Applicat ion Not e

Number 17 Author: Dominik Schneuwly Created: August, 2004

The Synchronization of 3G UMTS Networks





Oscilloquartz S.A., CH-2002 Neuchtel 2, Switzerland,
Tel. +41 32 722 5555, Fax +41 32 722 5556, e-mail: [email protected]
The Synchronization of 3G UMTS Networks 2
ABSTRACT
This Application Note presents a discussion of
the synchronization issues in UMTS networks,
and provides practical solutions for several UMTS
network configurations. The paper addresses all
variants of UMTS, i.e. UMTS-FDD and UMTS-
TDD. In UMTS networks all network elements
need some form of synchronization. The required
synchronization accuracy and stability depend on
the network element type. Most network elements
must be synchronized in frequency. Base stations
in the TDD variant of UMTS also require an
external phase-alignment signal. The way to solve
the synchronization question in UMTS networks
depends very much on the transport network used
to interconnect all the UMTS network elements.
This Application Note details a number of practical
solutions applicable to different transport network
configurations. These solutions can be used for
developing a synchronization design for any given
real UMTS network.
ABBREVIATIONS
3G 3
rd
generation mobile system
3GPP 3
rd
Generation Partnership Project
ADM Add-Drop Multiplexer
AMPS Advances Mobile Phone System
ATM Asynchronous Transfer Mode
BITS Building Integrated Timing Supply
CDMA Code Division Multiple Access
CN Core Network
FDD Frequency Division Duplexing
GERAN GSM/EDGE Radio Access Network
GGSN Gateway GPRS Support Node
GMSC Gateway MSC
GPRS General Packet Radio Service
GSM Global System for Mobile Communications
IMT-2000 International Mobile Telecommunications 2000
IP Internet Protocol
ITU International Telecommunications Union
Mcps Mega chip per second
MGW Media Gateway
MS Mobile Station
MSC Mobile Switching Center
PRS Primary Reference Source
PRC Primary Reference Clock
QoE Quality of Experience
QoS Quality of Service
SGSN Serving GPRS Support Node
SSU Synchronization Supply Unit
TDD Time Division Duplexing
TD-CDMA Time Division CDMA
TDM Time Division Multiplexing
TD-SCDMA Time Division Synchronous CDMA
TSG Timing Supply Generator
UE User Equipment
UMTS Universal Mobile Telecommunications System
UTRAN UMTS Terrestrial Radio Access Network
W-CDMA Wideband CDMA
1. WHAT IS UMTS?
UMTS stands for Universal Mobile
Telecommunications System and designates a
family of mobile systems of the 3
rd
generation.
There is a number of so-called 3rd generation or
3G system standards. Strictly speaking the term
3G applies to all mobile system standards which
are officially backed by the ITU under the IMT-
2000 program (International Mobile
Telecommunications 2000). Under this program,
the ITU adopted a limited number of mobile
system standards proposed by standardization
bodies from all over the world. The UMTS
standards were initially proposed by the European
Telecommunications Standards Institute (ETSI),
but now the standardization activity around UMTS
is headed by a new standardization body called
the 3
rd
Generation Partnership Project or 3GPP
(www.3gpp.org).

Currently there are three different UMTS
standards for terrestrial mobile communications
(UMTS also encompasses satellite systems for
personal communications, but these are outside
the scope of this Applications Note). Table 1 gives
a brief overview of the main characteristics of the
three terrestrial UMTS systems
1
. The three
standards differ in the technical solutions adopted
for the radio interface between the base station
and the user equipment. The differences concern
the way up- and down-link communications are
separated (duplexing), and the particular CDMA
coding techniques used for separating multiple
users (multiplexing). But more importantly it is the
targeted application environments that are
different. UMTS-FDD and UMTS-TDD 1.28 Mcps
are for general environments, whereas UMTS-
TDD 3.84 Mcps was specifically designed for
indoor pico-cells and outdoor micro-cells in
densely populated areas. UMTS-TDD 1.28 Mcps
was introduced only very recently as a result of a
Chinese initiative.

