The tech experts behind Germany's Coronavirus warning app

COVID-19 has brought about some big changes. And that includes the speed at which we conducted the research that enabled us, together with T-Systems and SAP and on behalf of the Robert Koch Institute (RKI), to unveil a functional app in June 2020. We worked with our usual tenacity and perseverance, and didn’t let the intense debate about data protection and availability hold back the pace of innovation: by the end of 2020, Germany’s Corona-Warn-App had been downloaded over 24 million times; when it comes to data protection, the app is as secure as it gets.

 

Since early March 2020, our Fraunhofer expert group has been demonstrating just how quickly an idea and a promising solution can unfold in the face of a challenge. Steffen Meyer, who manages the Corona-Warn-App project at Fraunhofer IIS, and his team have been heavily involved in the provision of the app’s Google/Apple exposure notification interface and in testing the technology. They have also been continuously adapting and updating the interface on the basis of expert advice and know-how gained from their research and development activities in the field of proximity detection for wireless technology, such as Bluetooth and WLAN. The interface makes it possible to determine the length and proximity of a contact. We discussed all this in an interview with Meyer.

Mr. Meyer, how did the project come about and how quickly did you and your team have to get started with the first development steps?

Steffen Meyer: A call came through to the director of our institute from the Fraunhofer Institute for Telecommunications, Heinrich Hertz Institute, HHI – and that was our cue to begin. As Bluetooth Low Energy specialists, we were asked to be part of the project team for a coronavirus warning app. From that point, we were all on board and we set to work. The goal was to use existing smartphone technology that everyone carries with them almost all the time. Introducing new technology wasn’t an option as it would have taken too long to develop and roll out across the population.

What were the particularities of using Bluetooth as a measurement tool?

We’ve been working in the field of proximity estimation using Bluetooth Low Energy-based signal strength measurement for more than ten years and have brought this technology to market in applications for manufacturing, logistics, transportation, trade shows and museums. Our experience meant we were well placed to assess the capabilities and limitations of Bluetooth based estimates. It was obvious that the technology could be useful here and, what’s more, that it was the only tool available to the majority of the population that was suitable for fulfilling the “proximity” and “duration” detection criteria of the Robert Koch Institute’s RISK score as a way to break chains of infection more quickly. One of the challenges that arose in our discussion of the app was about precision and how to make it absolutely clear to everyone that it wasn’t a question of making measurements down to the nearest centimeter.

How did your team contribute to the development?

The most challenging task lays in analyzing and optimizing how the measurement data was acquired and processed behind the exposure notification interface, which was necessary for Google and Apple to launch the app on Android and iOS, respectively. At the critical stage of the project, there were repeated discussions, tests and consultations taking place at different times in the team with T-Systems and SAP, but also with Google and Apple developers. It was a mammoth task and we were working pretty much around the clock, but it was also exciting. My team at Fraunhofer IIS is made up of Wi-Fi/Bluetooth specialists and system design and analysis experts. We worked with the necessary degree of scientific diligence but also with great commitment. We all knew how much was riding on this project.

The key part for us was the technological »translation« of the RKI risk model into the configuration parameters of the exposure notification interface for the various smartphones, plus the constant fine-tuning of the interface in order to map this model as accurately as possible. Other vital contributions from my team were the accompanying and comparative tests to implement the interface and new procedures for calibrating smartphone types.

Video "Corona Warn App - Tests at Fraunhofer IIS"

Privacy warning

With the click on the play button an external video from www.youtube.com is loaded and started. Your data is possible transferred and stored to third party. Do not start the video if you disagree. Find more about the youtube privacy statement under the following link: https://policies.google.com/privacy

What hurdles did you have to overcome in the app’s development?

I can give you two examples. Because timing of Bluetooth Low Energy signals isn’t coordinated, the receiver needs to be »listening« all the time to make sure it captures all of the Bluetooth Low Energy signals in its surroundings. So we proposed a method that enables longer and more frequent measurement intervals, and worked with Google and Apple to define it. This is one of the key features of the exposure notification interface that’s used around the world.

When using radio signals for measurement and converting them into other readings such as distances, various influencing factors must be taken into account in the calculation and estimation. These factors include movement, other radio signals and reflective materials in the surroundings, such as metal, walls, etc. Addressing the variance of these influencing factors and weighting them against an RKI risk assessment were among the most difficult tasks.

Because Bluetooth Low Energy can’t estimate the distance of a smartphone more accurately, we have to set priorities in terms of how we deal with this inaccuracy. The German federal government decided to accept a slightly increased number of false positives and opted to test more.

What developments are next on the map?

Alongside further parallel real-life tests, we need to address increasing demand from the corporate sector. Here, too, chains of infection ought to be broken as quickly as possible. In addition, efforts are underway to integrate methods for measuring distance using acoustic signals so as to achieve even greater accuracy for certain applications.

Thanks for the interview, Mr. Meyer.

The tech experts behind the Corona-Warn-App, pictured here with Fraunhofer president Prof. Reimund Neugebauer, who extended the German Chancellor’s thanks in person.