Cluster data amount by region ids

[oberinspector]

Proposal:
Each app instance has an e.g. 2 byte region id which is provided or selected at time of installation. Large countries may have multiple regions and small countries maybe only 1 or 2. Goal ist to get almost equipartition regions related to population and expected infection rate. Also the regions must be large enough to respect the privacy of the users.

Each app sents own region id with the BT beacon and the receiving app seperates and collects all received region ids. So 2 Bytes of the BT payload would be needed to transfer own region id.

In case of infection and authorisation of a health organization the app uploads the EphIds of the acquisition period together with the own region id.

Apps retrieve only EphIds from infected individuals with the, during the acquisititon period, received region ids. For the majority of users this will only be one region id.

With this proposal it should be possible to have only one app for all participating countries and no national app solutions.

[oberinspector]

This proposal was inspired by a thread and a comment from oberguru at heise.de:
Coronavirus: Schweizer App für Contact Tracing verzögert sich

[lbarman]

Hi @oberinspector, thanks for your input!
We did consider country codes a while ago.
One problem is that we're (very) limited in the number of bytes we can transmit on the Bluetooth layer.
Additionally, despite the region, embedding a regional/country code could bring privacy issues (e.g., when roaming).
In general, interoperability (between regions/countries) is a tough topic on which we're working.
Thanks!

[oberinspector]

i understand that 2 of 16 bytes may be already to much but i see no privacy downside compared to a national solution. Depends on the size of the regions. Upload in case of infection is alwas done with the one home region id. Download only for the region ids received during the past 14 days.

[keugens]

We did consider country codes a while ago.

What is your preferred alternative?

One problem is that we're (very) limited in the number of bytes we can transmit on the Bluetooth layer.

Don't see this. The RPI (TCN, ...) could include a region identifier and there would be enough bytes left to avoid collisions for the largest possible region.

Additionally, despite the region, embedding a regional/country code could bring privacy issues (e.g., when roaming).

To my perception, the consequences not to implement a region identifier will very likely outweight any benefit in privacy by outmost degree. And the fact, that Apple/Google supports local public health authorities to do their own thing, should not mean the public has to accept this.

[oberinspector]

2 Byte / 64k would be sufficient to split the worlds surface in regions with unique Identifier. Large enough to respect privacy and small enough to cluster the data effective compared to many different national solutions.
It doesn't matter if the app detects the region or the user selects one. To use the real home region will help to reduce amount of data since the propability is high that one will only receive data from home region.