Exploiting Wi-Fi Signals for Human Activity Recognition

Human activity recognition is gaining much attention due to its role in medical alert systems, interactive video games, smart home systems, and many more. One of the main objectives of any human activity recognition system is recognizing the different human activities without affecting them. In this work, we utilize the information embedded in the overflowing Wi-Fi signal to determine which activity is being performed. A dataset obtained by observing 20 subjects performing activities in two different environments adds to this study's credibility. The performed experiments show that an average activity recognition accuracy of 94% is possible.