The source code of the Corona warning app is available to the public, but questions remain as to how exactly the Bluetooth interface, which is developed directly by Apple and Google, works in the background. We therefore asked those responsible at the Fraunhofer Institute. The institute started work on a variant of the German Corona app that works using Bluetooth distance estimation as early as March.
The concept outlined relied on central data storage instead of decentralized as in the current proposal from Google and Apple; Steffen Meyer, head of the Cooperative Systems and Locating Group department at the Fraunhofer Institute for Integrated Circuits (IIS), confirms to us that the knowledge gained from the work on the early version of the Corona app also flowed to the partners of Google and Apple. The Fraunhofer Institute also tested the app for its reliability of the distance estimation before it started.
When asked why current Corona apps rely almost exclusively on Bluetooth as a technology, when it wasn’t designed for it, the expert answers that besides GPS, Bluetooth is almost everywhere in current smartphones. In order for the digital solutions to fight pandemics to work, users must be able to install them without any problems. This is only possible if the app uses a common technology such as Bluetooth.
In addition, Bluetooth is much more accurate than GPS when it comes to estimating the distance, says Meyer. With GPS, the error when determining a position can amount to ten meters, with two participants, i.e. users for whom the distance is to be measured, the error increases to 20 meters. Bluetooth, on the other hand, has always been designed as a near-distance technology, a Bluetooth module sends its field signals within a radius of 20 meters, the error in the distance estimate is a few decimeters, not several meters.
The technical background of the distance estimation is explained relatively simply. According to the Bluetooth specifications of the Covid 19 interface, a Bluetooth module in the smartphone sends data packets in the size of 15 to 20 bytes four times per second. Such a data packet contains, in addition to the rolling proximity identifiers, that is to say identifiers of the device which can be changed continuously, data on the signal strength (in dBm) with which the device sends these data packets. Once every five minutes, the device scans the area with the Corona warning app and receives the data packets from the smartphones from the immediate vicinity. It measures the signal strength with which these data packets have arrived at the target smartphone, the difference in strength is converted into the distance between the devices, because the field strength decreases with the square of the distance.
Steffen Meyer explains that with corona apps only the limit is really important, that is, whether another user is closer or farther than the 1.5 meters specified by the RKI. That is why the Fraunhofer Institute does not speak of distance measurement, but of distance estimation. However, together with the length of time that the two users are below the distance limit, the risk of infection can be assessed.
According to Meyer, the current app can correctly estimate the distance in 80 percent of cases. That would have been several tests in the test center L.I.N.K. shown by Fraunhofer IIS. The researchers organized a number of pre-tests during the development of the app, and the IIS team tested the Bluetooth interface for practicality for just under a week before it started.
The scientists assessed reliability and accuracy in everyday situations such as a trip on public transport, in a queue or at a party. After these tests, Steffen Meyer can reveal which popular location does not work so well with the Bluetooth distance estimation: in the back pocket. The body of the user serves as a shield for the Bluetooth signal, water is known to have a medium potential for interference for Bluetooth and WLAN signals, since it simply scatters and partially absorbs the radiation. On the other hand, according to Meyer, having a smartphone in your hand or ear, or carried in a handbag helps a lot.
The Corona-Warn-App is probably the most tested app to date, because in addition to the voluntary developers on Github, the Fraunhofer Institute dealt with the functionality of the app and the underlying interface. What emerged from this is certainly important for the current fight against pandemics, but comparable concepts can continue to underpin research in sociology or economics in the future, without having to forego the privacy and privacy of the participants. (Macworld)