1

CONTENTS


1. Executive Summary.................................................................................................................2
2. Introduction..............................................................................................................................2
2.1. The Limitations of Legacy Technology...........................................................................3
2.2. The Next Generation Challenge.....................................................................................3
3. 3GPP Bearer- Independent Circuit-Switched Core Network Architecture..............................4
3.1. 3GPP Release 4 Architecture.........................................................................................4
3.1.1 Mc interface............................................................................................................5
3.1.2 Nc interface............................................................................................................6
3.1.3 Nb Interface............................................................................................................6
3.2. Rel- 4 Network Elements................................................................................................7
3.2.1 MSC Server............................................................................................................7
3.2.2 GMSC Server.........................................................................................................7
3.2.3 Media Gateway......................................................................................................8
3.3. GSM to UMTS Evolution.................................................................................................8
4. Transmission Network.............................................................................................................8
4.1. Transmission Network Solutions ....................................................................................8
4.2. ATM or IP........................................................................................................................9
4.3. Packet Voice Transmission Efficiency..........................................................................10
4.4. Compression in Wireless..............................................................................................11
4.5. Quality of Experience....................................................................................................11
5. Beyond R4 IP Multimedia Subsystem (IMS)......................................................................12
5.1. Architecture Overview...................................................................................................14
6. Conclusions...........................................................................................................................15
Acknowledgments......................................................................................................................16

2
1. EXECUTIVE SUMMARY
In Release 4, 3GPP introduces a bearer-independent core network architecture that enables
packet networks as the bearer, in addition to the TDM bearers supported in current networks.
Both IP and ATM are supported.
This architecture allows the splitting of the monolithic MSC network elements into MSC Servers
that handle the control plane, and Media Gateways that handle the user plane.
The architecture supports all existing circuit-switched mobile services, which means that it will
enable the operators to provide all existing services to end users while reaping the benefits of
the split architecture and a packetized core network.
The separation of the control and user planes enables each function to be scaled independently
of the other, and be geographically independent of the other.
Utilizing packet networks in the user plane allows any-to-any node connectivity without
exponentially increasing the amount of connections, thus providing transmission savings.
Redundant capacity in the network can therefore be minimized.
Most of the new services introduced in emerging networks will be based on packet technology.
Utilizing a common packet network for both the legacy circuit-switched services and new
services eliminates the need of running parallel networks.

2. INTRODUCTION
3
rd
Generation wireless standards have defined an open architecture for the packetization of
wireless telephony networks, giving wireless operators a clear alternative to traditional TDM
technologies. The packetization of existing wireless telephony services is the first step toward
achieving the all packet network architecture, widely recognized as the ultimate goal for
network operators. This new architecture is defined in the 3GPP Release 4 Bearer Independent
Core Network (BICN).
This solution provides the same service capabilities as are available using traditional TDM
technologies. However, packet voice technology enables an operator to deliver these services
at a much lower cost. This substantially lower cost base enables an operator to be either more
competitive in their chosen market or to more profitably address new lower revenue markets.
The lower cost base is achieved through both capital and operational cost savings.
The consolidation and simplification of the core network into fewer higher capacity and higher
density network elements brings a variety of cost savings which are discussed in more detail
later. The consolidation and simplification of the core network is at the heart of the principles of
scalability and efficiency. These principles enable several benefits from a business perspective:
Distributed (flat-architecture) packet networks require less equipment than hierarchical TDM
networks, reducing overall capital investment in the network and the cost of ownership
associated with each network node.
The use of highly scalable network elements enables the network to be built using far fewer
nodes, thereby reducing internal network overhead and greatly improving network
efficiency.
Floor space always has a cost associated with it and the use of high-density packet
technology instead of traditional TDM equipment greatly reduces the floor space needed.
(This is the application of Moores Law.)
Packet networks enable more efficient use of transmission capacity facilitating the scalability
of network interfaces. The use of packet-based interfaces also has a direct bearing on the
scalability of network nodes.
3
Network equipment no longer has to be at legacy central office sites at geographically
constrained locations, but rather can be geographically distributed and scaled accordingly.
One common and scalable packet network is used for everything - voice, data and signaling
simplifying network operation so that there is no need for multiple over-lay networks,
thereby reducing network complexity and the operational overhead.
2.1. THE LIMITATIONS OF LEGACY TECHNOLOGY
Many mature legacy wireless network operators are experiencing high operational costs due to
the inefficiency and complexity of their legacy networks. With wireless markets continuing to
become more and more competitive, these operators are naturally looking for ways to lower
their costs. This is not only relevant to operators in mature networks but also to operators
building Greenfield networks, and those looking for some reassurance that their network
capacity investments will be future proof.
The traditional circuit-switched network architecture is based on monolithic network elements
which has close coupling of the control and bearer planes. This close coupling has
fundamentally limited the scalability of traditional network elements, such as in a MSC. Close
coupling of the control and bearer plane has also meant that a complete MSC must be installed
in each geographical area to be covered. Consequently, when MSC capacity limits are reached
or further geographical coverage is needed, network growth would be achieved by adding more
and more MSCs.
There are two types of growth to consider: growth due to increased geographical coverage, and
growth due to increased subscriber penetration. When the capacity of an MSC is exceeded, the
only solution is to add another MSC. In the early days, when the number of MSCs was low and
the geographical area covered by an MSC was quite large, adding more MSCs was not a major
problem. However, as subscriber numbers increased and switch coverage areas shrank, the
number of nodes in the network steadily increased, with each MSC needing to communicate
with all of the other MSCs in the network. So every time another MSC was added to the
network, the complexity of inter-MSC communications increased exponentially.
This is also a common networking issue in wireline networks, often referred to as The N
2

