PREDICT: Privacy and Security Enhancing Dynamic Information Monitoring
Abstract
The PREDICT project incorporates security and privacy in DDDAS systems to deliver provable guarantees of privacy and security while ensuring high fidelity for data acquisition, aggregation and analytics. Application scenarios include health surveillance data release, traffic analysis, situation awareness and monitoring, and fleet tracking. A novel two-stage scheme was devised for privacy-preserving task assignment, consisting of global server-side probabilistic assignment by an untrusted server using cloaked locations, followed by feedback-loop guided local optimization using precise participant locations, without breaching privacy and achieving high levels of target coverage with reasonable cost. Once data is collected, privacy preserving data aggregation and modeling with feedback control is performed. This project has developed techniques to deliver high data utility/integrity in aggregated data, with rigorous privacy guarantees such that source data is not disclosed. Finally, in many DDDAS settings, when local participants are mutually untrusted, and for increased responsiveness in the field, algorithms were investigated for secure analytics to be performed without disclosing individual inputs, true participant locations or other sensitive information.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 03, 2015
- Accession Number
- ADA623644
Entities
People
- Li Xiong
- Vaidy S. Sunderam
Organizations
- Emory University