Hi Felix, On 27.08.2014, at 21:51 , Felix Baumann <[email protected]> wrote: > We need a column that tells us whether the cell's signal is radial or whether > it covers only a sector > For that sector we would need an angle and a direction (as an angel).
Currently neither of the two projects tries to determine the shape of the cell area. The logical cell networks we are dealing with can be shaped like a cone or like a circle. The position estimate for the cell is generally “where most people are” or “where most observations where taken from” - so almost never tell you where the actual antenna or tower for the cell is. The range estimate is currently based on taking a bounding box around all observations for each cell and calculating the distance between the cell position estimate and the farthest away corner of the bounding box. These two approaches work fairly well for the basic cell id approach, where you see only a single cell network at a time and try to give a position and accuracy estimate back. When used for trilateration, this approach introduces quite some additional error. But generally the goal today is to give you a better position estimate than the one you get from GeoIP and see if we can get enough quantity of data to make this sort of service sustainable. Quality of estimation is something we’ll probably look at more towards the end of the year. Hanno _______________________________________________ dev-geolocation mailing list [email protected] https://lists.mozilla.org/listinfo/dev-geolocation
