General > General Technical Chat

Bluetooth Low Energy is unsuitable for COVID-19 contact tracing, say inventors

<< < (13/23) > >>

SiliconWizard:

--- Quote from: David Hess on May 14, 2020, 03:39:07 pm ---
--- Quote from: nali on May 14, 2020, 12:44:30 pm ---This is a pretty interesting blog post on the UK app:
https://www.ncsc.gov.uk/blog-post/security-behind-nhs-contact-tracing-app
--- End quote ---

The direct or of the National Cyber Security Centre (Ian Levy) is explaining how they are going to protect your privacy and security using a closed source application?  That is a laugh.

--- End quote ---

Indeed. ;D

cdev:
But you won't get indoor traces that are of any use, unless the person is in a flimsy 1 or 2 story wooden building and the antenna is pointing up and a whole bunch of other ifs.

let me point you to the subject area.. "SLAM" - Simultaneous location and mapping..
 Indoors is where people get infected, I suspect. 99% of the time.

Indoor mapping is a big technical field now because many people do want to build devices that can navigate indoor spaces and know where they are. Now I am by no means an expert but I do try to keep up on this and over the years Ive fooled around with it quite a bit, especially with GPS RTK.

There is a very very large difference between indoors and outdoors. As it stands right now all indoor mapping technology is basically crap. Outdoors is a different story but still very bad in "urban canyons" i.e. cities unless one goes to lengths to optimize performance in ways that are not feasible in a cell phone setting.

Also, constantly transmitting and receiving beacons is likely to drain battery life quite a bit. (GPS now can be done with very little battery) Presumably each device will contain a log of all the other BT IDs it receives and when. If somebody is in the range of another BT ID for an extended period of time, that might be valuable if the people were outdoors and it could be verified that they were proximate to one another. But then the risk of transmission is much reduced because they are outdoors. So pretty unlikely to be useful, except as a non-covid-19 surveillance tool.

 Indoors you'll notice that in cities, when people go indoors there usually is no GPS fix. Logged BT IDS will not be telling you where they are because there is no 3D fix. Unless they are right next to a window and then the lack of enough satellites will make the fix jump wildly around because  the GPS signal is boucing off buildings. It gets worse the further they are from windows.  newer GPS systems that are coming online use multiple bands and are substantially more accurate, outdoors, in urban canyons, but not inside buildings. And all GPS antennas like to be pointed up. Otherwise accuracy goes all to hell.

Read the literature. If you can find one paper that shows a technology that works with existing devices and which allows consistent indoor localization to the kinds of accuracies that can be reasonably used for contact tracing, I will be very surprised. In particular, the antennas in cell phones are not up to the task. Most of these experiments are done with robots (sometimes in this setting they are called "rovers") Look at the literature, the state of the art now has to use multiple inputs - particularly optical- video inputs and photogrammetry to get a reliable idea of location, based on the last good GPS fix, typically before they entered the building. They also use compasses and IMU. Thats called "dead reckoning"

Triangulation with cell phone sites is also nowhere near accurate enough indoors. What about wifi access ponts logged by Google, etc?  Again, not accurate enough to help much with this fine level of detail. What about visual tracking? That has a lot of romise, especially when using multiple cameras and especially D-RGB (RGB with depth sensors, similar to the MS Kinect devices)


--- Quote from: julianhigginson on May 14, 2020, 12:28:47 pm ---I think that going on and on about the RSSI strength detection thing and thinking it has to be perfect or it's useless is missing the point.

The contact tracing app is not meant to be 100% perfect.

if it can work even a bit, it can speed up transmission tracing.
which will catch some transmissions before the transmitted to person gets sick enough to be tested and get diagnosed (by which time they have passed it on themselves). so will bring down the reproduction number.

we see in the studies on clusters so far that exposure risk relates to the amount of viruses being expelled form the source vs time near the source. You can do worse in a room far away from someone who is just breathing if you are there for hours, than if you are right next to someone who is singing or shouting or exercising for a few minutes.

having the app log some kind of proxy for distance *and* time spent nearby would be useful, I suspect. Even if not perfectly accurate. seems that it only logs contact that has lasted around 15 minutes? seems a shame it can't measure times with a bit more accuracy. or at least multiples of 15 minutes.

--- End quote ---

You have to understand, once somebody drops off the GPS and goes underground they could literally be anywhere.  If they fill the metro stations etc. with BT beacons, thats still not going to get it to the level of accuracy it needs to be.

SiliconWizard:

--- Quote from: cdev on May 14, 2020, 03:09:12 am ---Security theater that is totally inaccurate is worse than nothing.

--- End quote ---

I do agree with that. If it's inaccurate enough that you can't actually deterrmine for sure that people have been in "contact" (thus in close proximity enough that it could matter as far as virus propagation is concerned), then it's basically useless, and worse than useless: if you still rely on it to take any kind of action or issue any kind of analysis, that's wasted time and money that could be better invested in something else.

cdev:
If it becomes wrong and or arbitrary, that is a way to guarantee it does not work for its intended purpose. because people will not cooperate if they feel that the system is not taking great pain to be accurate and fair.

 (Do we want a world like Terry Gilliam's "Brazil"?)


--- Quote from: SiliconWizard on May 14, 2020, 04:24:04 pm ---
--- Quote from: cdev on May 14, 2020, 03:09:12 am ---Security theater that is totally inaccurate is worse than nothing.

--- End quote ---

I do agree with that. If it's inaccurate enough that you can't actually deterrmine for sure that people have been in "contact" (thus in close proximity enough that it could matter as far as virus propagation is concerned), then it's basically useless, and worse than useless: if you still rely on it to take any kind of action or issue any kind of analysis, that's wasted time and money that could be better invested in something else.

--- End quote ---

Creating lots of jobs may be one of its main goals.

Look at Prohibition in the US during the Depression.

cdev:
I'm just saying, that the data recorded when somebody is indoors or in subways, (or tunnels) is likely to be useless, or close to it.

Trains, above ground? May be better but stil most trains now have metal cars with tinted windows. Tinted windows oftentimes eat up a lot of GPS signal, but trains without tinted windows, they may be able to get a fix from time to time.

Enclosed places, more so than outdoors, are where people get infected.


--- Quote from: Buriedcode on May 14, 2020, 04:50:03 am ---
--- Quote from: cdev on May 14, 2020, 03:09:12 am ---Security theater that is totally inaccurate is worse than nothing. What they need is contact tracing. Do you have a GPS tracker? Do this, turn it on, take a walk in a major city, going inside some buildings and going down into the subway. Then dump the trace and convert it into a KML trace. Since you cant upload KML here, maybe zip the file. Upload that. just do that. That will prove my point.

--- End quote ---

... And you.. expect the majority of citizens in an entire country to do this?

--- End quote ---

Arent they supposed to install an app that uploads all sorts of data that in essence, does that?  My point is that in any modern city, GPS traces recorded by a GPS tracker are going to be very very inaccurate. When they are indoors or underground, most of the time  there wont be any location data, with the occasional sudden apperance of a blit here a dot there, this results in wildly jagged traces all over the place (caused by multipath, which is the main problem with cheap GPS antennas, or antennas not pointed up. A qudrifilar helix is the best kind of antenna for this kind of situation. Its significantly better in rejecting multipath and funtioning when not oriented straight up..) .

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod