GPRS

GPRS is an always-on service. There is no need to dial up like you have to on a home PC for instance. This feature is not unique to GPRS but is an important standard that will no doubt be a key feature for migration to 3G. It makes services instantaneously available to a device.

GPRS is packet switched. Higher connection speeds are attainable at around 56-118 kbps, a vast improvement on circuit switched networks of 9.6 kbps. By combining standard GSM time slots theoretical speeds of 171.2 kbps are attainable. However in the very short term, speeds of 20-50 kbps are more realistic

Fixed IP addresses for mobile devices are not widely used due to shortages of Ipv4 addresses (see below). This information is stored in the HLR.

The second means of addressing is dynamic addressing. This is where a mobile device does not have its own IP address stored in the HLR. Instead the IP address is assigned to the GGSN domain. The method is also a type of dynamic IP addressing in which the IP address is assigned by RADIUS servers normally situated inside an IP network outside the mobile network, an example of this being when you dial up to an ISP from your home PC.

A mobile device is programmed with one or more Access Point Names which are commonly referred to as the APN's. An APN consists of a fully qualified DNS name e.g. rendc.org When a GPRS device wants to talk to rendc.org, the SGSN does a DNS look up and resolves the name to the correct GGSN. You could have multiple APN's programmed into your phone so you are not limited to a single service or GGSN.

Just because GPRS uses many of the components of a standard GSM network it would be foolhardy to assume that the same standards should apply. Things to be taken into account include provider general network architecture, radio interface and throughput. Here are some of the key elements briefly explained.

The Telecommunications Standard Institute has defined 3 new coding schemes for Radio Interface. When the GPRS device talks to the base station they can use 1 of the 4 schemes. The schemes are CS - 1 through CS - 3 where CS - 1 is the same as standard GSM. In simple terms CS - 1 is highly redundant but because of this is slow, 2 and 3 have less redundancy, whilst 4 has the least - removing all forward error control - but is capable of maximum throughput. If radio quality is bad then coding scheme 1 is used, as the quality improves less error control is needed.

An application can be assigned a Precedence Class 1, 2 or 3. If an application has a higher precedence (1) than another (3) then its traffic will be given a higher priority.

Applications can request predictive delay classes which guarantee an average and 95- percentile delay. There are 4 classes, 1 being the fastest.

Applications can request differing levels of reliability for its data depending on its tolerance to data loss.

Applications can choose different profiles for throughput. There are 2 distinctions in class, peak and mean. Peak throughput class is used mainly for bursty transmissions with a variable in octets per second describing the throughput required for burst of specified size. Mean is the average data transfer rate over a period of time measured in octets per hour.