3G mobile systems in general are being
deployed in order to match growing customer
demands. 3G systems provide much higher data
rates than 2
nd
generation (2G) systems such as
GSM, AMPS, and cdmaOne. 3G systems also
provide single platforms for both voice-centric and
data-centric services (in this document the term
voice-centric is used to designate real-time
services such as voice and also video, as
opposed to non-real-time data-centric services).
The latest enhancements of some 2G systems
(e.g. with GSM/GPRS) already provide such

1
In fact there are three Radio Access Network types called
UTRAN-FDD, UTRAN-TDD 3.84 Mcps and UTRAN TDD 1.28
Mcps, and one single UMTS Core Network type.
Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 3
unified platforms. But with 3G systems the
combination of voice- and data-centric services
comes with improved data rates: 144 to 384 kbit/s
for vehicular outdoor environments, and up to 2
Mbit/s for stationary indoor and city outdoor
environments. Within the family of 3G systems,
UMTS is the one that is closest to the 2
nd

generation GSM system. As a matter of fact,
UMTS was developed with smooth GSM-to-UMTS
migration in mind. During the migration UMTS
radio access networks and some later evolutions
of GSM radio access networks (e.g. GERAN) will
coexist and connect to a single UMTS Core
Network, thus providing similar services for both
GSM and UMTS handsets (though with very
different data rates).


Table 1: Terrestrial UMTS systems
UMTS system Intended Application
Multiple Access
1
Duplexing
2
UMTS-FDD All environments, especially
areas with medium population
density; suited for symmetric
traffic
W-CDMA
3
(3.84 Mcps chip rate
8
,
2 x 5 MHz channel bandwidth
9
)
FDD
6
UMTS-TDD 3.84 Mcps Indoor pico-cells and outdoor
micro-cells in densely
populated urban areas; suited
for both symmetric and
asymmetric traffic
TD-CDMA
4
(3.84 Mcps chip rate,
1 x 5 MHz channel bandwidth)
TDD
7
UMTS-TDD 1.28 Mcps All environments (expected to
be deployed in China first)
TD-SCDMA
5
(1.28 Mcps chip rate,
1 x 1.6 MHz channel bandwidth)
TDD
Notes:
1: the way multiple users share the available radio spectrum within a cell
2: the way a user shares the available radio spectrum for both communications directions
3: Wideband CDMA, actually a form of Direct Sequence CDMA
4: Time Division CDMA
5: Time Division Synchronous CDMA
6: Frequency Division Duplexing
7: Time Division Duplexing
8: CDMA code moments (chips) per second
9: radio spectrum bandwidth per user channel



2. UMTS NETWORK ARCHITECTURE
In order to understand the synchronization
issues of UMTS networks, it is necessary to have
a look at the general architecture of a UMTS
network. Figure 1 shows a rather detailed view of
the overall network architecture. The figure
depicts the constitutive functional entities and the
interfaces between them. UMTS entities are
grouped into three domains, the Core Network
(CN) domain, the UMTS Terrestrial Radio Access
Network (UTRAN) domain, and the User
Equipment (UE) domain. The Core Network is
itself divided into the Serving Network and the
Home Network. The Serving Network serves
that UTRAN where the calling user resides at the
moment of the call or session. The Home Network
on the other hand is the network to which the user
is linked by his subscription contract. From a
functional point of view, the Home Network
contains a database hosting the subscription
information; this database is called the Home
Subscriber Server or HSS (formerly Home
Location Register or HLR).

The Service Network is subdivided into a Circuit
Switched (CS) domain and a Packet Switched
(PS) domain. This partitioning reflects the fact that
UMTS mobile networks are designed for the
delivery of both voice-centric and data-centric
services. The two types of services are supported
in the Core Network by switching nodes adapted
to the circuit or packet switched nature of the
traffic.

