[OSM-talk-be] BIPT antennas

Mon Mar 9 21:17:38 UTC 2020

Hello everyone,

Already thanks for all the feedback! I answer some of your questions in the following topics:

Continuous update (@Midgard and @rodeo.be )

It was in the back of my head but I didn't want to plan it yet. I will probably work on that but not soon.
Note that the list of BIPT antennas is updated monthly. Is there server for scheduled scripts within OSM BE ?

Workflow (@Midgard)

The distance triggering manual review has been changed to 25m.

> 3) And even then, just dumping elements in OSM without manual review is not considered best
   > practice, but since it's only nodes, things are relatively simple and I won't object. I would
   > just like to see that they're not placed too close to any other existing node, but that can be
   > checked automatically.

Like one meter? Is there JOSM tool for that?

BIPT (@Thibault Rommel)

I agree. I updated the proposal

Latitude and longitude as tag (@s8evq and @Midgard)

That's a mistake. I only wanted to explain that I use the longitude and latitude provided in the dataset. It has been updated.

Open data Portal (@rodeo.be)

Good idea. I will inform BIPT of the positive feedback of the OSM BE community and I will kindly push them to do so.

Precision of the localisation

Looking alone at 9000 nodes for manual review is quite some work. Would it be acceptable to have a FIXME tag stating that the localization could be a few meters offset? Or is it considered to pollute the OSM db to do so?

Tagging (@Lionel, @Karel Adams, @midgard)

If I had found the discussion from 2018 before, I would have maybe not prepared the import :-D.

What to map?

The BITP dataset include a list of sites used in our mobile phone network. Each site can include multiple supports for antennas. And each support can include multiple directional antennas.
There is no info on the support in the dataset. That's why I focused on mapping the antennas (or the group of antennas). Are they enough visible in our landscape? 

I think so. On rooftop in cities, those groups of antennas are often visible. Their white color is often even more visible than the support itself. 

On other structures like tower, @lionel already mentioned that the antennas could be tagged with telecom=antenna https://lists.openstreetmap.org/pipermail/tagging/2018-October/040276.html.

I can understand that there is no proper way to tag the antennas and the group of antennas without specifying the support but it seems a bit limiting. 

How to map it?

@Lionel, you didn't comment on the man_made=antenna tag (Status: in use) which was also mentioned in the discussion of 2018.
There is also this proposal from nov 2018 where the man_made=antenna + antenna:application=mobile_phone is suggested (https://wiki.openstreetmap.org/wiki/Proposed_features/antenna:use)

In any case, I don't have enough experiences on tagging and I will follow what comes out of this discussion.

Best regards,

Vucodil

March 9, 2020 11:39:20 AM CET Lionel Giard <lionel.giard at gmail.com> wrote:
Cool news that they gave the authorization to use it ! And it is always great to have some interest on the telecom side. :D I'll give what i know and some opinion on the tagging. ^_^

For the tags to use, there was a (rather long) discussion in October 2018 on Tagging mailing list (https://lists.openstreetmap.org/pipermail/tagging/2018-October/thread.html) and one output was that the current scheme is probably not good (but nothing was decided) ! :p 

I had done some manual cleaning on the mast/tower tags 2 years ago i think - i looked at mapillary footage especially for mast/tower along motorways where it is often easy to spot them or did some survey (and we are not many to map these structure so it was quite easy :p ). And the current tagging scheme should be (following the wiki and what was clarified in the discussion) :
EITHER : 
- man_made=mast / tower (really subjective, as we don't have real difference but mainly: a tower is generally freestanding and often larger diameter/width (think about (often) concrete telecom tower), while a mast generally have often some guy wires and/or have a small diameter/width (think about metallic mast).
- tower:type=communication
- tower:construction=freestanding / lattice / guyed_lattice / guyed_tube (https://wiki.openstreetmap.org/wiki/Key:tower:construction?uselang=en-US)
- communication:mobile_phone=yes (if GSM, which should be the case for all thse one).
- height=* (if known)
OR
- telecom=antenna (there is no real other tag for antenna alone, and this one is using the telecom=* key as some people want to clarify things).
- communication:mobile_phone=yes (if GSM, which should be the case for all thse one).  
- height=* (if known).

