What is the basis for the dB Thresholds? Are they to low?

[OlympianRevolution]

Your Question

• Documentation File: https://github.com/corona-warn-app/cwa-documentation/blob/master/images/risk_calculation/risk_calculation_enf_v2_overview.pdf
• Line / Paragraph: https://github.com/corona-warn-app/cwa-documentation/blob/master/images/risk_calculation/risk_calculation_enf_v2_overview.pdf dB values
• Question:

1. What was the basis for the dB thresholds in the risk assessment?
2. How were these dB distance relationship values calibrated?
3. Should a person who is 4m away for 5 hours indoors be a zero risk contact?
4. Even if the distance is further should a very long duration contact be a zero risk contact? Indoors time means risk regardless of distance (within reason)

I have found that I could sit an entire day in a small office in about 4m distance from a colleague and the 79dB threshold would never be exceeded (in strength). Even if I were to occasionally be closer to that person it is unlikely that the ENF will scan at exactly that moment. This would lead to the improbable risk assessment that being 4m away from an infectious person indoors for an entire day has no risk of infection and I would not be warned should that person test positive.

Galaxy S21

[thomasaugsten]

I think you cannot compare the RSSI from the Bluetooth scanner because you are missing the transmission power of the sender to receive the real attenuation. You should try corona-warn-companion for rooted device or the Corona Contact Tracing Germany for detailed attenuation information.

The selected values are based on the Fraunhofer studies https://www.coronawarn.app/de/blog/2021-03-19-risk-calculation-improvement/

[vaubaehn]

Additionally to @thomasaugsten 's suggestions: Have a look here: https://developers.google.com/android/exposure-notifications/ble-attenuation-overview#calibration_per-device_tx_power_and_per-device_rssi_correction TX and RX power of devices are taken into account for risk assessment. For this, Google tested many commonly available phone models from different manufactuerers. When many phones of a specific model were available, they calculated averages of TX/RX power per model. When only some models were available per manufacturer, they calculated averages per manufacturer. There is a .csv that you can download to get a deeper impression. So, the values that the scanner displays, are not the values that are taken for risk assessment.

4\. Even if the distance is further should a very long duration contact be a zero risk contact? Indoors time means risk regardless of distance (within reason)

That was the main reason, why the feature "event check-in" was implemented into Corona-Warn-App. Unfortunately, neither politics, nor users or venue owners/hosts recognized the value of this feature well enough yet. I still hope, this will change soon.

[OlympianRevolution]

I think you cannot compare the RSSI from the Bluetooth scanner because you are missing the transmission power of the sender to receive the real attenuation. You should try corona-warn-companion for rooted device or the Corona Contact Tracing Germany for detailed attenuation information.

The selected values are based on the Fraunhofer studies https://www.coronawarn.app/de/blog/2021-03-19-risk-calculation-improvement/

Thanks for the info. I will do some more measurements with the CCTG app. (I hope they are using the same calibrations). Do you have a link to a publication of the actual Frauenhofer ENF attenuation research?

[Ein-Tim]

@OlympianRevolution Can this issue be closed?