Problem. In wireless networks, the complexity of this communications mesh is compounded by
the need to maintain up-to-date mobility information for both active and inactive subscribers -
the more mobility information to be kept up-to-date, the greater the network inefficiency from the
N
2
Problem. This is at the heart of the inefficiency of legacy wireless networks, which has a
direct bearing on the efficiency savings from more scalable equipment.
The scalability limitations of circuit switches constrains network growth as it becomes
increasingly difficult to implement a full physical mesh at only one network layer. To overcome
this problem, a whole new hierarchy of circuit switches has to be added to provide transit layer
interconnection of the MSCs. These transit layers typically do not provide end user services and
consequently represent increased operational cost for no extra service revenue to the operator.
Transit network layers are often implemented for both bearer and signaling traffic, creating
multiple overlay networks with corresponding capital and operational costs. Using closely
coupled circuit switches in a mature wireless network becomes a spiders web of hierarchical
complexity. In the end, a high proportion of the networks capability is simply focused on
maintaining this level of complexity, rather than on earning service revenue for the operator.

2.2. THE NEXT GENERATION CHALLENGE
The challenge for the target network architecture is how to enable an operator to grow from the
early network to the mature network without adding the complexity and inefficiency. Clearly,
complexity and inefficiency in a network is closely related to the lack of scalability of a MSC and
its interfaces, as measured by the added number of MSCs needed to match network growth.
The answer to this seeming conundrum is buried in the close coupling of circuit switches and
the two types of network growth.
4
First, the new target network architecture must decouple the control and bearer plane. This
means that the control plane and the bearer plane should scale independently of each other,
rather than limit the scalability of each as they do in traditional circuit switches. Bearer capacity
must scale according to the traffic needs of the geographical area that is to be covered. This
covers anywhere from low capacity levels in new market scenarios, all the way through to high
capacity levels in high-penetration mature markets - all achieved without nodal proliferation.
This enables efficient network expansion for geographical coverage followed by efficient nodal
growth as penetration rates increase.
Decoupling removes the geographical constraints of the control plane capacity from the bearer
plane, and eliminates the need for local presence when extending geographical coverage. Once
this is achieved, control plane capacity is free to scale as a network wide resource, scaling from
the early to late market as needed. The geographical independence of the control plane frees
the operator from any constraints of circuit switching, enabling control plane capacity to scale in
the most cost effective way. In practical terms, this might include the divestiture or consolidation
of property portfolios, as well as a more efficient application of highly skilled and costly labor.
This target architecture is further defined in 3GPP Release 4 Bearer Independent Core Network
standards.

