COVID-19: Bluetooth-tracing and distance measures
The infection risk depends on two parameters: the distance, and the contact time. Recording a contact time is easy; regular scans let us know precisely how long another device has been nearby.
Measuring distance is a challenge. Bluetooth signal is attenuated with distance, but there are hundreds of different devices, designs, and antennas. A device 5 meters away can have a stronger signal than another one just 2 meters away. To this, we see added complexity due to reverberation and interferences.
The OpenTrace developers have measured the decibel attenuation of different mobile phones in a fully anechoic chamber, and are proposing a calibration grid. Pr Hans-Jürgen Meckelburg has recently suggested a better calibration methodology. However, there is still no solution to correctly measure distance when a device is placed in a pocket or handbag.
In their latest technical documentation, Apple and Google are not providing any details on distance filtering. Contact-tracing based on their next-to-being-released feature will not allow the differentiation from 2 meters contact to 5 meters contact. But with these efforts going on to improve distance measurements, we can expect false alerts to diminish in later versions.
Ultrasound can more precisely measure distances and will be more effective to filter cases where a phone is close to another phone just through a wall or a window. It implies emitting sound and using the phone micro, which drains batteries. Using Bluetooth signals first and then calibrating with ultrasound is the best way forward, but will need another Apple and Google joint effort.