1. Qian Chen, Haibo Hu, Jianliang Xu Hong Kong Baptist University Authenticated Online Data Integration Services1

2. 2 Query Integration Server (IS) Result Authenticated Online Data Integration Services Data Sources Client Combining data from multiple sources Providing users with a unified query interface

3. 3 Client Airlines Integration Server (IS) Authenticated Online Data Integration Services CX105$617HK->MEL CX135$617HK->MEL CX105$617HK->MEL CX135$617HK->MEL QF30$594HK->MEL QF98$698HK->MEL MH73$691HK->MEL MH79$699HK->MEL QF30 Price: $594 QF30$594HK->MEL QF98$698HK->MEL MH73$691HK->MEL MH79$699HK->MEL Incorrect results Hacking attack Incomplete search Program bug In favor of sponsor HK -> MEL CX105 Price: $617

4.  Meta-analysis ◦ Life science research (e.g., virus spread and disease control) requires collection of disparate datasets ◦ Example: DataNet ◦ The server may be compromised by cyber attack  Collaborative data fusion ◦ Online collaborative data platforms ◦ Examples: Wikipedia, Wikisensing, Wikidata ◦ May alter some critical results due to political or financial reasons Authenticated Online Data Integration Services4 Authenticated Query Processing Enable clients to verify the correctness of query results

5. 0123 Authenticated Online Data Integration Services5 Service Provider Data Owner Client {1, 3, 4, 5} Merkle Hash Tree (MHT) Sign Dataset Q=[1,1] Verify 0123

6. Authenticated Online Data Integration Services6

7. 7

8. 8

9. 9

10. 10

11. Authenticated Online Data Integration Services11 Authenticated Prefix Tree Extended Queries & Optimizations Extended Queries & Optimizations

12. Authenticated Online Data Integration Services12

13.  Homomorphism ◦ Completeness  Embedded a secret sharing scheme ◦ Seal folding  Generate seals for internal nodes ◦ Update efficient  Cancel out the old seals  RSA based signature ◦ Publicly verifiable Authenticated Online Data Integration Services13

14. Authenticated Online Data Integration Services14 {1, 3, 4, 5} 0 0 1 1 Data Q=[1,1] 00011011 Integration Server Data Sources Client Verify Dataset & Seals

15. Authenticated Online Data Integration Services15 Efficient to update, but may deteriorate under skew distribution

16. Authenticated Online Data Integration Services16 Tightly clustered, but an update may cause cascading tree reconstructions

17. Authenticated Online Data Integration Services17

18.  Dataset: ◦ Gowalla dataset in Stanford Large Network Dataset Collection  6,442,890 user check-ins  1,280,969 unique locations with a non-spatial score ◦ Weather dataset from NWS Cooperative Observer Program  10,000 volunteers report daily weather observation  Server: Dual 4-core Intel Xeon X5570 2.93GHz CPU and 32GB RAM, running GNU/Linux, OpenJDK 1.6  Client: Core 2 Quad processor and 4GB RAM, WinXP  RSA (2048 bits), AES(256 bits)  h(): SHA-1 (160 bits) 18Authenticated Online Data Integration Services

19. 19 The construction cost is linear to the dataset size

20. Authenticated Online Data Integration Services20 Both index trees outperform the naïve solution. R-tree is better than G-tree since it is more compact.

21. Authenticated Online Data Integration Services21 G-tree updates more efficiently than R-tree. Both optimizations improve performance by 20-30%.

22. Authenticated Online Data Integration Services22

23. Authenticated Online Data Integration Services23

24. Authenticated Online Data Integration Services24 Cost is linear to the query range (i.e., the result size )

25. Authenticated Online Data Integration Services25 G-tree deteriorates faster since R-tree is more compact

26. Authenticated Online Data Integration Services26 G-tree wins when query ratio 60%

27. Authenticated Online Data Integration Services27

28. Authenticated Online Data Integration Services28