3. 3GPP BEARER- INDEPENDENT CIRCUIT-
SWITCHED CORE NETWORK ARCHITECTURE
3.1. 3GPP RELEASE 4 ARCHITECTURE
Within 3GPP Release 4, the concept of a Bearer-Independent Circuit-Switched (CS) Core
Network was introduced. This built upon Release 99 and provides for a flexible architecture
with the option for ATM or IP transport for the core network. The network architecture is
specified in 3GPP TS 23.002 Network Architecture and 3GPP TS 23.205 Bearer-Independent
Circuit-Switched Core Network. The option for IP bearer support on the Iu interface from the
core network to the radio network was added in 3GPP Release 5. 3GPP Release 4 is
completed and has been stable since 2003.
The purpose of the Bearer-Independent Circuit-Switched Core Network architecture is to
provide the capability to split the control and user planes into separate network entities and to
enable usage of a packet-switched backbone in the network. To ensure service continuity, it
was important that all end user services (such as supplementary services, SMS and bearer
services) as well as network services (such as IN, mobility management, and location services)
be unaffected.
At the same time and to ease deployment, the architecture needed to minimize the impact on
other network domains, e.g. the access network (BSS and RAN), the SS7 network, the
customer care and billing systems, the IN network entities, and the location service entities.
5
MGW
Signalling and Data Transfer
Interface
Signalling
Interface
UTRAN
PSTN/
Legacy/External
HLR
Applications
& Services
MSCserver GMSC server
Mc
D
C
MGW
Nb
Nc
Iu
Iu
CAP
CAP
Mc
GERAN
A
A

Figure 1: CS Core Network Logical Architecture

The Bearer-Independent Circuit-Switched Core Network (see Figure 1) enables the support of
different transports (e.g. ATM or IP) in a bearer-independent fashion. For the ATM and IP
transport, there is a strict separation between the call control level and the bearer control level.
In the case of ATM or IP transport, the passage of compressed speech at variable bit rates is
possible through the CS core network.
In order to split the control and user plane, the monolithic MSCs from Release 99 have been
separated into the MSC server, GMSC server and media gateways (see Figure 1). The GMSC
server and MSC server provides the call control and mobility management functions, and the
media gateway provides the bearer control and transmission resource functions, and contains
the stream manipulating functions. The GMSC server and MSC server are connected to the
media gateway via the Mc reference point. The MSC server and GMSC server are connected to
each other with the Nc reference point. There may be a number of call control transit nodes
between the MSC server and GMSC server in the Nc reference point. The MGWs are
connected with the Nb reference point.
By simply splitting the existing MSCs and defining interfaces between the MSCs and MGWs,
while maintaining the existing interfaces to other nodes (e.g. interfaces to the SCP, HLR), there
is no impact on existing services. Hence, the users connected to the Bearer-Independent
Circuit-Switched Core Network are not aware whether a MSC server/media gateway
combination is used, or a monolithic MSC is used.
The architecture can thus be seen as a CS core internal issue without impact on the other
reference points, network elements and subsystems.
3.1.1 MC INTERFACE
The Mc reference point defines the interface between the (G)MSC server and the MGW. The
H.248 protocol, together with 3GPP specific extensions (i.e. packages), is used over this Mc
interface.
H.248/MEGACO has been jointly developed within the ITU-T and the IETF, and supports a
separation of call control entities from bearer control entities, and a separation of bearer control
entities from transport entities. H.248 is used on the Mc interface between the (G)MSC servers
and the media gateway.
6
3.1.2 NC INTERFACE
The Nc reference point defines the interface between MSC servers. The Bearer Independent
Call Control (BICC) protocols defined by ITU-T are used over the Nc interface.
The BICC architecture as described in ITU-T Q.1902 consists of the following types of protocols:
call control protocol, bearer control protocols and a resource control protocol.
The call control protocol is based on ISUP and enables setting up different types of bearers
required in the Nb interface between MGWs.
3.1.3 NB INTERFACE
The Nb reference point defines the interface between MGWs. Bearer control and transport are
performed over the Nb reference point, which may be RTP/UDP/IP or AAL2/ATM for the
transport of user data. Both IPv4 and IPv6 can be used.
The user plane protocol used between two media gateways in the CS core network is referred
to as the Nb UP protocol. The Nb UP protocol is very similar to the Iu UP protocol used between
RNC and MGW.
The Nb UP framing is identical to the Iu UP framing (i.e. the same PDU types are valid for both
protocols).
MGW MGW
Nb Iu
T
r
a
n
s
p
o
r
t

L
a
y
e
r
SRNC
Radio
Protocols
Iu UP Iu UP
Nb
UP
Nb
UP

Figure 2: Nb UP Protocol Layer Occurrence in Overall Architecture
Figure 2 shows the logical location of the Nb UP protocol layer in relation to the Nb interface. Nb
UP defines initialization, rate control and time alignment procedures in addition to user data
transport.
The AAL2 Signaling Protocol defined in ITU-T Q.2630.2 is used for the establishment of AAL2
connections.
The ITU-T Recommendation Q.1970 BICC IP Bearer Control Protocol (IPBCP) is used for IP
bearer establishment. IPBCP is transported over the Mc and Nc interface by means of the ITU-T
Recommendation Q.1990 BICC Bearer Control Tunneling Protocol.
Nb supports AMR, EFR and G.711 codecs.

