In addition to getting the vaccine and wearing a mask, simply minimizing the size of crowds can help stop the spread of the covid-19 virus. To that end, researchers from the University of California, Santa Barbara, led by Yasamin Mostofi, have found a way to leverage a run-of-the-mill wifi network as a tool for counting the number of people in a given space.
Tools like this already exist, and allow places like airports and shopping malls to keep tabs on visitors and the flow of people through buildings. But they either rely on specialized hardware, like video cameras and sophisticated image recognition, or the interactions of wireless networks with people’s electronics as the devices constantly search for wifi connectivity. That means they’re either expensive or inaccurate, given that even in 2021, not everyone carries a connected smart device.
In 2015, another team of researchers led by Mostofi demonstrated how off-the-shelf wifi hardware could be used to count the number of people in a group without the need for any additional hardware, like them all carrying smartphones. But the technique that was demonstrated had one major limitation: It required all of those people to be walking around in a given space, which created measurable interruptions in a wifi signal. So while it could determine how many people were walking around in a restaurant, for example, it wasn’t able to take into account everyone sitting and eating.
Six years later, Mostofi and her team have demonstrated a new technique that can finally count people who are stationary, or who are at least mostly stationary, because it takes advantage of the fact that people are constantly fidgeting and making tiny unconscious movements. Like the previous technique, a wifi broadcaster and a receiver are placed on either side of a space filled with people who are seated or standing still, and by measuring and tracking the tiny fluctuations in the wifi signal strength, periods of high activity and low activity in the room can be determined.
But the data collected from the wifi receiver doesn’t necessarily tell the whole story. If five people in a room of 10 are all fidgeting at the same time, the data simply shows a period of movement which the researchers call a “Crowd Fidgeting Period” (periods of non-movement are instead called a “Crowd Silent Period”) that has no real information about how many people are actually moving. To create a headcount, the researchers had to develop a new mathematical model, but were actually able to borrow some tools from another field of mathematics that deals with something known as Queuing Theory: the study of waiting lines and queues that can be used to predict wait times.
More details about the mathematical model the team developed can be found in the results of their study which were recently published. After testing the new wifi counting technique in 47 experiments conducted in four different environments with groups of 10 people or less, the accuracy was found to be very high: 96.3% of the time, the estimated number of people seated in a space was either exact or off by one, with accuracy dropping to 90% when the wifi signals were also passing through walls.
As with the previous research, being able to accurately count the number of people in a space has some very useful applications, and not just for quickly figuring out how many people attended a TED talk, or stuck around until the end of a movie. For consumers it could be used to easily augment the intelligence of a heating and cooling system, allowing it to increase or decrease temperatures based on the number of people in a room. During an ongoing pandemic, it could also be used as tool for keeping tabs on how many people are in an establishment, such as a restaurant, in cities where seating capacities are limited to help curb the spread of the virus, automating an otherwise time-consuming task.
