DFB-PPGSQ: Dynamic Forward-Private and Epochal Backward-Private Graph Similarity Matching Query over Encrypted Sensor
Huiying Hou1,2,3, Yucong Ma2, Zisu Zhao2
1The Key Laboratory of Grain Information Processing and Control, Henan University of Technology, Zhengzhou 450001, China.
Sensors (Basel, Switzerland)
|August 13, 2026
Summary
This study introduces a new privacy-preserving graph similarity scheme for dynamic cloud data. It efficiently handles updates and searches while protecting sensitive graph information.
Area of Science:
- Computer Science
- Cybersecurity
- Data Management
Background:
- Graphs are increasingly used in cloud applications like IoT and industrial monitoring.
- Existing privacy schemes struggle with dynamic data (updates, deletions) and maintaining security.
Purpose of the Study:
- To propose DFB-PPGSQ, a dynamic, privacy-preserving graph similarity matching scheme.
- To enable secure similarity search on outsourced graph data without revealing content.
Main Methods:
- Implemented a dynamic forward-private and epochal backward-private graph similarity matching scheme.
- Utilized structured encryption with epoch-local feature tokens and update buffers.
- Employed shuffle-based branch-tree re-randomization for efficiency and security.
Main Results:
- Maintained low server-side filtering latency comparable to static methods.
- Achieved efficient handling of insertions and label updates with minimal overhead.
- Demonstrated low cross-epoch token linkage leakage after refresh in various workloads.
Conclusions:
- DFB-PPGSQ offers an efficient and secure solution for dynamic graph similarity search in cloud environments.
- The scheme balances pruning efficiency with robust privacy guarantees for outsourced graph data.