7
MGW MGW
MSC-Server MSC-Server
Nc
Mc
Nb
Mc
TS 29.232
BICC: Q.765.5
Tunnel: Q.1990
IPBCP: Q.1970

Figure 3: Transport of IPBCP

3.2. REL- 4 NETWORK ELEMENTS
3.2.1 MSC SERVER
The MSC Server mainly comprises the call control (CC) and mobility control parts of a MSC. It is
also integrated with a VLR to hold the mobile subscriber's service data and CAMEL-related
data. The MSC Server is responsible for the control of mobile-originated and mobile-terminated
CC CS domain calls.
The MSC server terminates the user-network signaling (see 3GPP TS 24.008) and translates it
into the signaling over the Nc interface. It also terminates the signaling over the Mc interface
with the media gateway. The MSC server controls the parts of the call state model that pertain
to connection control for media channels in an MGW. It also contains the 'Call Control Function'
in the BICC model.
3.2.2 GMSC SERVER
The GMSC server mainly comprises the call control and mobility control parts of a GSM/UMTS
GMSC as described in 3GPP TS 23.002.
The GMSC server terminates the signaling over the Nc interface and the call control interfaces
to the external networks. It also terminates the signaling over the Mc interface towards the
media gateway. The GMSC server controls the parts of the call state model that pertain to
connection control for media channels in an MGW. It also contains the 'Call Control Function' in
the BICC model.
If a network delivering a call to the PLMN cannot interrogate the HLR, the call is routed to a
MSC. The MSC will interrogate the appropriate HLR and then route the call to the MSC where
the mobile station is located. The MSC, which performs the routing function to the actual
location of the MS, is called the Gateway MSC (GMSC).
The choice of which MSCs can act as Gateway MSCs is for the operator to decide (i.e. all
MSCs or certain designated MSCs).
8
3.2.3 MEDIA GATEWAY
The media gateway (MGW) terminates the signaling over the Mc interface from the (G)MSC
servers. It also terminates the bearer part of the signaling over the Iu interface and the Nb
interface. A MGW terminates bearer channels from a circuit-switched network and media
streams from a packet network (RTP streams in an IP network or AAL2 channels in an ATM
network).
The MGW bearer control and payload processing capabilities also needs to support mobile
specific functions such as SRNS relocation/handover and anchoring. The MGW supports 3GPP
bearer and supplementary services. This means that it should support 3GPP specific CS data
handling, as well as have support for services such as conferencing, tones, DTMF
detection/generation and announcements.
For support of Iu interface, the MGW supports AMR codec. EFR support is also useful, as an
option for Iu, and also provided for codec compatibility in 2G to 3G relocation. For support of Nb
interface, AMR and G.711 codecs are supported over the NbUP.
At a minimum, echo cancellation and automatic level control are required for speech quality.
Regulatory requirements such as Lawful Interception (i.e. CALEA) are also supported.
3.3. GSM TO UMTS EVOLUTION
The CS services in GSM and UMTS are nearly identical. The most notable difference relates to
the capability of UMTS to support transparent 64 kbit/s synchronous CS data services that
enables services such as video calling. Thus, from the service control and control plane points
of view, the CS core provides excellent evolution possibilities. In most cases, the same
infrastructure can be used.
For the user plane, the UMTS architecture differs slightly from GSM with the transcoder being
located in the core network (vs. the access network with the GSM access). However, in most
cases the location of the codec is a minor issue compared to the support of the CS-related
bearer and supplementary services. Despite this difference, the Bearer-Independent Circuit-
Switched Core Network supports both UMTS and GSM networks. More information can be
found in the 3GPP Technical Report TR 23.977.
Introducing UMTS as an overlay network and the recommendation to utilize excess capacity in
GSM CS Core Network are common and valid approaches for supporting UMTS access.

