1. Statistical Identification of Encrypted Web-Browsing Traffic
Qixiang Sun
Stanford University
Daniel R. Simon, Yi-Min Wang, Wilf Russell, Venkata N. Padmanabhan, Lili Qiu
Microsoft Research

2. Outline Motivation & Problem Intuition Hypothetical Attacker
Attacker’s Success Rate
Countermeasures
Conclusion

3. Anonymous Web Browsing
Protect personal information from Attacker’s Inference
Medical (Online support group)
Questionable Activities
Question: Is this REALLY anonymous? R1 R2 R3 R4

4. What’s Different? In anonymous Web browsing Implication:
The chain of routers are used for both sending and receiving data
Can link HTTP requests and responses!
The target Web pages are publicly accessible
Responses are known!
Implication:
The first link/router is an exploitable weakness.

5. What Information is Available?
HTTP Get
Response
Browser
1st Router
Number of objects
Object sizes
Ordering of the objects
Delay between packets R1 R2 R3 R4

6. Intuition
Number of objects and object sizes are sufficient to identify a Web page!
On average, a Web page has 11 objects with each object yielding 8.4 bits of information
8.4*11 – log2(11!)  67 bits  1020 possibilities!!
Currently, there are about 109 Web pages

7. An Hypothetical Attacker
List of target
Sensitive sites URLs
Programmatic
Access to URL
& Traffic recording
Traffic pattern
Construction &
Database update
Traffic
Pattern
Database R1
Traffic recording
& Pattern construction
Traffic
Pattern
Browser
History
Similarity scores
Calculation
Decision
module
Negative
Positive

8. Guts of the Pattern Matching
Given two multisets of object sizes S1 and S2
Sim(S1, S2) = S1  S2 / S1  S2
Decision module uses an absolute threshold.
Traffic
Pattern
Database
Similarity scores
Calculation
Decision
module
For example:
S1 = {3KB, 3KB, 5KB}
S2 = {3KB, 5KB, 5KB}
Sim(S1, S2) =
= 0.5
| {3KB, 5KB} |
| {3KB, 3KB, 5KB, 5KB} |

9. Experiment Setup
Approximately 100,000 Web pages in total (URLs obtained from the Open Directory Project).
The hypothetical attacker chooses about 2200 pages as target pages.
Goal: Can these 2200 pages be identified without causing many false positives?

10. What is a Success and Failure?
Successful Identification:
A target page passes the similarity threshold and is not confused with other pages in the target set.
False Positive:
A non-target page is incorrectly identified as one of the target pages.
Potential False Positive:
A page passes the similarity threshold when compared with a single selected target page.

11. Attacker’s Success Rate
A threshold of 0.5 is sufficient.
80.4%
Is this small enough?
2.1%

12. A Detailed Look Inside False-positives are NOT generated uniformly!
HTTP 404s
Common-looking pages
0-identifiable pages

13. Dynamism in Web Pages Most pages are relatively static
One-day-old pattern database is sufficient

14. Countermeasures Padding Morphing Mimicking Individual objects
Add random-sized objects
Morphing
Pipelining the HTTP GET requests
Pre-fetching
Mimicking
Common templates or Web-hosting services

15. Padding Object Size Linear – Nearest multiple of padding size
Exponential – Nearest power of 2

16. Padding Random Objects

17. Two-chunk Pipelining
Approximately 36% of the target pages are 0-identifiable.
Very close to the theoretical limit of 1/e (assuming traffic patterns are random)
Implication: Can harness the total entropy in the Web page traffic patterns.

18. One-chunk Pipelining

19. Conclusion Encrypted Web browsing can be identified by
the target page’s “unique” traffic pattern.

21. Linear Padding

22. Exponential Padding

23. Pad Random Objects