Privacy protection framework for face recognition in edge-based Internet of Things

Area of Science:

• Computer Science
• Cybersecurity
• Artificial Intelligence

Background:

• Edge computing (EC) and the Internet of Things (IoT) offer solutions for resource limitations in face recognition.
• However, data privacy leaks remain a significant concern in these systems.
• Existing solutions often address only specific stages of data processing.

Purpose of the Study:

• To propose a comprehensive privacy protection framework for edge-based face recognition (EFR) systems.
• To address privacy concerns throughout the entire lifecycle of face data.
• To enhance the security of face images and training models during transmission between edge devices and the cloud.

Main Methods:

• Development of a general privacy protection framework for EFR systems.
• Design of a local differential privacy (LDP) algorithm based on feature information proportion differences.
• Integration of identity authentication and hash technology for data acquisition security.

Main Results:

• The proposed framework effectively protects face data privacy throughout its entire lifecycle.
• The LDP algorithm achieves a superior balance between data availability and privacy protection.
• Numerical experiments demonstrate the method's advantage over non-privacy protection and equal budget allocation.

Conclusions:

• The proposed framework provides a robust solution for privacy protection in edge-based face recognition.
• The novel LDP algorithm enhances security without significantly compromising data utility.
• This research contributes to the secure and privacy-preserving deployment of IoT-based face recognition systems.