Practical Traffic Analysis Attacks on Secure Messaging Applications - - PowerPoint PPT Presentation

▶
practical traffic analysis attacks on secure messaging
SMART_READER_LITE
LIVE PREVIEW

Practical Traffic Analysis Attacks on Secure Messaging Applications - - PowerPoint PPT Presentation

Feb 19, 2024 94 likes •517 views

Electrical and Computer Engineering Department Practical Traffic Analysis Attacks on Secure Messaging Applications Alireza Bahramali, Ramin Soltani, Amir Houmansadr, Dennis Goeckel, Don Towsley University of Massachusetts Amherst Instant

slide-1
SLIDE 1

Practical Traffic Analysis Attacks on Secure Messaging Applications

Alireza Bahramali, Ramin Soltani, Amir Houmansadr, Dennis Goeckel, Don Towsley

Electrical and Computer Engineering Department

University of Massachusetts Amherst

slide-2
SLIDE 2

Instant Messaging is Popular!

❖ Over 2 billion people use Instant Messaging (IM) applications ❖ Used to exchange various types of messages

2

slide-3
SLIDE 3

❖ A variety of IM services: Telegram, WhatsApp, Signal ❖ Most IMs have centralized structure ➢ All the communications are relayed through IM provider servers

3

Typical IM Providers

IM Server

slide-4
SLIDE 4

❖ Various types of communication: ➢ One-to-one communication ➢ Group communication ➢ Channel communication: admins and members

4

Typical IM Providers

IM Server

Admins Members

slide-5
SLIDE 5

❖ Extensively used to exchange politically and socially sensitive contents ❖ Therefore, IM services are attractive targets for government and corporation surveillance

5

IM Communications are Sensitive

slide-6
SLIDE 6

6

Examples

slide-7
SLIDE 7

7

How Confidential Are IMs?

The good news: content is protected by Encryption, End-to-Middle or End-to-End

IM Server Client Client

The bad news: traffic patterns leak information

slide-8
SLIDE 8

How Patterns Leak?

Admin Member1 Member 2 Member 3

8

slide-9
SLIDE 9

❖ This is a fundamental vulnerability!

➢ Major IM services do not obfuscate traffic patterns because it’s expensive

9

Objective of this study: investigate the threat of traffic analysis to popular IM services

slide-10
SLIDE 10

Our Attack

Goal Identify participants

  • f a target IM communication

Traffic Analysis

Adversary A surveillance organization

No need to cooperate with IM server

Timing Size

Meta-data Identity of IM users

10

slide-11
SLIDE 11

11

Attack Scenario

Surveillance Area

IM Server

Adversary observes target user traffic A d v e r s a r y

  • b

s e r v e s t a r g e t c

  • m

m u n i c a t i

  • n

t r a f f i c a s g r

  • u

n d t r u t h

Target User

Target channel: “Let’s protest”

slide-12
SLIDE 12

12

Adversary Ground Truth

IM Server

Adversary observes target user traffic A d v e r s a r y

  • b

s e r v e s t a r g e t c

  • m

m u n i c a t i

  • n

t r a f f i c a s g r

  • u

n d t r u t h

Adversary wiretaps an identified member/admin. Adversary joins the target channel as an admin. Adversary joins the target channel as a member.

Target User

Target channel: “Let’s protest”

slide-13
SLIDE 13

13

Target User

Surveillance Area

IM Server

Adversary observes target user traffic A d v e r s a r y

  • b

s e r v e s t a r g e t c

  • m

m u n i c a t i

  • n

t r a f f i c a s g r

  • u

n d t r u t h

Target User

Target user is the admin of the target communication Target user is the member of the target communication

Target channel: “Let’s protest”

slide-14
SLIDE 14

Outline

❖ Modeling IM traffic: We established a statistical model for regular IM communications ❖ Design attack algorithms: We use hypothesis testing to design attack algorithms ❖ Experiments: We perform experiments on Telegram, WhatsApp, and Signal ❖ Countermeasures: We design and implement an

  • pen-source countermeasure system called IMProxy

14

slide-15
SLIDE 15

Outline

❖ Modeling IM traffic: We established a statistical model for regular IM communications. ❖ Design attack algorithms: We use hypothesis testing to design attack algorithms ❖ Experiments: We perform experiments on Telegram, WhatsApp, and Signal ❖ Countermeasures: We design and implement an

  • pen-source countermeasure system called IMProxy

15

slide-16
SLIDE 16

Modeling IM Traffic

16

❖ Deriving theoretical bounds on our traffic analysis algorithms. ❖ Generating synthetic IM communication. ❖ Dataset: Traffic patterns of 1000 Telegram channels, each for 24 hours.

slide-17
SLIDE 17

17

Modeling IM Traffic

Inter-Message Delays (IMD) Message Sizes Communication Latency Message Types

IM Features

slide-18
SLIDE 18

Outline

❖ Modeling IM traffic: We established a statistical model for regular IM communications ❖ Design attack algorithms: We use hypothesis testing to design attack algorithms ❖ Experiments: We perform experiments on Telegram, WhatsApp, and Signal ❖ Countermeasures: We design and implement an

