2021 Fiscal Year Research-status Report
Achieving Differential Privacy under Spatiotemporal Correlations
| Project/Area Number |
19K20269
|
|---|
| Research Institution | Kyoto University |
|---|
Principal Investigator |
曹 洋 京都大学, 情報学研究科, 特定准教授 (60836344)
|
|---|
| Project Period (FY) |
2019-04-01 – 2023-03-31
|
| Keywords | differential privacy / spatiotemporal privacy / location privacy / data correlations |
|---|
| Outline of Annual Research Achievements |
Local differential privacy (LDP) has been used widely for spatiotemporal data analysis; however, recent studies demonstrate LDP is vulnerable to manipulation attack under malicious clients. In order to prevent this risk, we propose secure and efficient verifiable LDP protocols to prevent manipulation attacks by leveraging Cryptographic Randomized Response Technique (CRRT) as a building block to convert existing LDP mechanisms into a verifiable version. This work was accepted in 35th Annual IFIP WG 11.3 Conference (DBSec) 2021. Also, we are studying how to enhance the utility of differentially private data release under temporal correlations. Our idea is to design a new post-processing method to leverage the temporal correlations, which is assumed to be public. We had obtained preliminary results that verified the effectiveness of our idea. Currently, we are working on completing the extensive experiments and submitting to an international conference.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
We have achieved the original goal of this project: exploring the issues when applying differential privacy to spatiotemporal data. Besides this, we also obtained some new results and insights regarding the new security vulnerability of existing privacy-enhancing techniques under malicious clients.
|
| Strategy for Future Research Activity |
The extended period will be used to prepare the extensive experiments and submission for the topic of "boosting utility of differentially private stream data release".
|
| Causes of Carryover |
We will need more time to prepare the extensive experiments and submission for the work on "boosting utility of differentially private stream data release".
|
