Difference between revisions of "FAQ"

From OpenCellID wiki
Jump to: navigation, search
(How to extract all cells of one country from the cell_towers.csv file?)
Line 5: Line 5:
 
{| width=998px
 
{| width=998px
 
|
 
|
 +
 +
===I know where cell tower x exactly is but OpenCellID shows another position===
 +
 +
There are mainly two reasons for discrepancies between reality and the position of a cell tower reported by OpenCellID:
 +
 +
<u>1) Cell towers and cells are two different things</u>
 +
 +
It is pretty rare that on a physical cell tower only one antenna is mounted which emits the GSM signal in 360 degrees direction. More often several antennas are mounted on a cell tower, pretty often 3 or 4 per network access type (GPRS, UMTS, LTE...).<br>
 +
In this case each antenna serves one segment of the full 360 degrees circle, e.g. 120 degrees as shown here:
 +
 +
[[file]]
 +
 +
This is where the big discrepancy between the number of cell towers and the number of cell-ids comes from:
 +
 +
Vodafone for ex. reports less than 40.000 cell towers ("Basisstationen") in Germany, but OpenCellID reports more than 290.000 Vodafone cell-ids in Germany as of August 2014: [http://de.wikipedia.org/wiki/Vodafone Vodafone in Wikipedia], [http://opencellid.org/#action=statistics.cells&type=1&dateFrom=&dateTo=&mcc=262&mnc=2&sortBy=1 OpenCellID statistics]
 +
 +
This means, that in average on cell tower carries more than seven antennas (= cell-ids) in average.
 +
 +
After understanding that cell towers and cells are not the same, let´s see what that means for the computed GPS positions of each cell-id:
 +
 +
Imagine that many cell-id measurements have been collected equally distributed in one of the cake slices.
 +
In this case, the average of all recorded GPS positions would be as indicated in the graphics above (e.g. "centre of area 1"). This would then be the position reported by OpenCellID.
 +
 +
In case OpenCellID would know which antennas belong to the same cell tower, this information could be used to average the positions of all cell-ids (antenna sectors) of one cell tower which would then give a precise position of the cell tower. Unfortunately OpenCellID today has very little knowledge about the numbering schemes of the different GSM network providers and network access types. In case you can contribute such information for one or the other network, this would be very helpful for improving the data quality of OpenCellID.
 +
 +
 +
<u>2) The measurements of cell towers are not always distributed equally around a cell tower</u>
 +
 +
There are countless situations where it is not easily possible to approach a cell tower from each side by car, by bicycle or as a pedestrian. Just imagine a cell tower on a hill where just one road passes by on one side:
 +
 +
[[file]]
 +
 +
In this case, the measurements will not be equally distributed around a cell tower, which means that may be most of the measurements are on just one side of the cell tower. Averaging the GPS positions of all these measurements will most likely not have the centre of the area of the segment of a cell tower as a result.
  
 
===I am an <u>[http://www.openstreetmap.org OpenStreetMap]</u> mapper. How can I also map cell towers?===
 
===I am an <u>[http://www.openstreetmap.org OpenStreetMap]</u> mapper. How can I also map cell towers?===

Revision as of 23:06, 29 August 2014

OpenCellID banner FAQ.png

Frequently asked questions

I know where cell tower x exactly is but OpenCellID shows another position

There are mainly two reasons for discrepancies between reality and the position of a cell tower reported by OpenCellID:

1) Cell towers and cells are two different things

It is pretty rare that on a physical cell tower only one antenna is mounted which emits the GSM signal in 360 degrees direction. More often several antennas are mounted on a cell tower, pretty often 3 or 4 per network access type (GPRS, UMTS, LTE...).
In this case each antenna serves one segment of the full 360 degrees circle, e.g. 120 degrees as shown here:

file

This is where the big discrepancy between the number of cell towers and the number of cell-ids comes from:

Vodafone for ex. reports less than 40.000 cell towers ("Basisstationen") in Germany, but OpenCellID reports more than 290.000 Vodafone cell-ids in Germany as of August 2014: Vodafone in Wikipedia, OpenCellID statistics

This means, that in average on cell tower carries more than seven antennas (= cell-ids) in average.