4. TRANSMISSION NETWORK
The data networking industry has been debating whether to use IP or ATM in Carrier Data
Networks for many years now. Strong opinions are routinely voiced on both sides of this
argument and several claims of victory have already been made prematurely. In reality, there
are still no signs of this debate reaching a conclusion. Strong sales of both types of Carrier Data
products have continued, even through the depths of the telecommunications economic
downturn.
4.1. TRANSMISSION NETWORK SOLUTIONS
One of the key criteria behind the choice of VoIP or VoATM is the nature of the transmission
network infrastructure that is available to the wireless operator. One of the business drivers for
many operators is how to achieve operational cost reduction through transmission capacity
savings. Next Generation Networks can enable operators to achieve substantial efficiency
gains, which becomes an important factor when choosing the technology to use. As a business
driver, the relative transmission network efficiency of each technology must be considered.
9
Many wireless operators have traditionally taken the view that their business is concerned more
with wireless service provision than with transmission network provision. Consequently, this has
often been outsourced to other network operators who have specialized in this kind of
business. For this reason many wireless networks use leased line transmission capacity, which
represents a recurring operational cost to the business. An advantage to this approach is that
the wireless operator is free to focus on its core business of wireless service provision without
diversionary distractions into other business directions. If transmission network capacity is
abundant or transmission costs are low, the operator may not be concerned by this operational
cost.
However, if capacity is scarce or expensive, or if the distances involved are long, then a Next
Generation Networking business driver might be to achieve transmission cost savings. In this
case, the efficiency of the protocol used becomes important. There is no simple rule that can be
applied; however, as the cost of transmission capacity can vary enormously between countries,
and even within the same country. In recent years, some countries have seen substantial capital
investment in transmission capacity, which has led to increased competition and lower
transmission capacity prices.
Even so, some operators have chosen to invest in their own optical fiber infrastructure, which
requires substantial capital investment. This financial justification is often based on the savings
of transmission network costs that was discussed above. The availability of existing capacity
and its corresponding pricing trends must clearly be considered.
Next Generation Networks provide an alternative solution to implementing an optical network.
Wireless operators or their parent companies that have already implemented an optical network
are effectively exchanging the ongoing recurring operational cost of leased lines for an
immediate capital investment in their own optical infrastructure. In many cases this could be
classified as diversification from wireless into the transmission network business. Looking at
where investment in an optical network has already been made, the wireless operator has
created access to abundant low cost transmission capacity. It should be noted that in this
scenario, the operator is typically not necessarily concerned with packet voice transmission
efficiency.
4.2. ATM OR IP
In 3GPP Release 4, both IP and ATM transport is supported in the CS core network. In the Iu
CS interface, IP-based transport is introduced by 3GPP Release 5. From a standards point of
view, both transport solutions are valid choices.
ATM is the more conventional approach, with well-understood QoS mechanisms containing
hard guarantees. ATM is used as transport in current WCDMA access networks and Iu
interfaces, and is thus widely supported. It is also available in most markets. In the user plane,
3GPP defines the usage of AAL2 to enable efficient utilization of bandwidth - with AAL2
channels being created on a call-by-call basis.
However, ATM network management is inherently more complex. To be able to set up AAL2
channels on a call-by-call basis, ATM VCC must exist before the AAL2 channel can be set up.
This means that SVCs cannot be used on the ATM level. Network planning and management is
further complicated in big networks due to the limited number of ATM switches capable of
switching on the AAL2 level that are deployed or available. ATM is also generally perceived as
more expensive from the capital expenditure point of view.
IP transport on the other hand is often perceived as lacking on its ability to provide guaranteed
QoS solutions. This is partly based on perception, but also reflects how IP networks typically
have been designed and managed. With proper design and management, IP QoS is not an
issue on current networks. Since traffic is generated by elements controlled by the network
operators, the problem is further minimized when considering IP transport in the CS core. With
admission control at the edges of the network (in A- or Iu-interface and the PSTN-interface), the
traffic type and behavior is more predictable than, for example, the Internet. Due to the nature of
IP, the networks can be significantly simpler from a management point of view.
10