In the PS domain the switching nodes are called
GPRS Support Nodes. A GPRS Support Node is
nothing else than a packet switch or packet router
designed for the specific packet technology used
in UMTS and GSM systems (the acronym GPRS
stands for General Packet Radio Service). There
are two types of GPRS Support Nodes. The
Serving GPRS Support Node or SGSN connects
to another mobile network or to another part of the
Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 4
same mobile network (case where the called party
is also a mobile user). The Gateway GPRS
Support Node or GGSN is used in cases where
the called party is to be reached over a fixed
public data network such as the Internet or a
public X.25 network.

The situation is similar in the CS domain of the
Core Network. There are again two types of
switching nodes called Mobile Switching Center
(MSC) and Gateway Mobile Switching Center
(GMSC). The term MSC can be misleading, since
one could be led to believe, that this MSC is a
TDM switch (TDM: Time Division Multiplexing),
just like the MSC of a GSM network. This is not
the case. MSC and GMSC are actually hybrids
between TDM switches and packet routers (such
a hybrid is usually called a Media Gateway). This
is so because the MSCs and the GMSCs
interfaces towards the transit network (PSTN
interfaces in Figure 1) convey TDM signals (e.g.
E1) with a 125 s frame structure, whereas the
(G)MSCs interfaces towards the UTRAN (IuCS
interface in Figure 1) convey packets or cells. This
is entirely different from the GSM case, where
everything is based on TDM.

Finally there is the VLR (Visitor Location
Register). It is a database serving both the CS
and the PS domain; it keeps track of information
concerning a visiting user (roaming).




Figure 1: Architecture: entities and interfaces


In UMTS, the entire UTRAN is a packet or cell
switched network. The actual structure of the
UTRAN resembles the structure of the GSM
Radio Access Network. Base stations called
Node B in UMTS parlance are the transceivers
serving a radio cell. A number of Nodes B is
controlled by a so-called Radio Network Controller
or RNC (similar to the Base Station Controller or
BSC in GSM networks). The RNC handles control
functions such as the allocations of radio channels
to a requested call, execution of handovers and
many more. Both voice-centric and data-centric
traffic traverses the UTRAN either as IP packets
or as ATM cells. IP and ATM are the two
fundamental variants, which equipment
manufacturers can choose from. It is generally
expected that the first generation UTRAN
equipment will be predominantly based on ATM.
Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 5
Many operators seem to have higher confidence
in ATM when it comes to guaranteeing real-time
Quality of Service (QoS) parameters such as
packet/cell delay and packet/cell delay variation
(jitter). Being able to maintain these parameters
within acceptable limits is crucial for the Quality of
Service and the subjective Quality of Experience
(QoE) of real-time services such as voice or video
conferencing.

In the protocol stacks admitted by the UMTS
standards, ATM or IP can run over a variety of
Layer 1 or L1 protocols (physical layer). The
standards allows equipment manufacturers to
choose from a rather long list of possible L1
protocols or signals. Table 2 shows the complete
list. It is taken from the 3GPP Technical
Specification TS 25.411. Although the list
basically applies to all interface types, the higher
data rate signals are likely to be found in the Core
Network and on the IuCS and IuPS interfaces
(between Core Network nodes and RNC), while
the lower data rate signals are for the Iub
interfaces (between RNC and Node B). In the
current release of the UMTS standard (Release
6), all permitted L1 signals are synchronous
signals. This has important consequences for the
synchronization of UMTS networks, as will be
discussed in sections 4 and 5.
3. SYNCHRONIZATION ISSUES
In UMTS networks, just as in most other network
types, interconnected network elements interwork
correctly only if the signals they exchange comply
with given synchronization quality requirements.
The synchronization quality of the signals is
determined by the network elements equipment
clocks. Equipment clocks control some of the
important signal processing functions within the
network element (e.g. multiplexing, switching), as
well as the data rate (accuracy, stability) of
outgoing traffic signals. There are different clock
frequency accuracy and stability requirements for
different network element types. In most cases
frequency accuracy specifications are much
tighter than what the equipment clock could
deliver if it were free running. This means that
most equipment clocks need to be synchronized
or locked to an external and possibly remote
high accuracy frequency reference clock.
Furthermore there are cases where accurate
frequency is not enough, where all equipment
clocks of a given type are required to tick in phase
throughout the network. The issue of clock
synchronization is how to distribute a common
frequency reference and possibly a common
phase reference of appropriate quality to all clocks
in the network.

