@CroweDM ANY assertion that TraceTogether does not allow for individualised location tracing is entirely misleading and ignores correlation risk from telco metadata. There ARE ways of achieving this but THIS WAY is not it. #privacy #zkp #overreach #fakenews
Thread continues: Phones pinging off each other gives location data for 100 metre range to 10 metre radius. That is to say I can detect someone else's phone up to 100m away and measure that distance within 10m accuracy.
On the mobile tower side, 4G gives a radius of roughly 20KM while 5G shrinks this radius to 300 metre radius. Given 5G coverage is currently limited to Telstra primarily we shall discount this. Consequently we have towers (which overlap) with 20KM radius.
The metadata beyond cell login (which includes IMSI sim->mobile & IMEI sim->phone) the only other piece of data required is continuous signal strength logging.
Enabling this data is very common for diagnostic purposes with the only real reason not to turn it on related to capacity planning (ie. more logs). Presumably given the powers for "national security" the government can give this explicit directive now.
So, we have the following sets of data:
1) IMSI tower login with IMEI phone identifier mapped to MSN (mobile number)
2) Continuous signal strength logging from IMSI -> Transmit strength Strength
3) Bluetooth Identifiers ("BLE") + Signal strength
4) Timestamps
1) IMSI tower login with IMEI phone identifier mapped to MSN (mobile number)
2) Continuous signal strength logging from IMSI -> Transmit strength Strength
3) Bluetooth Identifiers ("BLE") + Signal strength
4) Timestamps
We collect "20 samples" preferably from 5 participants (Government wants to get 40% penetration so they are aiming for double this)
With this data alone we make the following assumptions:
1) Towers are distributed in overlapping configuration
With this data alone we make the following assumptions:
1) Towers are distributed in overlapping configuration
2) Towers are configured for a 20KM radius
3) Signal strength logging on towers is accurate to 10mW (2 decimal places for watts output)
4) Bluetooth signal strength is accurate to 1mW (pretty reasonable, it's usually 10-100 times more accurate)
3) Signal strength logging on towers is accurate to 10mW (2 decimal places for watts output)
4) Bluetooth signal strength is accurate to 1mW (pretty reasonable, it's usually 10-100 times more accurate)
5) Phone behaviour is known (ie. we know each users phone type and we download their respective FCC datasheet - we can identify the phone type directly from the transmitted IMEI during login)
This gives us a cluster accuracy using 3 towers to a 2KM radius.
This gives us a cluster accuracy using 3 towers to a 2KM radius.
Within this 2KM radius the 5 participants have 4 data points over a time horizon. Even with this small sample set we are now able to triangulate the 2KM radius to <100m. I'm actually being conservative here, realistically with 5G + recent phones the accuracy is likely <1m 
@threadreaderapp unroll
