PrivKV - Key Value Data Collection with Local Differential Privacy

Explore the innovative approach to privacy-preserving distributed data collection in this IEEE Symposium on Security & Privacy presentation. Delve into the concept of Local Differential Privacy (LDP) and its application to key-value data, a popular NoSQL data model. Learn about the PrivKV baseline approach and its limitations, then discover the iterative solutions PrivKVM and PrivKVM+ that enhance estimation accuracy. Understand the optimization strategy for reducing network latency and improving results through virtual iterations. Gain insights into the theoretical analysis and experimental results that validate these novel techniques for protecting user privacy while enabling accurate statistical estimation on sensitive data such as location and app usage.