General Packed Radio Service (GPRS)
GPRS Air Interface
GPRS Coding Scheme
GPRS States
GPRS Network
Enhanced Data rates for GSM Evolution (EDGE)
Usage of GPRS
Packet Switched Services for mobile communication
Two digital technology developments evolved in the 1990, the digital mobile communications and the Internet. The Internet became more and more popular, not only for research and industrial applications but also for „normal consumer“. Access took place via telephone lines. The trend to be „mobile“ increased the requirement for mobile phones to provide efficient access to the internet as well. In the beginning this was not to get Internet services with the phone but to use the phone as a modem for a mobile laptop PC.
The PCs had become mobile in the nineties. They were battery operated and could be taken anywhere. In the business world, PCs were already networked, via internal company networks and also via the Internet. Access to the Internet was achieved with modems (mostly already build into the PCs) and telephone lines. In the mid-1990s, Canadian engineer Brent Townsend invented the 56k modem. This achieved the highest possible transmission rate in voice based networks. These were limited by the speech bandwidth and, as we have seen, were transmitted at 64 kbit/s.
It would now be natural to also dial into the network via a mobile phone. This was actually possible. This service was called „circuit switched data“. However, the data rate that was achieved was only 9.6 kbit/s. Since a subscriber had to pay for the time of a connection was, this service was very expensive. It was also very inefficient. The subscriber paid for the connection, not the transmitted data. But if you are connected to the internet there is no continuous transfer of data and most of the time, the expensive line is unused.
General Packet Radio Service: GPRS
At the end of the 1990s, communication-engineers thought about how the GSM network could be used for packet-oriented data transmission. The target was not to build a completely new network, but rather to use the existing infrastructure as much as possible. It was useful, that the time slots of GSM, the bursts, somehow already resembled “packets”.
GPRS Air Interface
One straight forward way to introduced packet service would have been to use a single time slot to transfer a packet. The bitrate would still be less than 10 kbit/s with would be almost a factor of 6 less than the experience in the PC space. Therefore it was proposed to bundle multiple time slots. 4-5 such time slots or bursts could theoretically be used. On the other side this would be a challenge for the existing cell phones. More time slots would lead to even more signal processing, even if a speech codec would not be needed for pure data transfer. For this reason different classes of mobile phones were introduced with different performance capabilities.
Class | Time slots Downlink | Time Slots Uplink | Sum |
1 | 1 | 1 | 2 |
2 | 2 | 1 | 3 |
3 | 2 | 2 | 3 |
4 | 3 | 1 | 4 |
5 | 2 | 2 | 4 |
6 | 3 | 2 | 4 |
7 | 3 | 3 | 4 |
8 | 4 | 1 | 5 |
9 | 3 | 2 | 5 |
10 | 4 | 2 | 5 |
11 | 4 | 3 | 5 |
12 | 4 | 4 | 5 |
30 | 5 | 1 | 6 |
31 | 5 | 2 | 6 |
32 | 5 | 3 | 6 |
Due to the class of the receiver, the base station was able to allocate different numbers of time slots for transmission. When defining packet services, it was assumed from the start that more data was transmitted in the downlink than in the uplink. Normally, Internet applications load data and usually only send control commands. Data was uploaded less frequently. It was also recognized that a single time slot was too small to serve as a data packet. Therefore, four time slots were always connected for a transmission. This new transmission method was called General Packet Radio Service or GPRS.
GPRS Coding Scheme
If a normal convolution channel codec of 1/2 (2 bits per data bit) is used to protect data transmission only 8 kbit/s per time slot can be achieved, or 32 kbit/s for 4 combined time slots. This was already pretty decent, but it can’t really come close to the data rates of 56 kbit/s that were common at the time.
Various coding methods were therefore introduced for GPRS. For CS1 (Coding Scheme 1) the classic channel encoder with 1/2 was used. With CS2, individual bits were deliberately omitted from the transmission (punctuation). This allows more data to be transmitted in the hope that the remaining redundancy is sufficient to close the punctuation gaps again. If you have a very good transmission channel, you can even transmit the data unprotected. If errors still occurs, packet-oriented data transmission allows repetitions. Such repetitions can then take place with better data protection. There were a total of 4 coding schemes for GPRS.
Coding Scheme | Bitrate (kbit/s) per time slot | Code rate |
CS-1 | 8 | 1/2 |
CS-2 | 12 | 2/3 (punctuated) |
CS-3 | 14,4 | 3/4 (punctuated) |
CS-4 | 20 | 1 (unprotected) |
This way, a respectable 80 kbit/s could be achieved with GPRS. The transmission channels for GPRS are called PDTCH (Packet Data Traffic Channel) in GSM. Additional special control channels are Packet Associated Control Channel (PACCH) and the Packet Timing Advance Control Channel (PTCCH). All other necessary control channels are shared with GSM.
GPRS States
During normal GSM operation there are two states. After switching on, the phone is in idle mode. To get there, it logs on to the GSM network, is authenticated and the cell in which it is located is known. When a call is initiated, the phone transferres into the dedicated mode. In this mode, it is connected to the network via a dedicated “virtual line” (traffic channel).
With GPRS, the phone is initially in an idle state. This should not be confused with the idle state of GSM operation. In idle status the phone is not connected to the GPRS network. Through a GPRS attach, the end device logs on to the network and enters the Ready State. The Ready State is similar to a “normal connected modem” and is “always on”. In GSM operation this would corresponds to dedicated mode.
Packet channels can be assigned and used at any time in the downlink as well as in the uplink. However, it would be uneconomical to maintain this situation over a long period of time as it ties up too many resources. Therefore, a timer expires, typically after 44 s. If no activity occurs during this time, the ready mode is left and a third state is entered, which is called „standby“. This is similar to GSM Idle Mode. The connection to the network continues to be checked and cell changes are controlled if necessary. Data can only be transferred to the end device via prior paging. If a new transmission occurs in this way, the device transfers back to the Ready state.