Network operators must design an appropriate
synchronization network which guarantees that all
equipment clocks operate within the relevant
frequency (and possibly phase) accuracy and
stability specifications. Before discussing how
synchronization (frequency and possibly phase) is
distributed to the equipment clocks, it is necessary
to understand the synchronization performance
requirements. This is the subject of the next
section.
4. SYNCHRONIZATION
PERFORMANCE REQUIREMENTS
4.1. OVERVIEW
The method to be used for distributing
synchronization obviously depends on the
targeted synchronization performance. The
performance requirements differ from one network
element type to another. The traditional way of
specifying synchronization performance in a
network is to state frequency accuracy and jitter &
wander limits for certain types of interfaces. Table
3 summarizes the synchronization requirements
for the important interfaces of a UMTS network
(see Figure 1 for the location of the interfaces).
The way things are specified in the standards pre-
suppose or assume that synchronization
distribution flows from the Core Network to the
RNC, from the RNC to the Node B, and finally
from the Node B to the Mobile Station (see Table
3, column 2). The last column of the table
indicates the 3GPP Technical Specification where
requirements are taken from.
4.2. REQUIREMENTS FOR CN
NODES AND RNC
The important network elements of the Core
Network, i.e. MSC, GMSC, SGSN and GGSN, are
called Core Network nodes or simply CN nodes.
As mentioned earlier, the protocol stacks running
over the interfaces between CN nodes and RNC
are either cell- or packet-switched protocols.
Switching and forwarding cells and packets are
asynchronous operations. One could therefore be
led to believe, that CN nodes and RNC would not
require any synchronization. However this is not
the case at all. In cell- and packet-switched
networks, synchronization requirements are
driven by the needs of the Layer 1 (L1) protocols.
As mentioned earlier, there is a variety of possible
L1 protocols (see Table 2). In the current release
of the UMTS standard (Release 6), all permitted
L1 signals are required to be synchronous signals.
Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 6

Table 2: Layer 1 signal types
Designation Data rate [Mbit/s] Electrical specification
and frame structure
Network Limit for
wander
Network Limit for
jitter
STM-4 622 I.432.2
STM-1 155 I.432.2
STM-0 51 ETSI/TTC
STS-3c 622 T1.105
STS-12c 155 T1.105
G.825, 5.2
or
G.823, 6.2.3
G.825, 5.1
E3 34 G.751
1
E2 8 G.703/704
2
E1 2 G.703/704
G.823, 6.2.4 G.823, 6.1,
Table 1, PDH
synchronization
interface
T3 45 G.703/704
T1 1.5 G.703/704
J 2 6.3 J T-G.703/704
J 1 1.5 J T-G.703/704
G.824, 6.2.2,
Tables 5 and 6
G.824, 6.1
Notes:
1: G.751 specifies the mapping of four E2 into one E3; this means that it is not ATM cell to E3 mapping according to
G.804 (otherwise TS 25.411 would state G.832 which specifies the E3 frame structure for G.804 mapping).
2: The mapping of ATM cells into E2 is not defined yet in G.804 (where it is expected to be specified).



