Application of mmWave Radar Sensor for People Identification and Classification

Area of Science:

• Computer Science
• Electrical Engineering
• Signal Processing

Background:

• Device-free indoor identification is crucial for personalized services.
• Visual methods face limitations due to lighting, clear view requirements, and privacy concerns.
• mmWave radar offers a privacy-preserving alternative for human sensing.

Purpose of the Study:

• To develop a robust, device-free indoor identification and classification system.
• To overcome environmental challenges in object detection and recognition using mmWave radar.
• To enable accurate individual user identification and intruder detection.

Main Methods:

• Utilized mmWave radar technology for data acquisition.
• Applied an improved density-based clustering algorithm for accurate 3D point cloud processing and ground truth extraction.
• Employed a bi-directional Long Short-Term Memory (LSTM) network for classification and identification tasks.

Main Results:

• Achieved an overall identification accuracy of 93.9% for groups of 10 individuals.
• Demonstrated an intruder detection rate of 82.87%.
• Validated the system's effectiveness in varying environmental conditions.

Conclusions:

• The proposed mmWave radar system effectively addresses limitations of visual identification methods.
• The combination of advanced clustering and LSTM networks provides a robust solution for device-free indoor human identification.
• This technology enhances personalized services and security through accurate, privacy-conscious sensing.