From a network evolution perspective, IP is a more natural fit when considering current
development trends in networks and backbone infrastructure. This applies both to service
development trends, as well as technology development trends. ATM and IP co-existence, and
ATM to IP(/MPLS) migration issues are also well understood and supported by all major
backbone vendors.
Thus, it can be concluded that both ATM and IP (or a mix of technologies) are all valid options.
ATM might be more relevant in cases where a more conventional and familiar approach is
preferred, and might even be dictated by deployed infrastructure. IP on the other hand might be
preferred from a strategic viewpoint when taking operational and capital expenditure
considerations into account.
4.3. PACKET VOICE TRANSMISSION EFFICIENCY
The most efficient way to transport digital speech is through conventional TDM techniques,
primarily because there is no additional header information needed in TDM and consequently a
64kpbs constant bit rate voice stream is all that is needed. However, significant capacity savings
can be achieved by compressing speech or by suppressing unnecessary traffic, such as
silence. When this is achieved, the digital speech is no longer a constant bit stream of 64kbps,
but is a highly bursty variable bit rate bit stream. With silence, the bit rate can be close to zero.
But if this were to be transmitted in the conventional TDM way, the full 64kbps would still need
to be used, even if the actual compressed speech bit rate was close to zero.
In order to realize the capacity savings achieved through compression and silence suppression,
the variable-bit-rate traffic must be sent in packet form. To accomplish this, a packet header
must be attached to the payload, which has the opposite effect on compression in that it actually
increases the amount of information that has to be sent. If the amount of header information is
greater than the amount of bandwidth saved through compression, then the total capacity used
would be less efficient than TDM. If saving bandwidth is the objective, then there may be no
point in using packet speech in this case.
Mobile codec compression in the handset can be applied using high compression ratios to
optimize the efficiency of the radio network interfaces used. Mobile codec compressed speech
can provide significant savings against TDM networks even if large packet headers have to be
attached. However, this is only possible if the speech stays compressed for the complete end-
to-end speech path. To calculate the optimum bandwidth efficiency in the backbone for
compressed speech, we need to compare the packet header efficiency of the different transport
protocols for the whole protocol stack.
The transmission network capacity used by packet voice is influenced by several factors:
The rate of compression for the call achieved by the voice codec (if it is used on the
backbone).
The voice sampling rate.
Additional traffic reduction techniques, such as silence suppression.
The efficiency of the packet network protocol used for the call. That is, the overhead cost
varies depending on the protocol (e.g., VoIP or VoATM). Each protocol offers more than
one implementation option, each with its own efficiency implications.
11
4.4. COMPRESSION IN WIRELESS
The discussion on transmission network efficiency provided above introduces the utilization of
voice codecs in wireless voice transport. Wireless terminal devices compress speech prior to
being transmitted over the radio interface in order to make optimum use of the available radio
capacity. This is a major difference from wireline terminal equipment where this does not occur.
Consequently, in wireless networks it makes sense to leave this pre-compressed speech in its
compressed form since decompressing it would add to voice path delay, degradation in voice
quality and an increase in the amount of bandwidth needed to carry it on the network. In the
TDM networks operating today for both GSM and UMTS, these downfalls are prevalent. So
unlike wireline networks where bandwidth saving is achieved by adding compression, in
wireless networks it is achieved by simply avoiding decompression. This results in fewer delays,
higher voice quality and less traffic on the network. This benefit is equally applicable to both
VoIP and VoATM, and although is not a major consideration in the choice of packet protocol, it
is a significant factor in choosing to move from TDM to packet.
4.5. QUALITY OF EXPERIENCE
Quality of Experience highlights important issues such as service continuity and Quality of
Service.
From a service point of view, the 3GPP Rel-4 CS core network architecture provides seamless
continuity. All Rel-99 and earlier services are supported. This includes tele-services, bearer
services and supplementary services. The end user is unaware of which type of CS core
network he is being served by.
Quality of Service in mobile networks is mainly affected by the nature of the radio interface
(need for compression, packet loss, jitter and delay). These factors also apply to using packet-
switched technologies for the fixed portion of the call.
The following table further elaborates on end user / application QoS requirements according to
3GPP TS 22.105 Service and Service Capabilities.
Table 1: End-user Performance Expectations - Conversational / Real-time Services
Medium Application
Degree of
symmetry
Data rate Key performance parameters and target values

