GPRS Services

GPRS is designed to provide packet data services at higher speeds than
those available with standard GSM circuit-switched data services. In theory, GPRS could provide speeds of up to 171 Kbps over the air interface,
though such speeds are never achieved in real networks (because, among
other considerations, there would be no room for error correction on the
radio frequency [RF] interface). In fact, the practical maximum is actually
a little over 100 Kbps, with speeds of about 40 Kbps or 53 Kbps more realistic.
Nonetheless, once can see that such speeds are far greater than the
9.6-Kbps maximum provided by standard GSM.
The greater speeds provided by GPRS are achieved over the same basic
air interface (that is, the same 200-kHz channel, divided into eight timeslots).
With GPRS, however, the mobile station (MS) can have access to more
than one timeslot. Moreover, the channel coding for GPRS is somewhat different
than that of GSM. In fact, GPRS defines a number of different channel
coding schemes. The most commonly used coding scheme for packet
data transfer is Coding Scheme 2 (CS-2), which enables a given timeslot to
carry data at a rate of 13.4 Kbps. If a single user has access to multiple
timeslots, then speeds such as 40.2 Kbps or 53.6 Kbps become available to
that user.Table 5-2 lists the various coding schemes available and the associated
data rates for a single timeslot.
The air interface rates in Table 5-2 give the user rates over the RF interface.
As we shall see, however, the transmission of data in GPRS involves a
number of layers above the air interface, with each layer adding a certain
amount of overhead. Moreover, the amount of overhead generated by each
layer depends on a number of factors, most notably the size of the application
packets to be transmitted. For a given amount of data to be transmitted, smaller application packet sizes cause a greater net overhead than
larger packet sizes. The result is that the rate for usable data is approximately
20 to 30 percent less than the air interface rate.
As mentioned, the most commonly used coding scheme for user data is
CS-2. This scheme provides reasonably robust error correction over the air
interface. Although CS-3 and CS-4 provide higher throughput, they are
more susceptible to errors on the air interface. In fact, CS-4 provides no
error correction at all on the air interface. Consequently, CS-3 and particularly
CS-4 generate a great deal more retransmission over the air interface.
With such retransmission, the net throughput may well be no better than
that of CS-2.
Of course, the biggest advantage of GPRS is not simply the fact that it
allows higher speeds. If that were the only advantage, then it would not be
any more beneficial than High-Speed Circuit-Switched Data (HSCSD),
described later in this chapter. Perhaps the greatest advantage of GPRS is
the fact that it is a packet-switching technology. This means that a given
user consumes RF resources only when sending or receiving data. If a user
is not sending data at a given instant, then the timeslots on the air interface
can be used by another user.
Consider, for example, a user that is browsing the Web. Data is transferred
only when a new page is being requested or sent. Nothing is being
transferred while the subscriber contemplates the content of a page. During
this time, some other user can have access to the air interface resources,
with no adverse impact to our Web-browsing friend. Clearly, this is a very
efficient use of scarce RF resources.
The fact that GPRS enables multiple users to share air interface
resources is a big advantage. This means, however, that whenever a user
wants to transfer data, then the MS must request access to those resources
and the network must allocate the resources before the transfer can take
place. Although this appears to be the antithesis of an “always-connected”
service, the functionality of GPRS is such that this request-allocation procedure
is well hidden from the user and the service appears to be “always-on.”
Imagine, for example, a user that downloads a Web page and then waits
for some time before downloading another page. In order to download the
new page, the MS requests the resources, is granted the resources by the
network, and then sends the Web page request to the network, which forwards
the request to the external data network (such as the Internet). This
happens quite quickly, however, so that the delay is not great. Quite soon, the new page appears on the user’s device and at no point does the user
have to dial-up to the ISP. 173