. 37
( 132 .)


3GPP TS04.60: General Packet Radio Service (GPRS); Mobile Station (MS)“Base Sta-
tion System (BSS) interface; Radio Link Control/Medium Access Control (RLC/MAC)
3GPP TS04.64: General Packet Radio Service (GPRS); Mobile Station“Serving GPRS
Support Node (MS“SGSN) Logical Link Control (LLC) layer speci¬cation.
3GPP TS04.65: General Packet Radio Service (GPRS); Mobile Station (MS)“Serving
GPRS Support Node (SGSN); Subnetwork Dependent Convergence Protocol (SNDCP).
3GPP TS05.01: Physical Layer on the Radio Path (General Description).
3GPP TS05.02: Multiplexing and Multiple Access on the Radio Path.
3GPP TS05.03: Channel coding.

3GPP TS08.14: General Packet Radio Service (GPRS); Base Station System (BSS)“
Serving GPRS Support Node (SGSN) interface; Gb Interface Layer 1.
3GPP TS08.16: General Packet Radio Service (GPRS); Base Station System (BSS)“
Serving GPRS Support Node (SGSN) Interface; Network Service.
3GPP TS08.18: General Packet Radio Service (GPRS); Base Station System (BSS)“
Serving GPRS Support Node (SGSN); BSS GPRS Protocol.
3GPP TS22.60U: Mobile multimedia services including mobile Intranet and Internet ser-
3GPP TS23.060: General Packet Radio Service (GPRS) Service description; Stage 2.
3GPP TS23.107: Quality of Service (QoS) concept and architecture.
3GPP TS24.008: Mobile radio interface Layer 3 speci¬cation; Core network protocols;
Stage 3.
3GPP TS29.060: General Packet Radio Service (GPRS); GPRS Tunnelling Protocol (GTP)
across the Gn and Gp interface.
3GPP TS43.051: GSM/EDGE Radio Access Network (GERAN) overall description;
Stage 2.
A list of the current versions of the speci¬cations can be found at http://www.3gpp.org/
specs/web-table specs-with-titles-and-latest-versions.htm, and the 3GPP ftp site for the
individual speci¬cation documents is http://www.3gpp.org/ftp/Specs/latest/
IP Applications for


From its earliest inception, the Universal Mobile Telecommunications System (UMTS)
(Release 99) makes use of the Internet Protocol (IP) for data transport within the core
network. In this ¬rst release of UMTS, IP is used to route general packet radio service
(GPRS) traf¬c between the user equipment (UE) and external IP networks (for example,
the Internet) and across the UMTS core network using the GPRS tunnelling protocol
(GTP). This role of IP expands through release 4, culminating in release 5, where IP may
be used for the entirety of the transport (traf¬c and control) across the UMTS terrestrial
radio access network (UTRAN) through the core network and beyond. The use of IP
provides a number of bene¬ts. The IP protocol suite is a mature, well-tested technology
that has proven to be a highly robust and scalable architecture supporting many millions
of nodes on the Internet. IP has a vast range of applications (e.g. WWW and email) and
services (e.g. e-commerce and security) already developed for it. With the use of IP as a
transport mechanism these applications and services can be ported directly to the UMTS
environment. Finally, IP has the capability to carry mixed voice and data traf¬c ef¬ciently
through the use of a range of IP quality of service (QoS) protocols. To facilitate a smooth
progression to IP version 6 (IPv6) the UMTS speci¬cations state that both IP version 4
(IPv4) and IPv6 must be supported within each and every network element.
This chapter and Chapters 8 and 9 examine how IP is applied in the various releases
of UMTS. This chapter focuses on the IP protocol suite and describes how IP is used
within the UMTS R99 network for GPRS service. In particular, the provisioning of QoS
and security for GPRS networks is discussed in some detail. Later chapters show how the
role of IP is expanded within the network for releases R4, R5 and R6. With R4 all of the
UMTS core network will use packet switching but is still split into separate domains for

