[New software] openBmap logger (GSM positioning)

Onen onen.om at free.fr
Fri Feb 20 19:17:04 CET 2009


Yorick Moko wrote:
> isn't more data always preferable? you could still decide not to use them...
> cells that we are connected to would recieve a higher "precision" off course,

Agreed. That is the point of view of Nick (from the project). Add a 
field to know if the log is a connected cell, or a neighbour one. On the 
phone the idea is to have the areas of the cells in a database. Not 
every GPS point uploaded to the "raw" database. But still, the "raw" 
database could grow up pretty fast. You should see with Nick, he takes 
care of the server side.

> but if I walk five times around in a city, and I never get connected
> to a certain cell, but I know a lot of points where that cell is in
> range, and a lot of points where it isn't? I don't find it impossible
> that one could distill some information from this.

Probably. But again the database could grow quickly. But as stated in my 
previous email, I think we need to experiment with this, to see what 
works or not, and what takes not too much space.

> And even a precision of 30km or more helps the assisted gps from
> ublox; so neighbouring cells can certainly be of some help, don't you
> agree?

I definitely think that using the neighbour cells to improve accuracy at 
locating time, will be much useful. What I wonder, is if to log them, 
increasing the amount of data, but stretching the areas related to them 
will bring us anything.

> y
> On Fri, Feb 20, 2009 at 2:42 PM,  <onen.om at free.fr> wrote:
>> For the moment our position is:
>> We log the cell ids we get connected to.
>> We do not log the neighbour cells. The reason is: if we are far from the cells,
>> we will (maybe, this needs experimentation to my point of view) see the cells
>> even very far. I fear that it would stretch the areas, and thus decrease the
>> precision, nullifying the benefit of recording those additional cells.
>> But, my idea, is to use the neighbour cells, to detect the overlapping region
>> among areas of cells I currently see, to improve the precision.
>> Any comments?
>> Onen

More information about the community mailing list