Erik Anderson and Kim Laine and Melissa Chase}, title = {Precio: Private Aggregate Measurement via Oblivious Shuffling}, howpublished = {Cryptology ePrint Archive, Paper 2021/1490}, year = {2021}, note = {\url{https://eprint.iacr.org/2021/1490}}, url = {https://eprint.iacr.org/2021/1490} }