Table 3: Synchronization performance requirements
Interface Direction of synchronization
distribution
Requirements References
IuCS, IuPS CN node
to
RNC
y <1E-11
N.L.: G.823/824/825
4
TS 25.411, 4
Iub RNC
to
Node B
y <1E-11
N.L.: G.823/824/825
4
TS 25.431, 4
Node B
Synch. Input Port
2, 3
External source
To
Node B
y <5E-8
x <1.25 s
2, 3

TS 25.402, 6.1.2
2, 3
Uu Node B
to
MS
y <5E-8
x <1.5 s
2, 3
TS 25.104, 6.3
1
TS 25.105, 6.3
2, 3
TR 25.836, 6
2
TR 25.868, 6
3
Notes:
x: phase-time
y: fractional frequency
N.L.: Network Limit
1: FDD
2: 3.84 Mcps TDD
3: 1.28 Mcps TDD
4: depending on layer 1 signal type, see Table 2

Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 7



Figure 2: MTIE Network Limits for Iub, IuCs and IuPS interfaces




1E-9
1E-8
1E-7
1E-6
1E-5
1.00E-02 1.00E-01 1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04
Observation Interval [s]
Network Limit for STM-n and STS-n signals
Network Limit for E-n signals
Network Limit for T-n and J -n signals


Figure 3: TDEV Network Limits for Iub, IuCs and IuPS interfaces

1E-8
1E-7
1E-6
1E-5
1E-4
Network Limit for STM-n and STS-n signals
Network limit for E-n signals
1.00E-01 1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 1.00E+05
Observation Interval [s]
Network Limits for T-n and J -n signals
Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 8
For all these signal types the required accuracy of
the data rate is

y <1E-11 ,

where y is the fractional (or relative) frequency
deviation. This accuracy figure is the same as the
one specified in ITU-T Rec. G.811 for a Primary
Reference Clock
1
(PRC). This means that CN
nodes and RNC must be synchronized directly or
indirectly to a PRC, their equipment clocks must
be traceable to a PRC. The jitter and wander
present on the IuCS, IuPS and Iub interfaces are
required to be within the so-called Network Limits
specified in ITU-T Rec. G.823, G.824 and G.825.
The last two columns of Table 2 give the exact
bibliographic references for these important
Network Limits. G.823 and G.824 contain several
Network Limits specifications for different network
configuration cases. Because IuCS, IuPS and Iub
interfaces are used to carry synchronization from
CN node to RNC and from RNC to Node B
respectively
2
, it is the so-called Network Limits for
Synchronization Interfaces which apply. Figures
2 and 3 show the Network Limits for wander
expressed as MTIE and TDEV thresholds. The
curves shown in the two figures are taken from
the ITU-T Recommendations mentioned in Table
2.
4.3. REQUIREMENTS FOR NODE B
Because synchronization flows from the Node B
to the Mobile Station (user equipment), the
standards specify synchronization performance at
the Node Bs radio interface Uu (the transmitting
antenna). For UMTS-FDD there is only a
frequency accuracy specification:

y <5E-8 .

The specification basically means that the Node
Bs transmitter must deliver a radio signal whose
carrier frequency and carried data rate exhibits
this accuracy. Visibly this specification is less
stringent than the one for the CN nodes and the
RNC. Normally the Node B in UTRAN-FDD
derives frequency synchronization from a traffic
signal traversing the Iub interface. For that
purpose, the traffic signal carrying synchronization
to the Node B must comply with the Network
Limits indicated in Table 2.

The 5E-8 frequency accuracy is needed for a
number of reasons. First, accurate
synchronization of all Nodes B is required for
successful handover processing. Degraded
synchronization has a direct and measurable
impact on the number of user calls or sessions
lost during the handover. Secondly, accurate radio
carrier frequencies at the transmitter outputs
(antennas) guarantee that radio channels of
neighboring cells do not overlap in the spectrum
domain. Spectral channel overlap causes channel
cross-interference, which in turn causes a
noticeable and at times very annoying
degradation of voice quality (signal-to-noise ratio).
Thirdly, accurate synchronization is a pre-requisite
for all the frame synchronization processes that
take place over the various interfaces. Signal
frame structures are aligned by special protocol
procedures 1) between CN nodes and RNC, 2)
between RNC and Nodes B, and 3) between
Node B and Mobile Station. These frame
synchronization processes optimize handover
times, user signal delay and user signal jitter.
Although frame synchronization is achieved by
software-driven protocol procedures, they function
only if the accuracy of the physical
synchronization is within the mentioned 5E-8
frequency accuracy specification.

