Secure kNN computation on encrypted databases
read more
Citations
Privacy-Preserving Multi-Keyword Ranked Search over Encrypted Cloud Data
A Secure and Dynamic Multi-Keyword Ranked Search Scheme over Encrypted Cloud Data
Enabling Personalized Search over Encrypted Outsourced Data with Efficiency Improvement
Achieving Efficient Cloud Search Services: Multi-Keyword Ranked Search over Encrypted Cloud Data Supporting Parallel Computing
A Privacy-Preserving and Copy-Deterrence Content-Based Image Retrieval Scheme in Cloud Computing
References
k -anonymity: a model for protecting privacy
L-diversity: Privacy beyond k-anonymity
t-Closeness: Privacy Beyond k-Anonymity and l-Diversity
Privacy-preserving data mining
Related Papers (5)
Frequently Asked Questions (19)
Q2. What are the future works mentioned in the paper "Secure knn computation on encrypted databases" ?
The possibility of integrating different schemes in the SCONEDB model to support a wide range of applications makes EDBMS a practical solution to service outsourcing. A future research issue is the systematic study on different operators that can be supported on an encrypted database w. r. t different security levels and goals. It is possible to extend the attack model to include other aspects, e. g., the amount of available computational power. How to include security goal as another component into the SCONEDB model is a subject for future work.
Q3. How many d+1 points can be recovered from E(DB)?
If there are d+1 points xi (1 ≤ i ≤ d+1) in P such that the vectors (xi,−0.5||xi||2) are linearly independent, then the attacker can recover DB from E(DB).
Q4. What is the way to deal with security threats?
The conventional way to deal with security threats is to apply encryption on the plain data and to allow only authorized parties to perform decryption.
Q5. How many kNN equations can be used to solve M?
Since the authors know P = {x1, x2, ..., xd+1} and the corresponding encrypted values I(xi), the authors can set up the following equations to solve M : Mx̂i = I(xi) where x̂i = (xi,−0.5||xi||2)T for i = 1 to d +
Q6. How many queries do players need to perform to break a DRE?
5.2.4 Query encryption and result decryption0 1 2 3 4 5 650 60 70 80 90 100d'En cryp tion time (ins )For each query, player 2 needs to perform one encryption and k decryptions.
Q7. what is the weakness of the scheme?
A weakness of Scheme 1 is that given an enough number of points in P , a level-3 attacker can set up enough number of equations to solve for the unknowns in the transformation matrix M .
Q8. How many tasks are used to evaluate the performance of the schemes?
The authors evaluate the performance of the schemes under 4 tasks: (i) key generation; (ii) database encryption; (iii) kNN computation and (iv) query encryption and result decryption.
Q9. What are the advantages of cloud computing?
Emerging computing paradigms such as database service outsourcing and utility computing (a.k.a. cloud computing) offer attractive financial and technological advantages.
Q10. What is the key required to the encryption and decryption processes?
A key K is required as a parameter to the encryption and decryption processes (note that a key may contain a number of components, e.g., RSA requires a pair of numbers as the key).
Q11. What is the scalar product of p and q?
The scalar product of p and q (represented by column vectors) can be represented as pT Iq, where pT is the transpose of p and The authoris a d × d identity matrix.
Q12. What is the weakness of the encryp-?
The weakness of this simple method is that the unencrypted query points q̂’s all lie on a d-dimensional hyperplane with the unit vector in the (d+1)-st dimension being the normal of the hyperplane.
Q13. What is the way to find a unique ordered set of Q?
Given a set P = {x1, x2, ..., x|P |} ⊂ DB in a level-2 attacker’s knowledge H, the authors want to find a unique ordered set Q ⊂ E(DB) such that sig(Q) = sig(P ).
Q14. What is the simplest way to solve the transformation matrices?
The equations for solving the transformation matrices are: MT1 p̂a = p ′ a and M T 2 p̂b = p ′ b, where M1 and M2 are two d′ × d′ unknown matrices.
Q15. what is the tradeoff between the two proposed schemes?
there is a tradeoff between Scheme 2, which is resilient to level-3 attacks, and Scheme 1, which allows more efficient query processing.
Q16. How can the authors show that DRE has poor resistance to level-2 attacks?
the authors can show that DRE has poor resistance to level-2 attacks by showing that the attacker can “upgrade” his level-2 knowledge to level-3 using signature linking attack.
Q17. how to set up equations to solve for M?
the attacker can set up equations to solve for M and use Pv to verify the hypothesis: if the recovered database contains Pv, the hypothesis may be correct; otherwise, the hypothesis cannot be true.
Q18. How many known points in P can be broken to break a DRE?
The authors have shown that signature linking attack only requires a small number of known points in P to break a DRE and the attack cost is not expensive.
Q19. what is the scalar product of these two (d+1)-dimensional points?
The scalar product of these two (d+1)-dimensional points can be represented as(p1 − p2)T (rq) + (−0.5||p1||2 + 0.5||p2||2)r = 0.5r(||p2||2 − |p1||2 + 2(p1 − p2)T q) = 0.5r(d(p2, q)− d(p1, q))So, the condition is equivalent to 0.5r(d(p2, q)− d(p1, q)) > 0⇔ d(p2, q) > d(p1, q).