Privacy Protected Spatial Query Processing

Transcription

1 Privacy Protected Spatial Query Processing Slide 1 Topics Introduction Cloaking-based Solution Transformation-based Solution Private Information Retrieval-based Solution Slide 2 1

2 Motivation The proliferation of mobile devices Smartphones, laptops, tablets, etc. The popularity of positioning technologies GPS, Galileo, etc. Privacy threat from accessing Location-Based Services e.g., Find me the closest ATM (bank). How to protect mobile users location privacy? Slide 3 Presentation Outline Introduction Cloaking-based Solution Transformation-based Solution Private Information Retrieval-based Solution Slide 4 2

3 System Architecture Slide 5 System Architecture (Cont.) Access Points Tablets, cell phones, laptops, etc. Location-based Service Providers e.g., Find my nearest gas station - users have to reveal their identity. Location Cloaker A trusted server which implements K-Anonymity mechanisms and manages the location information of users. User privacy policy K-anonymous and the minimum cloaked region size. Need new query processing algorithms. Slide 6 3

4 K-Anonymity The mechanism blurs the identities of K users [Swe02]. One trusted server (i.e., the location cloaker) is needed to cloak K users locations for protecting location privacy. Slide 7 The Data Structure for Managing Mobile User locations Slide 8 4

5 Algorithm for Nearest Neighbor Queries Slide 9 Algorithm for Nearest Neighbor Queries (Cont.) Slide 10 5

6 Solution for Road Networks Slide 11 Solution for Road Networks (Cont.) Slide 12 6

7 Solution for Road Networks (Cont.) Slide 13 Privacy Protected Nearest Neighbor Query (k = 1) Slide 14 7

8 Privacy Protected Range Query Slide 15 Topics Introduction Cloaking-based Solution Transformation-based Solution Private Information Retrieval-based Solution Slide 16 8

9 Preliminary We observed that one-way function can be an ideal mechanism to preserve users privacy. A one-way function is easy to compute but difficult to invert, meaning that some algorithms can compute the function in polynomial time while no probabilistic polynomial-time algorithm can compute an inverse image of the function with better than negligible probability. Slide 17 Preliminary (Cont.) The applied transformation function has to allow fast computation of its preimage with special information, called trapdoor. In mathematical terms, if f is a trapdoor function there exists some secret information y, such that given f(x) and y it is easy to compute x. Multiplication of two large prime numbers. Slide 18 9

10 Space Encryption A space-filling curve is a continuous curve, which passes through every point of a closed space. The important property of these curves is that they retain the proximity and neighboring aspects of the indexed data. The paper investigates the applicability of space-filling curves as ciphers for preserving location privacy of mobile users. Slide 19 Space Encryption (Cont.) A user applies space-filling curves to encrypt the locations of spatial objects in a 2-D space before sending requests to service providers. Since the user knows the curve parameters (i.e., the curve order and orientation), the query results returned from the service provider can be decrypted with the corresponding curve order and orientation as the trapdoor. Slide 20 10

11 Space Encryption (Cont.) Slide 21 Space Encryption (Cont.) For a two dimensional space, we define as the m th order Hilbert curve with the o th orientation. We can formalize the relationship as where x and y are the coordinates of a 2-D space point. Slide 22 11

12 Query Evaluation Range Query Slide 23 Query Evaluation - knn Slide 24 12

13 Improving query processing accuracy Slide 25 Tradeoff Modeling Slide 26 13

14 System Architecture Slide 27 System Architecture (Cont.) The number of encryption keys to be stored in a client? The relationship between encryption key number and location privacy. Encryption key updating frequency. Slide 28 14

15 Topics Introduction Cloaking-based Solution Transformation-based Solution Private Information Retrieval-based Solution Slide 29 Private Information Retrieval (PIR) A user needs to gain access to a specific record of a database but does not want to reveal the record in which s/he is interested. A PIR protocol allows a user to retrieve the i th record from a database of size n stored at an untrusted server, without revealing i to the server. Theoretical PIR and practical PIR. Slide 30 15

16 Theoretical PIR The client and server follow a secure twoparty computation which allows the client to privately retrieve the i th bit from a bit string of size n owned by the server. Slide 31 Practical PIR Slide 32 16

17 Practical PIR (Cont.) Slide 33 17