1
called Primary Reference Source (PRS) in North America
2
See 3GPP Technical Specification TS 25.411, 4.2.1

In the two UMTS-TDD systems, the Node B and
the Mobile Station not only need frequency
synchronization, but also phase synchronization.
In other words: the equipment clocks of all Nodes
B and all Mobile Stations are required to deliver
clocking signals which are phase-aligned against
each other. There is a phase accuracy
specification for the Node Bs radio interface. The
specification is the same for both UMTS-TDD
systems (3.84 Mcps and 1.28 Mcps):

x <1.5 s .

Phase synchronization is needed in order to
assure that frames transmitted by neighboring
cells are aligned in time. This is called Inter Node
B Synchronization. With the TD-CDMA and TD-
SCDMA techniques used in UMTS-TDD,
degradation or loss of Inter Node B
Synchronization results in high cross-interference
between use channels.

In the UTRAN-TDD, it is not possible for the
Node B to get a phase reference from the distant
RNC. Instead the Node B must be supplied with
an external phase synchronization signal over its
so-called Synchronization Input Port. The
electrical characteristics of the port (RS-422) and
the format of the reference signal are specified in
TS 25.402. The signal is essentially a 100 Hz
pulse train (0.5 to 1 ms pulse-width) with
additional frame markers every 2.56 s (2 to 3 ms
pulse-width) and every 40.96 s (4 to 5 ms pulse-
Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 9
width). The three superposed periodic signals are
used to align the frames, the multi-frames and
the SFN periods of the radio signals (SFN
stands for Single Frequency Network). The phase
reference signal must have an accuracy of

x <1.25 s .

The only practical way to supply such a signal is
to generate it locally with a GPS-receiver. GPS-
receivers are able to deliver frequency, phase and
even time signals with very high accuracy. These
signals are indirectly (i.e. via the GPS satellites)
derived from the ground master clock of the GPS
system.
5. SYNCHRONIZATION SOLUTIONS
5.1. REFERENCE SOLUTIONS
Said simply, the task at hand is to distribute

frequency with an accuracy of 1E-11 to all
CN nodes (MSC, GMSC, SGSN, GGSN) and
all RNC,
frequency with an accuracy of 5E-8 to all
Nodes B, and
phase with an accuracy of 1.25 s to all
Nodes B (in the UMTS-TDD case),

and make sure that the distributed
synchronization signals also comply with the
relevant Network Limits for jitter & wander.

The way this is achieved depends essentially on
the type of transport network used for
interconnecting the UMTS network elements. This
section presents six reference solutions, three for
UMTS-FDD and three for UMTS-TDD. For each of
the two cases there are three solutions
corresponding to three different transport network
configurations:

1) Trusted SDH or SONET transport network
2) Trusted PDH transport network
3) Leased lines

For the clock synchronization issue it makes a
difference, whether the transport network is itself
synchronous or not, and whether the transport
network is able to transport synchronization
signals transparently or not. The three transport
network scenarios listed above reflect these
differences. In the first scenario the transport
network (SDH or SONET) is itself synchronous.
This means that all transport network elements
are synchronized to a PRC, and synchronization
derived from the transport network elements can
be used to synchronize other equipment. In the
second scenario, the transport network (PDH) is
transparent for the timing of the transported traffic
signals (tributaries). This means that the
transported traffic signals can be used to carry
synchronization from one place to another. The
case where the transport network elements are
not synchronous and the transport network is not
transparent for timing defines the third scenario.

