evaluating lbs privacy in dynamic context

Post on 22-Feb-2016

38 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

Evaluating LBS Privacy In DYNAMIC CONTEXT. Implement report (12/05/2011). Outline. Architecture Implement Merge module Algorithm module Reciprocity module Experiment Conclusion. Outline. Architecture Implement Merge module Algorithm module Reciprocity module Experiment - PowerPoint PPT Presentation

TRANSCRIPT

1

EVALUATING LBS PRIVACY IN DYNAMIC CONTEXTImplement report (12/05/2011)

2

Outline Architecture Implement

Merge module Algorithm module Reciprocity module

Experiment Conclusion

3

Outline Architecture Implement

Merge module Algorithm module Reciprocity module

Experiment Conclusion

4

5

6

Outline Architecture Implement

Merge module Algorithm module Reciprocity module

Experiment Conclusion

7

Outline Architecture Implement

Merge module Algorithm module Reciprocity module

Experiment Conclusion

8

Implemented algorithms Nearest-neighbor ASR (nnASR)

R-Tree Index Different results when run many times with

same input Interval Cloaking

Quad-Tree index Same input – same result

Grid Sorted List Same input – same result

9

Outline Architecture Implement

Merge module Algorithm module Reciprocity module

Experiment Conclusion

10

Checking reciprocity module Input: issuer id + MBR Output: number of users which have same

MBR – real k Algorithm:

Set k_anonymity = 0 Run anonymizing algorithm to get AS For each id ui in AS, run algorithm to get ASi

If AS = ASi then k_anonymity = k_anonymity + 1 If k_anonymity >= k, return true Else return false

11

Problem with reciprocity property

An assumption about anonymizing algorithm: In snapshot, same input same result

Problem Algorithm: same input different results

Example: nnASR

II

1st time 2nd time

12

nnASR: an attack proposed Assumption

K is known Idea

Find the chosen users Its k-nn must be in or be the original MBR

Forecast the candidate issuer For each user in original MBR (exclude chosen

users) Check whether its k-nn include one of chosen users

& expand MBR is equal to original MBR True candidate

13

Illustration k = 4

Candidate

14

Illustration k = 4

Candidate

15

Refine algorithm Just refine value k of request Brute-force:

Increase k until request satisfies reciprocity property

Suitable for algorithms which: Same input same result

Problem: nnASR

16

Outline Architecture Implement

Merge module Algorithm module Reciprocity module

Experiment Conclusion

17

Experiment Implement the algorithms in Java System configuration:

OS: Window 7 Processor: AMD Phenom II X4 B40 3.0Ghz RAM: 2GB

Data: users’ locations in Sanfrancisco with 17000 users

Run 500 tests and take the average to get output values

18

Experiment

4 6 8 10 12 14 16 18 200.000

0.200

0.400

0.600

0.800

1.000

1.200

1.400

1.600

1.800

2.000

Average computation time

GridnnASRInterval

k

Tim

e (m

s)

19

Experiment

4 6 8 10 12 14 16 18 200.000

20,000.000

40,000.000

60,000.000

80,000.000

100,000.000

120,000.000

140,000.000

160,000.000

GridnnASRInterval

k

Perim

eter

(m

)

Average size of the generalized region

20

Experiment

1 2 3 4 5 6 7 8 90

200

400

600

800

1000

1200

Average computation time for checking & refin-ing

GridInterval

k

Tim

e (m

s)

21

Conclusion nnASR algorithm: how to improve

privacy of user? The performance of reciprocity module

top related