OpenCellID (was OpenBmap logger (GSM positioning)])

Thomas Landspurg t.landspurg at 8motions.com
Wed Aug 26 08:38:05 CEST 2009


  Sorry for not replying this email. I am not an active member of the
open moko community so I was a little bit overloaded by the mail rate
at that time. Hopefully, the OpenVMap guy pointed me this unreplied
email, so here are my answer, and it's basically very short:


2009/2/21 Nick <realtimeblog at gmail.com>:
> Thomas,
>
> I agree with you about the train and people in the train.
> what about bad gps hdop,... ?
>
> I agree your approach  will be probably enought for assited GPS.
> (anyone knows the precision needed ?)
>
> but I don't think this is the right one for the other services mentionned.
>
> Thinking of a high quality "my position" service (google kind),
> we will not achieve it not treating gps hdop, gps speed...
>
> so to be constructive in order to merge our databases,
> would you be ready to collect gps speed, gps (hvp)dop and make these
> info available
> in your (very large) measures file ?

  YES


>
>  From this, I could provide you with a new mapping manager that identify
> the cells position calculated with gps (hvp)dop, gps speed when these
> values are available
> the cells position calculated with no gps (hvp)dop, gps speed because
> these values are not available
>
> We would have only one database with both quantity (the whole database)
> and quality for some cells. (Hoping that the best quality,
> would be available for all the cells in the future)
>
> what do you think about it ?
>

  Sounds great!

 The diifculty will be to do the new computation of all cells with
data of various level of quality. Hard but not unpossible

