Counting people with Wi-Fi
To accomplish the feat of people-counting, the researchers put two Wi-Fi cards at opposite ends of a target area, a roughly 70-square-meter space. Using only the received power measurements of the link between the two cards, and not requiring people to carry Wi-Fi-enabled devices, the approach can estimate the number of people walking in that area. So far, the researchers have successfully tested with up to and including nine people in both indoor and outdoor settings.
"This is about counting walking people, which is very challenging," said Mostofi, a professor of electrical and computer engineering. "Counting this many people in such a small area with only Wi-Fi power measurements of one link is a hard problem, and the main motivation for this work."
This people-counting method relies in large part on the changes of the received wireless signal, according to the researchers. The presence of people attenuates the signal in the direct line of sight between the Wi-Fi cards if a person crosses the line of sight, and human bodies also scatter the signal – resulting in a phenomenon called multi-path fading – when they are not in the direct line-of-sight path. By developing a probabilistic mathematical framework based on these two key phenomena, the researchers have then proposed a way of estimating the number of people walking in the space.
With the near-ubiquity of Wi-Fi in many settings, the researchers’ findings have the potential for many diverse applications. For instance, the ability to estimate the number of people in a given space could be used in smart homes and buildings, so air conditioning and heating could be adjusted according to the level of occupancy. "Stores can benefit from counting the number of shoppers for better business planning," noted Mostofi.
Security and search-and-rescue operations could also take advantage of occupancy estimation. Previous work in the research lab involved imaging stationary objects/humans through walls with Wi-Fi signals, and Mostofi plans to eventually bring the two projects together in the future.
Project home page: Head Counting with WiFi: Counting People Without Relying On Personal Devices