After understanding that cell towers and cells are not the same, let´s see what that means for the computed GPS positions of each cell-id:

Imagine that many cell-id measurements have been collected equally distributed in one of the cake slices. In this case, the average of all recorded GPS positions would be as indicated in the graphics above (e.g. "centre of area 1"). This would then be the position reported by OpenCellID.

In case OpenCellID would know which antennas belong to the same cell tower, this information could be used to average the positions of all cell-ids (antenna sectors) of one cell tower which would then give a precise position of the cell tower. Unfortunately OpenCellID today has very little knowledge about the numbering schemes of the different GSM network providers and network access types. In case you can contribute such information for one or the other network, this would be very helpful for improving the data quality of OpenCellID.


2) The measurements of cell towers are not always distributed equally around a cell tower

There are countless situations where it is not easily possible to approach a cell tower from each side by car, by bicycle or as a pedestrian. Just imagine a cell tower on a hill where just one road passes by on one side:

file

In this case, the measurements will not be equally distributed around a cell tower, which means that may be most of the measurements are on just one side of the cell tower. Averaging the GPS positions of all these measurements will most likely not have the centre of the area of the segment of a cell tower as a result.

I am an OpenStreetMap mapper. How can I also map cell towers?

OpenCellID mainly collects MCC, MNC, LAC, and CID information combined with a GPS position.
In most cases, it is not possible to obtain this data by going to a cell tower for various reasons:

  • the cell tower might be on a rooftop which is not accessible
  • there is no sign at the base station indicating such information
  • there are many antennas all together and it is not possible to find out which antenna has which Cell ID

In addition, one of the basic rules of OSM is to map visible things. This is not the case here as MCC/MNC/LAC/CID are not visible in most cases.
Therefore, only a very small number of cell towers is currently mapped in OSM with MCC/MNC/LAC/CID tags.

The most effective way to contribute to OpenCellID is to use one of the smartphone applications listed here and to collect measurements while mapping something else.
The Keypad-Mapper 3, for example, has a built-in feature to record cell tower data while mapping house numbers / addresses.

How to extract all cells of one country from the cell_towers.csv file?

Due to the size of the cell tower file it is not possible to use common programs like Excel or Microsoft Word for this task.

One way to extract the required data is as follows:

  1. Download cell_towers.csv file here.
  2. Install the free Windows software CSVed.exe on your windows PC.
    Last version we tested successfully was version 2.3.2 (2014). It creates a shortcut on your desktop during its installation.
  3. Run CSVed.exe and load the CSV file
  4. Use “Filter & Dups” feature for filtering and saving the required data; filtering of the data takes a few minutes but works perfectly good.
  5. Open the extracted data with Excel or any other CSV compatible software for further examinations.

Thank you very much to Sam Francke for this great tool!

What is a CLF format?

CLF files contain information about mobile network cells.
They were used in some Nokia mobile phones for logging cell data: Cell Id, LAC, MCC, MNC, location etc.
OpenCellID supports import of CLF files as a source of information about cells locations. During the upload of the data it can be defined, if the uploaded data is measurements or if the file contains precise information of cell tower positions which would then override the average of all GPS-positions of all measurements of a given cell.

The following versions are supported: 2.0, 2.1, and 3.0.

CLF version 2.0

Files have the following format:
CellId(hex.)+LAC(hex.)+MCC+MNC<TAB>data
(plus symbol means that values are not separated; <TAB> — tabulation character)
Example: 239731EF26202<TAB>City Square

CLF version 2.1

Files have the same format as in version 2.0 but CellId and LAC are stored as decimal values.

CLF version 3.0

This format uses more information about cells. It has the following format:
MCC+MNC;CellId;LAC;RNC;Latitude;Longitude;Ratio;Data;RFU
CellId and LAC can be decimal or hexadecimal.
Example:
26202;07812;03101;00000;45.894375;31.51312;0;City Square;0

LAC vs. RNC

One RNC can handle multiple LACs. RNC ID does not necessarily have a connection to LAC ID - it depends how the network is planned. For example you can find in the same RNC (12) LACs 33, 21 and 78 - no connection between them. No relative connection of RNC IDs and LAC. This is based on how to plan the network. LAC depends on paging. RNC ID is same as Site IDs. It only serves as identification.