SLIDE 1 DOLPHIN ATTACK
GUOMING ZHANG, CHEN YAN PRESENTED BY JACOB BEDNARD WAYNE STATE UNIVERSITY CSC6991
SLIDE 2 Overview
- Soundwaves and Digital Signal Processing (DSP)
- Attack Methodology
- Defense Mechanisms
- Demonstration Videos
SLIDE 3
Soundwaves and Digital Signal Processing (DSP)
SLIDE 5
SLIDE 6
SLIDE 7
SLIDE 8
SLIDE 9
SLIDE 10
SLIDE 11
SLIDE 12
SLIDE 13
Attack Methodology
SLIDE 14 Dolphin Attack Major Contributions
- Attackers can use inaudible sounds (>20kHz) to activate and control
voice controllable systems such as cell phones, home entertainment systems, vehicles, etc.
- The attack is sneaky in nature. Device owners may not be aware that
they are being attacked due to the remote distance and inaudible sounds that the attack utilizes.
SLIDE 15 Dolphin Attack…?
What does this have to do with Dolphins?
SLIDE 16 Motivation
- Eaves Dropping (Leak Personal Data, Authentication By-Pass, etc.)
- “Sneaky” Attack
- Defeat the “Air-Gap”
- Drive-By Attack
SLIDE 17 Threat Model
- No Target Device Access
- No Owner Interaction
- Inaudible
- Attacking Equipment
SLIDE 18
Feasibility Analysis (Test Setup)
SLIDE 19
Feasibility Analysis (Exploiting Linear Amps)
SLIDE 20
Feasibility Analysis (Speakers & Microphones)
SLIDE 21
Attack Design
SLIDE 22
Human Voice Samples
SLIDE 23
Text-to-Speech Samples
SLIDE 24
Command Modulation (Conversion to Inaudible)
SLIDE 25
Low-Cost Attack Implementation
SLIDE 26
Test Results
SLIDE 27
Test Results
SLIDE 28
Defense Mechanisms
SLIDE 29 Hardware-Based Defenses
- Microphone Enhancement
- Inaudible Voice Command Cancellation
SLIDE 30 Software-Based Defenses
- Supported Vector Machines
- (Machine learning)
- 100% Success Rate w/ 24 voice samples
SLIDE 31 Demonstration Videos
http://usslab.org/projects/dolphinAttack.html
SLIDE 32 Summary
- Soundwaves and Digital Signal Processing (DSP)
- Attack Methodology
- Defense Mechanisms
- Demonstration Videos
SLIDE 33
Questions / Comments?