End-to-end
one-way
delay
Delay
variation
within a
call
Information loss
Audio

Conversational
voice


Two-way

4-25 kb/s

<150 msec
preferred
<400 msec limit
Note 1

<1 msec

<3% FER
Video


Videophone Two-way
32-384
kb/s
<150 msec
preferred
<400 msec limit
Lip-synch : <
100 msec

<1% FER

Data

Telemetry
- two-way
control
Two-way
<28.8
kb/s
<250 msec N.A

Zero
Data
Interactive
games
Two-way <1 KB <250 msec N.A Zero
Data Telnet
Two-way
(asymmetric)
<1 KB <250 msec N.A Zero
Note: The overall one-way delay in the mobile network (from UE to PLMN border) is
approximately 100 msec.
12
In practice, there exists a delay of 80-100ms (UL) with the mobile and the access network,
which means that even with no delay in the core network, the preferred values of <150ms will
not be reached in a mobile-to-mobile call.
When using packet-switched transport in the CS core network, the core network introduces
further delay. The delay mainly comprises packetization delay and jitter buffers required to keep
the jitter within acceptable values. The IP or ATM backbone itself, as well as the routers and
switches in the backbone, introduce insignificant delay compared with the delay of the end
points and transmission.
As in traditional mobile CS networks, QoS is highly affected by the used codecs. 3GPP has
defined Tandem Free Operation (TFO) to provide codec transparency when originating and
terminating parties support the same codecs. TFO can also be supported over 3GPP Rel-4 CS
core networks. 3GPP has further defined Out of Band Transcoder control (also called
Transcoder Free Operation or TrFO) in UMTS access cases to provide both codec transparency
and bandwidth savings when the Rel-4 CS core network architecture is used. In TrFO, the
transcoders in the core network are effectively bypassed and inserted only when the core
network needs it for inter-working reasons. 3GPP is currently working on providing similar
benefits for a TFO type of solution for the TrFO, enabling the same level of service for GSM as
for UMTS accesses.

5. BEYOND R4 IP MULTIMEDIA SUBSYSTEM
(IMS)
The Always On subscriber is slowly coming of age. Ubiquitous broadband access is ushering
in new consumer and lifestyle convergence communication, entertainment, work, leisure,
distance, and time. The demand for new and sophisticated ways of being in constant contact
everyday, enlarging the community for leisure and entertainment anywhere and anytime, and
enhancing the effectiveness and the efficiency in the workplace, have all dramatically increased.
For operators, business success can no longer be guaranteed by making incremental
improvements in their existing products, services and business models within this new
environment.
The third generation body 3GPP is driving these rapid changes by specifying the IP Multimedia
Subsystem (IMS), starting with Release 5, as the enabling vehicle by which this demand will be
met. IMS is a new layer in the packet domain and will provide person-to-person, real and non
real-time integrated multimedia services such as voice, video, instant messaging, on-line
gaming, web sharing and many other services over a single IP network. IMS will bring the full
potential of Internet technology to mobile users, allowing the creation of a new business cycle
for wireless operators.
13

















At network level ubiquity broadband access, Packet Switching technologies, SIP protocol, have
been adopted as the multimedia convergence technologies. However, one of the key questions
that the industry will need to ask itself is: to what extent will mobile services integrate and inter-
operate with Internet services in order to satisfy these new market demands. Deploying
profitable and cost effective solutions that enable multimedia services to be compelling, flexible
and easy to use represents the new challenge for network operators.
GSM infrastructure providers are leveraging their own unique data and networks expertise to
help this transformation by building convergent networks able to deliver services across various
networks and media. This results in bringing new value to the end user experience and new
streams of revenues to the operators.


Enhanced Experience
Simplified usability
Seamless services access
Single User ID
Single bill
Revenue generation
Stimulate data usage
Churn Reduction
Services control
OPEX/CAPEX optimized
Seamless services offer
end-user
operator
14
5.1. ARCHITECTURE OVERVIEW
The following diagram indicates the main components of the IMS architecture:















CSCF - Next generation call control: The Call State Control Function (CSCF) is an enhanced
SIP proxy (i.e. an entity that routes SIP messages). Additional wireless capabilities are added
(i.e. SIM-based authentication, billing, QoS control). Three "flavors" of CSCF have been defined
(P-CSCF for Proxy, I-CSCF for Interrogating, and S-CSCF for Serving) potentially allowing a
better repartition of the functionalities in different nodes, thereby providing more scalability, as
well as roaming capabilities.
HSS - Centralized Subscriber Data/Authentication: The 3GPP IMS relies on the packet core
for connectivity to the handset/client. The provisioning and authentication data is held at the
HLR. The 3GPP IMS also needs its own provisioning and authentication information, therefore
the HLR is extended to hold this information in one place the Home Subscriber Server (HSS).
The Home Subscriber Server (HSS) is the master database for a given subscriber in the IMS
domain. It is a key network element in the IMS network environment that facilitates mobility and
service management for the IM-subsystem domain.
Open Applications Architecture: 3GPP defined a high-level architecture to allow SIP
application servers, 3rd party applications and CAMEL SCPs to inter-work with the call control.
This concept would allow the wireless operator to easily mix and match applications from
different vendors and re-use legacy IN services. The service-triggering mechanism (through the
standardized ISC interface) would offer the possibility of providing many different services
based on a dedicated application server, controlled by either the wireless service provider or
through a 3rd party. The ISC would also allow interfacing with OSA GW or existing IN/CAMEL
platforms by means of the IM-SSF gateway.
Connectivity: Packet and Access enhancements to allow real time services to work, many
changes are required in the access and packet domains. This ensures appropriate Quality of
Service (QoS) behavior for real time services (jitter and delay) and ensures the IP bearer is
efficiently used. In addition, the packet core needs to be able to handle IPv6 addressing as
defined by the 3GPP for the IMS.
PDF: Quality of Service and admission control in the wireless domain, control of radio
resources is critical. Hence, 3GPP has added procedures to ensure the application layer
(CSCF) and bearer layer (GGSN) are well coordinated. The Policy Decision Function (PDF) is
GGSN
MGCF
Application Layer
Control Layer
Connectivity Layer
MGW
HSS
Go
PSTN/
PLMN
Internet/
Intranet
I/S/P
CSCF
R4 BICN
MGW
MSC
Server
Internet/
Intranet
15
responsible for coordinating the set up of bearers (secondary PDP context establishment) with
session setup (SIP Invites). In addition, application layer and bearer layer billing IDs are
exchanges for billing correlation.
MGCF, IMS-MGW: Circuit Domain Inter-working the 3GPP standards define the PSTN inter-
working architecture in R5, with detailed call flows in R6. The Media Gateway Control Function
(MGCF) is roughly equivalent to the Call Server in a R4 BICN architecture. It terminates SIP
sessions and provides ISUP (or equivalent) signaling to the PSTN. The IMS-Media Gateway
(IMS-MGW) provides inter-working and transcoding from IP bearer to TDM (or potentially VoIP
or VoATM).

