Experiments on Detecting and Counting People via Ambient Wi-Fi

Abstract:
This paper investigates how well the presence and number of people in an indoor space can be detected using channel state information from ambient Wi-Fi signals using off- the-shelf hardware. Experiments show that well-known machine learning algorithms like k-nearest neighbors, neural networks, and support vector machines can be trained to detect human presence almost perfectly. In addition, the number of people (up to six) can be estimated with moderate success. Potential applications include smart homes as well as industrial and commercial environments.