Convergence Technologies for 3G Networks: IP, UMTS, EGPRS and ATM J. Bannister, P. Mather and S. Coope
™ 2004 John Wiley & Sons, Ltd ISBN: 0-470-86091-X

circuit- and packet-type service. The voice traf¬c within the core for R4 will be moved
in packets using technologies such as voice over IP (VoIP) or voice over ATM. In R5
and R6 the UMTS network evolves to an all-IP architecture. In this, both the UTRAN
and the core network can use IP transport. The core network at this stage will consist of
only one packet switch domain and all services will be converged over a single network


The history of IP dates back to the 1970s, when its development was part of the US
Advanced Research Projects Agency (ARPA). At that time the primary objective was
the development of a network which was robust against a single point of failure. The
original network interconnected the US military with several research centres throughout
the US, and was known as ARPANET. In 1983, as the network development was coming
more from the research community, ARPANET split into MILNET for the military, with
the academic community retaining the rest of ARPANET. As the scope of ARPANET
increased, it eventually became known as the Internet.
This led to the development of a technology called packet switching, in which data
is broken up into variable size chunks called packets, marked with a header containing
a source and destination address, then sent into the network for delivery (Figure 5.1).
As the packet is sent between the internal network switches (called routers), each router
makes a decision on where to send the packet next to get it to its ¬nal destination based
on the destination address in the packet, the beauty of the scheme being that if a given
router in the network fails, packets can be re-routed via another path, thus keeping the



Station A Router

B A Data
Destination Source Router
Address Address
Station B

Figure 5.1 IP network operation

network in operation. Note in the context of IP, the term gateway is commonly used, and
is synonymous with the term router.
IP is an open standard, and its standardization process is under the coordination of
the Internet Engineering Task Force (IETF, www.ietf.org). The standards are published
through a process of Internet drafts, Request for Comment (RFC) documents and standards
(STD) documents.

5.2.1 IP protocol
The IP protocol suite was originally developed to support transport of data across the
Internet. Each packet in an IP network is referred to as a datagram, with the primary
purpose of the IP protocol being to route these datagrams across the network. Other
functions provided include such services as fragmentation and reassembly of packets
that are too large to be transported over a given network transport technology and the
appropriate handling of traf¬c with different priorities within the network. Figure 5.2
shows a diagram of an IP header, which is added to each packet before entering the
The header starts with the version number of the protocol. Currently version 4 is
used but the newer version, v6, is already being implemented by many manufacturers.
The following description actually applies to IPv4 (IPv6 is covered later in the chapter).
Following the version number is the Internet header length (IHL) ¬eld giving the length
of the header in 32-bit words. The usual value for this ¬eld is 5, resulting in a minimum
header length of 20 octets. The type of service ¬eld speci¬es reliability, precedence, delay
and throughput parameters. Historically, many routers have ignored this ¬eld; however,
it has been rede¬ned as the differentiated services (DiffServ) codepoint and can be used
to help provide differentiated QoS. The importance of this ¬eld will be seen when QoS
within the IP network is covered later in the chapter.
The total length ¬eld indicates the total length of the datagram before fragmentation.
The next line of the header consists of ¬elds which handle fragmentation within the
network. When fragmenting a packet, IP sets the identi¬cation ¬eld to the same value

0 4 Number of bits

Version IHL Type of Service Total Length
Identification Fragment Offset
TTL Protocol Header Checksum

Source Address

Destination Address

Optional part zero or more words. Each word must be 32 bits long.

