search

camaraproject / DeviceLocation

Repository to describe, develop, document and test the DeviceLocation API family
Apache License 2.0
21 stars 32 forks source link

Units for radius #60

Closed jlurien closed 1 year ago

jlurien commented 1 year ago

Currently is kilometers [km] but there is a proposal to use meters [m] instead

Opinions?

bigludo7 commented 1 year ago

I have no enough technical background to provide a precise answer but my understanding is cell tower triangulation precision at best is around 300/500 meters with sometimes a 150 m precision in city where there is a high density of tower. But it could be several kilometers in rural area.

My 2 cents: I tend to think that km radius is just good enough and not provided unrealistic promise.

alpaycetin74 commented 1 year ago

There are a variety of location technologies available , some of which may provide very accurate results within a radius of a few meters. I believe we favor the more accurate systems and go with meters.

JoachimDahlgren commented 1 year ago

It would be beneficial if we can use the same unit in both verify location and "retrieve device location" and I would then prefer to use meters as accuracy will differ depending on the situation and which underlying system might be used.

jlurien commented 1 year ago

Related to this is that currently the location verification API specifies a minimum for accuracy of 2 km. If we move to meter we should probably review this limit as well.

bigludo7 commented 1 year ago

Related to this is that currently the location verification API specifies a minimum for accuracy of 2 km. If we move to meter we should probably review this limit as well.

This is exactly that make me a bit nervous about this switch.... for me we should align on the 'minimal' capability and not the 'best' one (based on gmlc for example) in order to get as many as possible telco implementations. And so in this case we have to consider use of cell-based location. Doest it make sense using meter if our accuracy is around 500 meters (and we did not have this 500 meters accuracy in remote place) ?

jlurien commented 1 year ago

I'm waiting for some internal feedback, but in case that now or in a near future, accuracy for this type of service may be in the order of 100s of meters in some cases, it would may sense to define accuracy as, for example:

description: accuracy in meters [m] type: integer minimum: 500

even if we have to return 5000 in other cases

bigludo7 commented 1 year ago

Discussion occurred in Jun 20th call --> use meter