Privacy-Safe Network Trace Sharing via Secure Queries

Members

Problem Statement

Contemporary network researches are difficult, because

• Users don't want their information to be leaked, therefore ISPs try to avoid such thing from happening by sanitizing traces or not publishing at all.
• Researchers, on the other hand, need rich data for their research purposes.
• Researchers also will benefit from a more powerful tool that is easy to use with more designated features for trace analysis.

Trol/Patrol

Our solution: Privacy-safe network trace sharing framework, with privacy and security in mind via Trol and Patrol.

Patrol is designed to solve the problems mentioned above. It provides a framework that securely stores traces and provide a query interface for users, and return results that are fine-tuned to avoid leaking sensitive data, while still with enough statistical information for research purposes. Queries are expressed in an SQL-like language, Trol, that supports most common trace processing primitives.

Trol overview

• A trace analysis tool: support common data query functions including grouping, keeping, foreach, conditional/relational/arithmetic operations, output in histogram, CDF, etc.
• A privacy enforced tool: protect data privacy using enhanced K-Anonymity, with protection against tracker and faker.

Software

Trol/Patrol are in active development, and our source code will be publicly available soon.

Publications

Presentations

Acknowledgment