Human Activity Recognition using Wireless Signals and Low-Cost Embedded Devices

Developing affordable and privacy-aware smart home applications remains a challenge. Existing camera-based and sensor-based solutions often raise concerns about cost and user privacy. This paper proposes a low-cost WiFi-based sensing system utilizing ESP32 microcontrollers and Jetson Nano edge devices for human activity recognition (HAR) within smart homes. The system captures Channel State Information (CSI) data using an ESP32-based WiFi transmitter and receiver, eliminating the need for expensive hardware while ensuring privacy preservation. A mix of conventional machine learning and deep learning models are evaluated on the obtained dataset, achieving an accuracy of up to 95.57%. The best-performing model is then deployed on the Jetson Nano edge device for efficient activity classification and high throughput. Furthermore, the system seamlessly integrates with existing communication protocols such as Message Queuing Telemetry Transport (MQTT), and provides a user-friendly interface for visualization. We demonstrate the effectiveness of our approach by achieving accurate HAR using a single transmitter-receiver pair, highlighting its practicality and scalability potential. This work aims to further exploration of WiFi sensing technology for smart home applications by enabling cost-effective and privacy-preserving HAR.