Mobile phone networks, like any other network, cannot be operated unless the location of each associated device is known. That cannot be avoided. Current mobile phone networks in the UK record location in terms of cell ID. Cells are defined by base stations, also known as "radio masts" or "antennae". Each base station has an ID and defines a cell. Assuming that an omni-directional antenna is used:
(In fact, the topology of cells varies. Cells can be divided into sectors. Cells can comprise multiple micro-cells. Omni-directional antennae do not necessarily define spherical cells ...)
When a mobile phone associates with a base station, i.e. when there is at least one blob of signal on the screen, its location can be narrowed down to the given cell and the mobile phone network operators can detect the location of the phone accurately to within several kilometres in the countryside, 150 metres in the city and sometimes better.
The main location-detection technologies being used in the US and Europe
are EOTD and AGPS:
While EOTD and AGPS approach greater accuracy in location-detection, other related technologies are overtaking them. 802.11 wireless networks can provide 5ft accuracy, see for example:
Ekahau's RFID products can achieve 50cm accuracy. They provide a useful starting point for comparing the various location-detection technologies. Start with those, and then try Geolocation technologies and applications for third generation wireless.
EOTD looks as though it has problems achieving accurate location-detection, AGPS seems to be well on the way and the possibility of using multi-protocol mobile phones, incorporating 802.11, looks most promising, given the growing prevalence of WiFi hotspots.
(The paragraphs above are adapted from a paper produced in May 2003 (para.4.6). The research has not been updated since then. Many references in the May 2003 paper have been removed, having disappeared from the web in the interim.)