exposure time vs. distance

[traut21]

From the FAQs I learned that the risk of exposure is calculated from distance and time.

What are the details how this is computed?

Is there a chart of time vs. distance or could someone provide the algorithm behind?

I wonder what the minimum contact time is,
e.g. when you pass someone in the underground train.

[codejus]

Here you can find more about the risk calculation and assessment in the documentation:
https://github.com/corona-warn-app/cwa-documentation/blob/master/solution_architecture.md#risk-score-calculation
and https://github.com/corona-warn-app/cwa-documentation/blob/master/cwa-risk-assessment.md

[traut21]

Yes, I can find 'more' there - but I do not find the details. How is this dB scale applied?

I do see the 55 dB estimation (attenuation for 1.5 m) and a 63 db (for 3 m), but I do not see any dependency on time. Does this indicate that you need an exposure of at least 10 minutes at a close distance < 1.5 m?

[FMCumhaill]

@traut21 this might help you https://github.com/corona-warn-app/cwa-documentation/blob/master/transmission_risk.pdf

[traut21]

Thanks, that's plenty of stuff. Is this just the theory or is it actual how the app does work?

[traut21]

I've neither read nor fully understood the content of this description. But it reads as if it would leave the hard work to the ExposureNotification API. It might define some paramaters, but it does not name or explain the actual parameters in use.

Nothing about the essential factors of time, duration, distance or the aforementioned dB values

[daimpi]

@traut21 you're interested how the risk calculation works, right? We recently walked through some examples in the community Slack which you can join here.

The config values which are currently live can be found here 🙂.

[ndegendogo]

@traut21 and don't forget to take into account that both dimensions (distance and duration) are not a precise measurement but only a rough estimation.

Distance is estimated based on the signal strength (attenuation), but the signal depends also on how you carry your device (open in your hand / in your back pocket), and on reflexions at the environment.

Duration is estimated based on the number of beacons received. But reception is performed only every 5 minutes for a few seconds. So a single beacon can mean anything between 1 second and 10 minutes. And two readings (5 minutes apart) can mean anything between 5 and 15 minutes.

[MikeMcC399]

The question about the relationship between distance and dB levels was raised also in #484.

The time element is also discussed in #485 together with a new diagram on https://github.com/corona-warn-app/cwa-documentation/blob/master/images/risk_calculation/risk_calculation_enf_v2_overview.pdf.

[Ein-Tim]

Related: corona-warn-app/cwa-website#712

[heinezen]

The comment from #484 (comment) should also answer this question I hope.

The translation of signal attenuation to distance is not described by an equation, i.e. there is no formula that directly calculates the distance from signal attenuation. Given examples of attentuation-distance pairs (such as 73 dB --> >8m) should be seen as guidance values. The attenuation ranges used by the app's calculation are based on empiric research that - as far as I understand - examined which attenuation intervals would indicate that devices are close enough for the transmission of the virus.

• 73dB is very far away
• 63-73 is far away
• 55-63 is mid-range
• <55 is very close

The question about minimum exposure time is answered in the FAQ: https://www.coronawarn.app/de/faq/#encounter_19_calc

---

Corona-Warn-App Open Source Team

[Ein-Tim]

Is this issue solved?

[traut21]

not really. Some input is too much. The most relevant source https://github.com/corona-warn-app/cwa-documentation/blob/master/images/risk_calculation/risk_calculation_enf_v2_overview.pdf does not load properly and the downloaded version requires more detailed knowledge.

I do not understand the concept yet. If it's > 79 dB, the t_att is multiplied with 0,0 and then multiplied with a TRV. I do not understand the concept why you compute something by multiplying it with zero, making it zero that way. I would not expect an "x 0.0", but an "@ 0.0 s". Thus a contact that close would be added immediately.

https://github.com/corona-warn-app/cwa-documentation/blob/master/cwa-risk-assessment.md#current-configuration is more helpful, although I cannot find all matches between its tables and the other PDF. One is from 2021-01-19, the other from 2021-03-17. I guess the examples would need an update for the current status

[MikeMcC399]

@traut21
https://github.com/corona-warn-app/cwa-documentation/blob/master/cwa-risk-assessment.md is very out of date. See #453.

[MikeMcC399]

@traut21

The documentation has been updated.

See https://github.com/corona-warn-app/cwa-documentation/blob/main/solution_architecture.md#mobile-applications for current information.

The document https://github.com/corona-warn-app/cwa-documentation/blob/main/cwa-risk-assessment.md is now deprecated.

[Ein-Tim]

@traut21 See #446 (comment), does this answer your question? If not: @dsarkar please mirror this issue to JRIA.

[traut21]

I'd prefer a more simple presentation