=> Those two are two different things : the first one is a structure that support some antennas (typical GSM mast support multiple antennas), and the second is a standalone antenna (on a rooftop for example). The BIPT only give antennas, so we must first determine if it is standalone or on a mast or tower.

There is no approved tagging of multiple antennas on 1 mast or tower (you mention the "Radio antennas mapping proposal" but it seems really complicated and easy to break with the relations...). Maybe we should just create a custom belgian tag (similar to how the french are tagging their own infrastructure) for the antenna present like : 
- ref:BE:BIPT=21292
- ref:BE:Proximus=10DLT_01    (or ref:BE:PXS if we want an abbreviation ?! I did use that in the past on street cabinet but i could change it)
- ref:BE:Orange=1-32264-W1
- ref:BE:Telenet=_BW4629P
Following what is in the technical data and their ID (i took one example having the three operators ;-) ). It would keep the different operator information like it is done on street_cabinet for exemple. It would also be easier to maintain and more difficult to break, because if we put 3 nodes next to each other (1 for each antenna), it would be easily broken by anyone editing the area (especially in ID editor). 

Note that, the operator tag is difficult to assess for the mast or tower structure as it could be any one of the multiple antenna operator or even someone else (and they don't give this information publicly). So i would not use the operator tag except on individual antenna or mast/tower that would only have 1 antenna. 

We could also use a subtag like antenna=1/2/3/... if we want to give the number of antenna on a same support (mast or tower) ? 

Note that there was some discussion of a "potential" proposal in the discussion to change the tagging of "telecom mast and tower" into something looking more like the "power" scheme. Something like that :
- telecom=tower (similar to power=tower grouping everything into one tag) 
- structure=guyed_mast, tubes_mast, lattice, tubular/tubes, ...
- tower:type=communication
- communication:mobile_phone=yes

=> This proposal is mainly re-using the common tags used for power scheme : structure=* instead of tower:construction (François Lacombe - a french mappers involved a lot in telecom scheme - was proposing that); and telecom=tower instead of man_made=tower or mast (i was proposing that). It would simplify the tagging as we would tag everything easily and refine only in the structure tag. But that was never formally proposed and approved AFAIK. 
I don't know if you want to go into the rabbit holes of trying to adapt a new tagging scheme for this ahah. Anyway we can use the current scheme as it would be easier now. ;-) 

Kind Regards,
Lionel

Le lun. 9 mars 2020 à 00:36, Midgard <midgard+talkbe at janmaes.com> a écrit :

Replying inline to s8evq and Karel:

Quoting s8evq (2020-03-08 20:20:34)
> What is the point of adding longitude=* and latitude=* to the nodes?

I had overlooked them, but these tags definitely have to be dropped.

> How precise are the locations of the antennas in the BIPT dataset? Do we know what the quality of this data is before importing?

The ten or so that I checked were pretty close, within 5 metres. One was either very recent, or
20 metres off. (BIPT has location 51.151194,3.235139 but there's no structure visible there on
the most recent imagery.)

In any case, we would get higher quality with a manual review instead of fully relying on the
source: we can correct errors when the structure is visible on imagery.

> Perhaps my questions sound a bit tough, but I appreciate the effort you put into this.

Such is an import discussion. Original Poster has my appreciation too :)

> On Sun, 8 Mar 2020 17:46:38 +0000, Karel Adams <fa348739 at skynet.be> wrote:
> > didn't we
> > have a rule to map only those features visible in the scenery? The BIPT
> > antennae (sic!) are usually attached to existing structures, such as
> > church spires or GSM masts or so? Of course we map those highly visible
> > carrying structures, but to map the individual antennae seems to me like
> > overdoing things.

Looking at the source data, it's going to be one node for one mast, which typically has several
directional antennas mounted on it. A node per antenna is not something I'd like to see either.

Off-topic: when referring to the electrical part, "antennas" is actually the most common form. By the way,
could you maybe start trying to behave more constructive and socially acceptable? I believe you can
do it with some effort.

Kind regards,
Midgard

_______________________________________________
Talk-be mailing list
Talk-be at openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

_______________________________________________
Talk-be mailing list
Talk-be at openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-be/attachments/20200309/510c5124/attachment.htm>