mmWave sensor

With the mmWave project we're investigating a privacy-friendly alternative for camera surveillance in the city.

With the mmWave project we're investigating a privacy-friendly alternative for camera surveillance in the city. Partners in this project are TU Delft, City of Rotterdam and beep beep. To illustrate, the corona crisis makes it more urgent to measure crowds in popular places. The mmWave sensor sees moving objects (such as people) as a collection of dots in space and cannot be used to identify individuals. This limitation exists in the hardware of the sensor and cannot be removed by updating the software on the sensor. Thus, the mmWave sensor registers crowdedness, whilst collecting as little data as possible, to preserve privacy and to adhere to the ambition of data minimization.

After undergoing some revisions, the mmWave sensor is returning to the Marineterrein in October. The following months, we are testing if the sensor – with renewed housing – works properly in the outside area of the Marineterrein. In a next phase, the measurements of the sensor will be compared to the current crowdmonitoring system.

mmWave technology

Millimeter-wave radar technology is a contactless sensing technology for detecting objects and providing the range, velocity and angle of those objects (in our case humans). mmWave sensors transmit signals with a wavelength in the high frequency spectrum between 30GHz and 300GHz, also called the millimeter (mm) range. In the case of this pilot, a mmWave sensor with a low ‘resolution´ (step size) selected. This makes it practically impossible get to any identifiable information from the device and therefore it is a privacy-conscious sensors.

By collaborating with the Responsible Sensing Lab, the mmWave project is testing the knowledge of TU Delft in the real world in order to monitor crowd movements in a privacy-friendly way.”
— Remy van Rooijen, Technical product developer en medeoprichter van beep beep

Video

This video shows a radar image of what the mmWave sensor observes. When the algorithm decides the dots form a block, it detects a person (who can not be identified).

Watch video