> regards,
> Nick
>
> Thomas Landspurg a écrit :
>>
>>
>> 2009/2/20 Nick <realtimeblog at gmail.com <mailto:realtimeblog at gmail.com>>
>>
>>     Thomas,
>>
>>     After trying to reach you a few times last year,
>>     i am really glad to have some news from you now.
>>
>>
>>   The easiest way is to use the email mentionned in the Web page! ;-)
>>
>>
>>
>>     I am reponsible for the openBmap website. And yes, it would be a great
>>     thing to merge our projects !
>>
>>     The number of logs you have is very impressive ! well done !
>>
>>     My concern would be about the quality of your data .
>>
>>     you still mention on the front page
>>
>>     "Note:If you want a professional CellID Database, I suggest you to
>>     go to
>>     Navizon
>>     who provides top services and databases."
>>
>>     what do you mean ?
>>
>>
>>
>>  Tihs mean that there are company that have created huge databases of
>> high quality by spendig a lot of money on it, and they make money by
>> selling these data. Navizon pays his users for this, and other are
>> throwing a lot of money on this too, by sending people doing measures,
>> or by buying operators database. That's exactly the same difference
>> between OpenStreetMap and Navteq/Teleatlas. OpenStreetMap is free,
>> provided by the community, but of a lower quality than their
>> commercial counterpart except on some specific area not covered by
>> these equivalents. Of course, the objective is to reach the same
>> quality, but this will take time.
>>
>>
>>
>>
>>     and after importing 800 000 of your gps points in january, I see that
>>     for instance
>>     gps speed,  gps hdop, gps pdop, gps vdop are not available
>>      (at least at the begining of this huge measures.txt file !),
>>
>>
>>     A bad pdop, vdop, hdop and i am positioned at 1 km from my real
>>     position ...
>>     What if i am in a high speed train at 300 km/h? or in plane ? (yes, it
>>     can work in planes...)
>>
>>
>>
>>   Let's go back to the basics: the objective of these database if to
>> provied an positionning to create services using localisation on top
>> of this database. Let's take the sample of an high speed train: there
>> is high chances that all the sample will came from people in the train
>> itself, and not from the neighbour outside (train don't go at 300 km/h
>> in high density area). So this mean that the REAL position of the cell
>> will never be accuratly computed, which is not an issue, because the
>> only interesting information is the user position. So sampling mesures
>> even some errors is fine as long as it works fine to get user position.
>>   So the philosophy behing opencellid was to reach the 80/20 ratio:
>> acheiving the 80% of functionality will require only 20% of the time
>> needed to do these 100% functionality. So that's why we have a simple
>> and elegant API, that is used by different devices with different
>> capacities, while acheiving exactly this objective: providing an
>> accurate cell id positionning.
>>
>>
>>
>>     In my opinion, considering the following services
>>
>>     *** asisted gps
>>     *** "cell id" to google "my position" kind of service
>>     *** "cell id" to "town name" service
>>
>>
>>
>>      the real questions are:
>>     what precision do we need for openmoko location service through
>>     gsm cell
>>     id ?
>>     what precision our possibly merged database would provide ?
>>
>>
>> Assisted GPS does not require huge precision. I am not an expert, but
>> I would be curious to know what is the precision needed to get an
>> assisted GPS. Using triangulation is a different story, and this
>> obvisouly will work fine only in high density area, where you also
>> have high density cells. So again, there is a direct correlation
>> between the precision and the density of the area.
>>   Note also that OpenCellID use LAC (Local Area Code) to provide an
>> alternate positioning with a lower precision if a cell is not know,
>> but if the LAC is know. Typically other cells has been discovered in
>> the same area.
>>    So once you get CellID positionning, you can use reverse geocoding
>> service to get the town for instance.
>>
>>
>>
>>
>>     what do you think about the above considerations ?
>>
>>     really glad to hear from you !
>>
>>     regards,
>>     Nick
>>
>>     Thomas Landspurg a écrit :
>>     >
>>     >
>>     > 2009/2/20 Onen <onen.om <http://onen.om>
>>     <http://onen.om>@free.fr <http://free.fr> <http://free.fr>>
>>     >
>>     >     Hi Thomas,
>>     >
>>     >     Thomas Landspurg wrote:
>>     >     >
>>     >     >   Dear OpenMoko community (and thanks ed for pointing this
>>     out).
>>     >     >
>>     >     >   I am behind the opencellid.org <http://opencellid.org>
>>     <http://opencellid.org>
>>     >     <http://opencellid.org> project, and it
>>     >     > seems that there are some discussion around it these day
>>     on the
>>     >     mailing
>>     >     > list.
>>     >     >
>>     >
>>     >     Last month, and today, indeed.
>>     >
>>     >
>>     > Yes, I've get to it today! It's a pity not to have been notified of
>>     > such discussion before. I've been through them, and I want to
>>     add some
>>     > clarification about difference between database:
>>     >
>>     > - OpenCellID can also store signal strengh. But one of the issue, is
>>     > heterogenity between datas. Some client don't have this information,
>>     > and this create some additional complexity. That's why this
>>     > information is for now only stored but not yet used. And in all was,
>>     > if you want it, it's in the measure table, and not in the cell
>>     table.
>>     >
>>     > - About the barycenter of the area/instead of barycenter: it's not
>>     > always the best way to do it, as it give more value to false datas.
>>     > while the "simple" barycenter reduce these. Another option would
>>     be to
>>     > exclude data that would be completely "out of range"
>>     >
>>     >  And more "general" information about the databse: more than 500
>>     > developers have registered to get an API key. Obvisouly not 500
>>     > application are out, but show the interest of the community
>>     >
>>     >
>>     >
>>     >
>>     >     >   So let me clarifiy:
>>     >     >
>>     >     > - As described in the web site, the license is under
>>     creative common
>>     >     > share alike 3.0. I had several request today stating that just
>>     >     linking
>>     >     > to the license was not clear enough, so I will re-clarify
>>     it on
