Advances, Systems and Applications
Table 4 Performance categorization of existing approaches
From: Confidential database-as-a-service approaches: taxonomy and survey
Transmission overhead | Communication rounds | Touched entries | Computation | ||
|---|---|---|---|---|---|
SP | Mediator | ||||
Deterministic Indexes | O(k) | 1 | O(k) | none | low |
Deterministic Indexes (Flattened) | O(k + e) | 1 | O(k + e) | none | low |
Bucketization | O(k + e) | 1 | O(k + e) | none | low |
Bucketization (Flattened) | O(k + e) | 1 | O(k + e) | none | low |
Order-Preserving Encryption | O(k) | 1 | O(k) | none | low |
Searchable Encryption | O(k) | 1 | O(n) | moderate | moderate |
Encrypted B-Trees | O(log(n) + k) | log(n) | O(log(n) + k) | none | moderate |
Encrypted B-Trees (Shuffled) | O(2 ∗e∗log(n) + k) | log(n) + 1 | O(2 ∗e∗log(n) + k) | none | moderate |
Fragmentation | O(k + e) | 1 | O(k + e) | none | none |
Fragmentation (non-colluding SPs) | O(k) | 1 | O(k) | none | none |
Homomorphic Encryption | O(1) | 1 | O(k) | high | high |
Oblivious RAM | Depends on the specific approach | ||||
Example [35] | O(log (n)2+ log(n) ∗k) | log(n) +1 | O(log (n)2+(l o g(n))∗k) | none | moderate |
Example [36] | SP →Client: O(log(n) + k) | log(n) + 1 | O(log (n)2+(l o g(n))∗k) | low | low |
(non-colluding SPs) | SP →SP: O(log (n)2+(l o g(n))∗k) | ||||
Private Information Retrieval | Depends on the specific approach | ||||
Hash Index: O(\({n}^{\frac {1}{3}}) +\) k | 1 | O(\({n}^{\frac {1}{3}}) +\) k | low | low | |
(non-colluding SPs) | B-Tree: O(log(n)\(*\textit {n}^{\frac {1}{3}} +\) k) | log(n) | O(log(n)\(*\textit {n}^{\frac {1}{3}} +\) k) | ||