Smartphones as trackers
In her research, Fatemeh is studying how people move through public spaces over time. One common method used in crowd monitoring has been detecting the Media Access Control (MAC) address of Wi-Fi-enabled devices. When a smartphone searches for nearby wireless networks, it transmits a request containing a MAC address. Sensors installed in public spaces can detect those MAC addresses, and the information can be used to estimate the number of unique devices at a certain location. By comparing devices detected at different locations, it is possible to estimate how many devices, and potentially people, moved from one location to another.
“If we count detected smartphones, it is possible to estimate the number of individuals; however, not everyone carries a smartphone, and some people carry multiple devices,” Fatemeh says. “But this method has become less reliable. Now, the MAC addresses change dynamically, so one smartphone can send four different MAC addresses, which can make a single smartphone appear as four different ones.”
Is privacy really possible?
This is the problem her research is trying to solve: how can crowd monitoring be performed while preserving individuals’ privacy? She has developed a framework for a camera system that counts people moving across locations over time without identifying who they are. Fatemeh: “The system doesn’t store identifiable images but transforms facial features into privacy-preserving biometric templates.”
But to comply with European privacy laws, she adds that this isn’t enough. Crowd monitoring systems need to be designed to minimise data collection: collecting only information necessary for crowd safety tasks, such as crowd density and movement patterns. Another prerequisite is processing the data directly on cameras or edge devices so that raw video footage doesn’t need to be transmitted to or stored on centralised servers.
She also stresses the difference between privacy and security. “Security is about who can access the data. Privacy is about what we can learn from the data.” Should an incident occur in a stadium, such as a fire or a confrontation between opposing sports fans, a camera-based monitoring system can help law enforcement. Fatemeh: “The system is secure in the sense that only the police can access the data. But privacy can be violated if the data reveals personal identities or behavioural patterns.”
Technical challenges
One of the main challenges during her research was reducing noise, that is, lighting conditions, camera angles, and facial features such as eye movements. All these make a face appear slightly different each time it is captured and, as a result, one person’s face may be counted as several people.
The framework, which she and her research team have developed, uses biometric protection methods to ensure it works reliably despite changes in facial appearance. So far, they have tested the system on publicly available facial datasets, and they need to evaluate it under different situations. But the results already show that camera-based crowd monitoring and privacy protection aren't mutually exclusive.
And as for whether football fans can rest assured that their privacy is protected during the World Cup, there is no straight answer. Technology alone can't guarantee privacy, as much depends on how governments apply these systems and how strictly they are regulated, particularly in the United States, where most matches are due to be held.
