What are Acoustic Side Channel Attacks and how is AI used to increase its accuracy?

What are Acoustic Side Channel Attacks and how is AI used to increase its accuracy?

Subject :Science and Technology

Section: Awareness in IT and Computers

Introduction

• Research paper: “A Practical Deep Learning-Based Acoustic Side Channel Attack on Keyboards”
• Published and supported by the ethics committee of Durham University, U.K.
• AI used to decode passwords by analyzing sound from keystrokes
• Highlighted accuracy of Acoustic Side Channel Attacks (ASCA) using deep learning models for laptop keystrokes classification and mitigation

Understanding ASCA

• Side Channel Attacks (SCAs) method for hacking cryptographic algorithms
  • Analyzing auxiliary systems
  • Collecting signals: electromagnetic waves, power consumption, mobile sensors, sound from keyboards, printers
• ASCA: using keyboard sound to analyze keystrokes and leak sensitive information
• Users underestimate the misuse of keyboard sound, not hiding keystroke sounds

Impact of AI on ASCAs

• AI and deep learning increase the risk of side channel attacks
• Laptop models with same keyboard increase ASCA scope
• More microphones near keyboards due to modern technology
• AI-enabled deep learning models can interpret acoustics

Accuracy of ASCA Attacks

• Research by Cornell University, Durham University, University of Surrey, Royal Holloway University of London
• Used audio recordings from Zoom calls, smartphone mics, off-the-shelf equipment
• ASCA attack accuracy: 95% with nearby phone keystrokes
• Deep learning model achieved state-of-the-art accuracy on MacBook Pro keyboard
• Smartphone microphone data achieved 95% accuracy, dropped to 93% with Zoom calls

Historical Context of ASCA Attacks

• ASCA attacks date back to 1950, using acoustic emanations to crack encryption
• 1982: United States NSA declassified documents listing acoustic emanations as compromise source