>>     >     the web
>>     >     > site, but also in this list.
>>     >     >
>>     >
>>     >     Last time I checked (and other people, see the post I
>>     pointed out
>>     >     earlier today), it was not clear. But today, as people had a
>>     look
>>     >     again,
>>     >     it seems to me pretty clear :-)
>>     >
>>     >
>>     > Last time, it was written under creative common license, with a link
>>     > to the creative common share alike license. I am sorry if this
>>     was not
>>     > clear enough, but as you see, a simple mail is enough to get it
>>     corrected.
>>     >
>>     >
>>     >
>>     >
>>     >     > - I am surprised to see statment that I did not answer to some
>>     >     > questions. I've verified,and all openmoko request have been
>>     >     answered. I
>>     >     > am not perfect, and may be some emails have been missed, but a
>>     >     search on
>>     >     > openmoko on my mail box did not raise any pending question.
>>     >     >
>>     >
>>     >     Not sure what you mean by openmoko request... When I stated that
>>     >     neither
>>     >     us nor openmoko did get an answer, I should have written:
>>     "If I recall
>>     >     correclty openmoko tried to reach opencellid, but I have not
>>     heard of
>>     >     any response. We tried to contact opencellid, but got no
>>     response." I
>>     >     personally have not tried to reach you. But Nick yes,
>>     without answer,
>>     >     for what I have understood.
>>     >
>>     >
>>     >    Would be curious to have his email just to check. I am quite sure
>>     > I've answered to all demands like this.
>>     >
>>     >
>>     >     >     There is more than 100 000 cells covered, with 5.5
>>     millions of
>>     >     > measure, and more cells will be "donated" soon. We expect to
>>     >     reach 200
>>     >     > 000 cells in the coming weeks thanks to a new project
>>     donation.
>>     >     There is
>>     >     > also more than 10 different clients (windows mobile, symbian,
>>     >     > blackberry, j2me,...) gathering the database.
>>     >     >
>>     >
>>     >     Is there different countries?
>>     >
>>     >
>>     > Yes, the stats page show all countries:
>>     >
>>     > http://www.opencellid.org/cell/stats
>>     >
>>     >
>>     >
>>     >     Good to see there is no client for openmoko, otherwise I may
>>     have
>>     >     worked
>>     >     for nothing ;-)
>>     >
>>     >
>>     >  Yes, I've heard that other where working on such client too!
>>     >
>>     >
>>     >
>>     >     >   I've been running this project since more than one year
>>     , with an
>>     >     > objective to push community efforts around cell id. So I would
>>     >     be more
>>     >     > than happy to see new effort joining this project instead of
>>     >     creating
>>     >     > separate efforts.  So let's join effort and create
>>     something big!
>>     >     >
>>     >
>>     >     I am very glad to read this, especially as I was very
>>     disappointed not
>>     >     being able to leverage the existing work you have done. For
>>     my part I
>>     >     work on the client side. A logger/uploader. I guess it would be
>>     >     easy to
>>     >     modify it to upload to your database if we go that way. But
>>     for now, I
>>     >     think it would be good Nick (who takes care of the server side)
>>     >     and you
>>     >     keep discussing, in order to evaluate a possible merger.
>>     >
>>     >     As there are also plans on embedding the database on the
>>     phone, and
>>     >     using it to locate, I would like to know if this part would
>>     interest
>>     >     you? Or only the server side and upload?
>>     >
>>     >
>>     > The idea is to provide all the means to do so. So if there is
>>     anything
>>     > that is needed to help you to do this, I would be happy to
>>     provide it.
>>     > For instance, a way to send an area and get the list of cells in
>>     that
>>     > area. I amalready working on such  functionality.
>>     >   But the switch to from OpenBMap to OpenCellID should be quite fast
>>     > as I assume that the API is probably the same, or very close to. (
>>     > http://www.opencellid.org/api )
>>     >
>>     >
>>     >
>>     >     >  And as a reminder, the complete data base is available for
>>     >     download.
>>     >     >
>>     >     >  So if you have any question/interrogation, feel free to share
>>     >     them with
>>     >     > me so we can clarify this.
>>     >     >
>>     >     >  Regards,
>>     >
>>     >     Great we can move along. I hated this feeling of reinventing the
>>     >     wheel!
>>     >
>>     >
>>     > So do I!
>>     >
>>     >
>>     >
>>     >     Onen
>>     >
>>     >
>>     >     _______________________________________________
>>     >     Openmoko community mailing list
>>     >     community at lists.openmoko.org
>>     <mailto:community at lists.openmoko.org>
>>     <mailto:community at lists.openmoko.org
>>     <mailto:community at lists.openmoko.org>>
>>     >     http://lists.openmoko.org/mailman/listinfo/community
>>     >
>>     >
>>     >
>>     >
>>     > --
>>     > Thomas LANDSPURG
>>     > 8Motions
>>     > Founder/CTO
>>     > http://www.8motions.com
>>     > http://www.opencellid.org
>>     >
>>     >
>>     >
>>     ------------------------------------------------------------------------
>>     >
>>     > _______________________________________________
>>     > Openmoko community mailing list
>>     > community at lists.openmoko.org <mailto:community at lists.openmoko.org>
>>     > http://lists.openmoko.org/mailman/listinfo/community
>>     >
>>
>>     _______________________________________________
>>     Openmoko community mailing list
>>     community at lists.openmoko.org <mailto:community at lists.openmoko.org>
>>     http://lists.openmoko.org/mailman/listinfo/community
>>
>>
>>
>>
>> --
>> Thomas LANDSPURG
>> 8Motions
>> Founder/CTO
>> http://www.8motions.com
>> http://www.opencellid.org
>>
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Openmoko community mailing list
>> community at lists.openmoko.org
>> http://lists.openmoko.org/mailman/listinfo/community
>>
>
> _______________________________________________
> Openmoko community mailing list
> community at lists.openmoko.org
> http://lists.openmoko.org/mailman/listinfo/community
>



-- 
Thomas LANDSPURG
8Motions
Founder/CTO
http://www.8motions.com
http://www.opencellid.org



More information about the community mailing list