GPRS Network
GPRS caused major investments and changes, especially on the network side. Many elements could be shared with the GSM network, especially the base stations. But beyond the base stations, packet services went into a different network than the GSM switch network.

The base stations consist of the Base Transceiver Stations (BTS) and the Base Station Controllers (BSC). Normally, the BSC is responsible for allocating time slots and transmits user data to the so-called TRAU (Transcoding and Rate Adaptation Unit), which converts voice-encoded data into 64 kbit/s voice data and vice versa. From there, the voice signals go to the Mobile Switching Center (MSC), which is connected to the Home Location Register (HLR) and possibly to the Visiting Location Register (VLR). These provide information about the identity and location (cell) of the mobile phone subscribers. The MSC also connects to the classic telephone network (PSTN).
In packet switched services, the so-called Packet Control Unit (PCU) ensures the allocation of time slots in the base stations and thereby takes over essential functions of the BSC. The PCU forwards the packets to the Serving GPRS Service Node (SGSN). This is essentially a counterpart to the MSC. It is also appropriately connected to the Home Location Register to manage the subscriber’s location and control encryption tasks and authentication tasks. The gateway GPRS Service Node (GGSN) provides access to the actual Internet. The GGSN is used to trick the Internet into believing that the subscriber has a (fixed) IP address. Even if the subscriber moves in the mobile network, his access point in the GGSN does not change.
Enhanced Data rates for GSM Evolution (EDGE)
There was a relatively simple way to increase GPRS data rates even further. Pure GPRS still used GMSK to modulate the data bits. So just one bit per transmission is used. As we have already seen, IS-54 (digital AMPS) already used a modulation scheme that allows 2 bits per transmission (Pi/4 DQPSK).
Now the GPRS networks went one step further and introduced 8-PSK modulation which allows 3 bits per transmission.

With 8 PSK coding, various coding schemes were introduced which protected the data strongly or not at all. Without error protection, a bit rate of 59.20 kbit/s per time slot was achieved. By bundling 4 time slots, it was possible to get a respectable transmission rate of 236.8 kbit/s.
EDGE was first introduced in the USA in 2003, probably primarily to face competition from Qualcomm. In Europe, especially in Germany, the introduction was rather late and almost parallel to UMTS, the actual 3G standard. T-Mobile introduced it in Germany in 2006. This was also the main reason that T-Mobile was the first network operator to be allowed to sell the famous iPhone, which in its first expansion stage could only use GPRS EDGE as a transmission standard.
Usage of GPRS
Initially, GPRS was used to connect a mobile LapTop PC to the internet. The mobile phone served as a wireless modem like a WLAN Modem. Some manufacturers also built plug-in cards that could be inserted into a PCMCIA slot on a LabTop, similar to early versions of WLAN modems.
However, mobile devices from around 2000 were hardly able to directly use Internet services such as browsing the World Wide Web. The limitations came from the display, which only had limited graphics capabilities, and from the lack of keyboards.