Figure 5.2 IP header

for each fragment of the original datagram. This allows the receiver to work out which
fragment belongs to which datagram when reassembling.
The fragment offset, on the other hand, is used to indicate where an individual fragment
belongs in the whole datagram. This ¬eld is used to help put the fragments back together
in the correct order when they reach their ¬nal destination. As well as these two ¬elds
there are a couple of fragmentation ¬‚ags. DF stands for ˜don™t fragment™ and can be set
by a host that wishes to send data without fragmentation. The MF (more fragment) ¬‚ag
is set to 1 if there are more fragments to come; a value of 0 indicates the end of a list of
If the router itself has to perform fragmentation this will introduce an extra overhead in
terms of processing when forwarding the packet across the network. The use of fragmen-
tation headers also leads to an increased number of headers, decreasing the bandwidth
available for the payload. For these reasons fragmentation is only offered as an optional
service for IPv6, with the fragmentation information being relegated to the extension
headers. In this case each host is expected to send only datagrams no longer than the
maximum transfer unit (MTU) of the path across the network from source to destination.
The time to live (TTL) ¬eld is used to prevent misrouted packets circulating forever
on the network. Before forwarding a packet the router decrements this ¬eld by one; if
the value in the TTL ¬eld is zero then the packet will be discarded. The protocol ¬eld
indicates the higher layer protocol that is using IP to move its packets; in most cases
this will be TCP or UDP (see Sections 5.2.4 and 5.2.5). The header checksum protects
the header against corruption, and packets with an incorrect header checksum will be
The base header shown in Figure 5.2 can be followed by a number of extra optional
headers. The presence of these extra headers is indicated by an IHL greater than 5.
Table 5.1 describes some of the IP options available.

5.2.2 IP addressing and routing
IP addressing is based on the concept of hosts and networks. A host is anything on a
network that is capable of transmitting and receiving IP packets, such as a workstation or
a router. Hosts are connected together through one or more networks, therefore to send
a datagram to a host both its network and host address must be known. IP addressing

Table 5.1 IP options
Name Description
Record route Each IP router that forwards a packet adds its address to packets marked
with the record route option. This allows a packet™s route to be traced
across the IP network
MTU probe/reply Used to discovers the MTU for a given path across the network
Internet timestamp Allows router to timestamp packets as their route is recorded across the
Source routing Allows the source host to constrain the route of the packet across the

combines the network and the host into one single data structure called an IP address. A
mechanism called subnet masking is used to determine which part is the network address
and which part is the host address. This subnet mask can vary from situation to situation.
An IP address consists of 32 bits. By convention, it is expressed as four decimal
numbers separated by full stops, for example:

IP address
Subnet mask

Since there are four bytes, addresses range from to, which
gives a total of over 4 billion unique addresses. In fact, because of the way IP addresses
are allocated in practice, fewer addresses than this are actually available.
There are various classes of address, as shown in Table 5.2. Note for each class there
is a limit on the total number of network and host addresses available.
When trying to identify an address as class A, B, C, D or E, the simplest way is to
look at the address ranges given in Table 5.2. Class A addresses start with 1“126, class B
addresses start with 128“191, class C addresses with 192“224 and multicast addresses
with 224“239.
For example, is easily identi¬ed as a class B address and therefore supports
addresses in the range to
Subnetting is the division of the single network with many hosts into smaller subnet-
works with fewer hosts. We do this by using part of the 32-bit IP address to indicate
the subnetwork and another part for the actual host address. For example, with class C
we could have 6 subnets each with 30 hosts. Figure 5.3 shows how we can increase the
number of subnetworks at the expense of the number of hosts on those subnetworks.
It is possible to increase the number of 1s to generate a subnetwork, as shown in
Table 5.3 (a class B network is assumed in this example). This mechanism of using

Table 5.2 IP address classi¬cation
Class Network Host Total network Total host Address range,
address address addresses addresses dot decimal form
¬eld size ¬eld size
A 7 24 126 16 777 214“
B 14 16 16 383 65534“
C 21 8 2 097 151 254“
D (multicast) 28“
E (reserved)“

32 bits

Network Subnet Host

Figure 5.3 Subnet and host tradeoff


. 37
( 132 .)