  • pen-source countermeasure system called IMProxy

18

slide-19
SLIDE 19

19

Attack Algorithms

Event-Based Algorithm Shape-based Algorithm

slide-20
SLIDE 20

Attack Algorithms: Event-Based

Target User

20

1- Event Extraction 2- Correlation Function 3- Comparing to a Threshold Event MATCH! If two events are close enough:

slide-21
SLIDE 21

21

Hypothesis Testing

slide-22
SLIDE 22

Attack Algorithms: Shape-Based

Target User

22

1- Event Extraction 2- Traffic Normalization 3- Correlation Function Event 4- Comparing to a Threshold

Cosine Similarity Traffic Bars

slide-23
SLIDE 23

Outline

❖ Modeling IM traffic: We established a statistical model for regular IM communications ❖ Design attack algorithms: We use hypothesis testing to design attack algorithms ❖ Experiments: We perform experiments on Telegram, WhatsApp, and Signal ❖ Countermeasures: We design and implement an

  • pen-source countermeasure system called IMProxy

23

slide-24
SLIDE 24

❖ We perform experiments extensively on Telegram, WhatsApp, and Signal ❖ We use patterns of 500 channels. ❖ Scenarios ➢ Identifying Admin of a Telegram channel ➢ Wiretapping an identified user (one-to-one)

24

Experimental Setup

slide-25
SLIDE 25

Attacks’ Performance

25

Event-based algorithm Shape-based algorithm

Even with 15 minutes of traffic both algorithms have 94% confidence while FP rate is 0.001

slide-26
SLIDE 26

We compared our work with DeepCorr

26

Why Not Deep Learning?

We perform better than DeepCorr for smaller false positive rates!!? 1- IM flows are sparse. 2- IM flows are less noisy.

slide-27
SLIDE 27

Outline

❖ Modeling IM traffic: We established a statistical model for regular IM communications ❖ Design attack algorithms: We use hypothesis testing to design attack algorithms ❖ Experiments: We perform experiments on Telegram, WhatsApp, and Signal ❖ Countermeasures: We design and implement an

  • pen-source countermeasure system called IMProxy

27

slide-28
SLIDE 28

How to defend?

1- Using circumvention systems: Tor, VPN

They are not effective without any background traffic.

Event-based detector

28

slide-29
SLIDE 29

❖ A proxy-based

  • bfuscation system

❖ No IM cooperation required ❖ Can be applied to any IM service just by proxy the IM traffic through it

29

IMProxy

❖ Algorithms: ➢ Adding delay ➢ Adding dummy packets ❖ Main components: ➢ Local proxy ➢ Remote proxy

slide-30
SLIDE 30

30

How It Works?

IM Server sender (admin) receiver (member) Adversary Watching Adversary Watching Local proxy Padding packets Local proxy Not observable by adversary Removing padded packets Remote proxy Remote proxy Padding packets Adding delay Removing padded packets

slide-31
SLIDE 31

Evaluating IMProxy

❖ Latency: A Laplacian distribution with parameter ❖ Adding dummy packets based on a Uniform Distribution

31

❖ SOCKS5 proxy ❖ Event-based attack With 10% bandwidth

  • verhead, we

have 30% decrease in confidence

slide-32
SLIDE 32

Conclusions

❖ We show that despite the use of encryption, popular IM applications leak sensitive information about their client’s activities. ❖ The reason is that IMs do not use any obfuscation algorithms because it is expensive ❖ We hope that our results warn IM providers to take proper measures

32

Thanks to

slide-33
SLIDE 33

33

How It Works?

SIM Server Local proxy Remove Padded packets Local proxy Padding packets Remote proxy Remove Padded packets Remote proxy Padding packets and adding delays

S u r v e i l l a n c e A r e a

slide-34
SLIDE 34

34

A Fundamental Vulnerability

We show that despite the use of encryption, popular IM applications leak sensitive information about their client’s activities.

Why? Obfuscation of traffic is expensive for IM

  • perators.

How? Merely watching encrypted IM traffic. (Traffic Analysis)

slide-35
SLIDE 35

How to defend?

2- Using IMProxy

IMProxy: A proxy-based obfuscation system

Obfuscate timings by adding delays Obfuscate sizes by adding dummy traffic

How it works?

35

slide-36
SLIDE 36

Evaluating IMProxy

❖ Evaluating against IMProxy aware adversary ❖ Adversary trains a classifier on traffic flows

36

slide-37
SLIDE 37

Attack Algorithms: Shape-Based

Target User

Cosine Similarity

37

slide-38
SLIDE 38

Evaluating IMProxy

  • Latency: A laplacian distribution with parameter
  • Adding dummy packets based a Uniform Distribution
  • SOCKS5 proxy

Oblivious adversary IMProxy-aware adversary

38

slide-39
SLIDE 39

Generalizing to other IMs

Viber Signal WhatsApp Telegram

Messages in IMs have the same shape of traffic They appear as bursts of packets

39

slide-40
SLIDE 40

40

Telegram

❖ 200 million monthly active users. ❖ Most users are in countries with strict media regulations.

Iran Russia

❖ Telegram consumes 60! percent of Iran’s Internet bandwidth! ❖ It has the concept of channels.