The standard is open, non-proprietary and still evolving. It has obtained wide support especially in Europe, where it is the major standard.
GSM differs from its predecessors most significantly in that both signalling and speech channels are digital. It has also been designed for a moderate level of security.
GSM exists in four main versions, based on the band used: GSM-900, GSM-1800, GSM-850 and GSM-1900. GSM-900 (900 MHz) and GSM-1800 (1.8 GHz) are used in most of the world, excluding the United States and Canada. The United States and Canada uses GSM-850 and GSM-1900 (1.9 GHz) instead, since in the U.S. the 900 and 1800 bands were already allocated.
In some countries the GSM-900 band has been extended to cover a larger frequency range
In Europe and other areas outside North America the GSM system initially used a frequency of 900MHz, shortly afterwards the PCN network used the 1800MHz frequency, nowadays the PCN networks are considered part of the GSM system and many phones are dual-band operating on 900/1800MHz.
The following data capability text needs to get relevant hyper links and references inserted
In GSM, a call is dedicated either as voice or data. A voice call uses a GSM specific codec to transmit the audio over a 9600 BPS digital link to the base station.
A data call lets the user use the phone as a modem with 9600 BPS bandwidth (some networks may also handle 14400 BPS). All newer GSM phones can be controlled by a standardised hayes AT command set through a serial cable or a wireless link ( irDA or bluetooth) . The AT commands can control everything in the phone from ring tones to data compression algorithms. An extension to the GSM data capabilities, High Speed Circuit Switched Data (HSCD), allows data transmission speeds up to 43.3 K[[bit]/s by allocating several data channels into one logical link. Realistic bandwidth is usually about 30 kbit when standing still. Expect 10 kbit/s if moving.
GPRS is backward compatible with GSM. This eases the migration path for a GSM operator, who can gradually upgrade the infrastructure to GPRS as the market expands.
Packet switched data under GPRS is achieved by allocating unused cell bandwidth to transmit data. As dedicated voice (or data) channels are setup by phones, the bandwith available for packet switched data shrinks. A consequence of this is that packet switched data has a poor bitrate in busy cells. The theoretical limit for packet switched data is approx. 170 kbit/s. A realistic bitrate is 30-70 kbit/s. A change to the radio part of GPRS called EDGE will allow higher bit rates of between 20 and 200 kbit/s.
GPRS packet switched data is IP-based. Each phone has one (or more?) IP addresses allocated. GPRS will store and forward the IP packets to the phone during cell handover (when you move from one cell to another). TCP's inability to differ between radio noise induced pauses and network congestion makes the protocol unsuitable for GPRS (or any radio based IP traffic). A radio noise induced pause will make TCP (unnecessarily) throttle back its transmission speed.