6. CONCLUSIONS
Operators with mature wireless networks that have grown organically over several years will be
familiar with the rising network costs at a time of limited revenue growth. The question of how to
reverse this trend, or even simply contain it, is the focus of many operators today. However, this
question is also relevant to operators of less mature networks and for those operators building
Greenfield networks. Operators need to learn the lessons gained from the first generation of
mature wireless networks.
A major component to the rising costs from mature networks has been identified as a function of
network growth. Network growth has been achieved through the proliferation of multiple
network nodes, versus scaling up of a smaller number of nodes. The proliferation of network
nodes creates significant network overheads that enable network nodes to communicate with
other networks nodes. This problem increases exponentially with the number of nodes in the
network. Communication between nodes is necessary to maintain important functions such as
location updates and active mobile handovers.
The large number of network layers further adds a dimension to this problem. The emerging
wireless service proposition is no longer simply based on TDM telephony, but is a diverse
package of services of which many will be based on packet technology. Implementing different
network layers for each service type, network command and control means that the wireless
operator is no longer operating a single network, but rather several networks at once. The
consolidation of these multiple overlays into a single multi-service network is a key goal of the
Wireless Next Generation Core Network.
Nodal scalability is also addressed through Wireless Next Generation Core Networks with the
complete separation of the bearer and control plane, enabling each function to be scaled
independently of the other and be geographically independent of the other. Containing the
optimum number of network elements needed for geographical coverage and network resiliency
ensures that the network cost model is optimized for the operators own circumstances. The
scaling of each node in the network can then be implemented as growth requires it. Nodal
scalability is achieved through the application of packet network techniques specifically to avoid
the limitations of traditional TDM technology.
3G Americas believes that the application of Next Generation technology can bring a diverse
range of benefits to operators of any type. GSM operators have the advantage by having 3GPP
specifying the IP multimedia subsystem (IMS), starting with release 5. Long term benefits can
be achieved through the simplification and consolidation of a network into smaller numbers of
scalable network elements. This means that a mature network can be consolidated and
simplified to contain escalating network costs. In the same way, less mature networks can avoid
the pitfalls of network growth that were prevalent in first generation wireless networks.
IMS is a powerful network strategy providing GSM operators a new layer in the packet domain
that will provide person-to-person, real and non-real time integrated multimedia services such
as voice, video, instant messaging, online gaming, web-sharing, and other services over a
single IP network.
16

ACKNOWLEDGMENTS
The mission of 3G Americas is to promote and facilitate the seamless deployment of GSM, GPRS, EDGE,
and UMTS throughout the Americas for the benefit of consumers. 3G Americas' Board of Governor
members include AT&T Wireless (USA), Cable & Wireless (West Indies), Cingular Wireless (USA),
Ericsson, Gemplus, HP, Lucent Technologies, Motorola, Nokia, Nortel Networks, Openwave Systems,
Research In Motion, Rogers Wireless (Canada), Siemens, T-Mobile USA, Telcel (Mexico), and Texas
Instruments.

We would like to recognize the significant project leadership and important contributions of Harshad Patel
of Nortel Networks to this white paper.