The synchronous nature of SDH/SONET
networks and the timing transparency of the PDH
networks can only be exploited for
synchronization, if these properties can be relied
upon. The synchronization delivered by an
SDH/SONET network element can be used only if
the required synchronization quality is there with a
high availability. A PDH tributary can be used for
the transport of synchronization only if the timing
transparency does not get degraded by failure
conditions or other disruptive events that are not
under the control of the UMTS operator. This kind
of trust pre-supposes either that the UMTS
operator controls the transport network, or that
there is a contract between UMTS operator and
transport service provider concerning
synchronization. With traditional leased lines this
is precisely not the case. Here the transport
network is owned by the leased line provider, and
the standard leased line service definitions do not
describe synchronization aspects in an
appropriate manner. So leased lines provide
neither trustworthy synchronization nor
trustworthy timing transparency.
5.1.1. UMTS-FDD, TRUSTED
SDH/SONET TRANSPORT NETWORK
Refer to Figures 4 and 5. SDH and SONET
networks are synchronous networks. They
possess their own synchronization distribution
system which guarantees that all SDH/SONET
network elements are synchronized to a PRC
under normal operating conditions (no failures). A
well-designed SDH/SONET synchronization
distribution network also guarantees that the
synchronization performance complies with the
relevant Network Limits, again under normal
operating conditions. Provided that the
SDH/SONET transport networks own
synchronization distribution can be trusted,
synchronization reference signals for the UMTS
network elements are simply taken from co-
located SDH/SONET network elements. This is
always possible, since all SDH/SONET network
elements have so-called External Timing Output
ports delivering a non-traffic-carrying
synchronization signal (usually 2.048 MHz, 2.048
Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 10
Mbit/s or 1.544 Mbit/s). All it takes is to connect
the SDH/SONET network elements External
Timing Output to the UMTS network elements
External Timing Input. Since PRC-traceability can
get disrupted under certain failure conditions
occurring inside the SDH/SONET network, it is
highly recommended to add an SSU
1

(Synchronization Supply Unit) in each site
containing a CN node (MSC, GMSC, SGSN,
GGSN) or an RNC. In case the traceability to the
PRC gets lost because of some failure condition,
the SSU enters an operation mode called
holdover. In holdover mode the SSU becomes an
autonomous frequency generator. The frequency
accuracy in holdover mode, though slowly
degrading with time, is sufficient for operating the
connected UMTS equipment for a period of 1 day
to over a week, depending on the SSU model and
the temperature conditions. SSU deployment is
important for all CN nodes and all RNC, because
a disruption of synchronization supply at these
places would affect a large number of user calls
and sessions, and have very negative impact on
customer satisfaction.

Nodes B which are co-located with an SDH or
SONET network element take synchronization
directly from the SDH/SONET network element.
The situation is somewhat different when there is
no SDH or SONET network element co-located
with the Node B. In this case the traffic is carried
from the egress SDH/SONET Add-Drop-
Multiplexer (ADM) to the Node B over a lower
order tributary signal such as E1, E2, E3, T1, T3,
J 1 or J 2 (see Table 2). These signals are
generated by an RNC which is normally traceable
to a PRC (or to an SSU in holdover mode). The
lower order tributaries traverse the SDH/SONET
network from the RNC site to the egress ADM
closest to the Node B. Whether these tributaries
are suitable or not for carrying synchronization to
the Node B depends on the timing transparency of
the SDH/SONET transport network. The timing
transparency must be good enough to keep the
jitter & wander of the transported tributary within
the relevant Network Limits (refer to Figures 2 and
3). This is the case in most real situations. The
condition is that the SDH/SONET transport
network is well synchronized, so that pointer
adjustments do not occur too frequently. There
can be situations, however, where these
conditions are difficult to fulfill at all times. In
sparsely meshed SDH/SONET networks the
probability of synchronization failures can be too
high (availability issue). This may lead to
increased pointer activity, which in turn can
degrade the wander on dropped E1, T1 and J 1
signals well beyond the specified Network Limits.
In such cases the solution is to re-time the E1, T1
and J 1 tributaries in the site where they leave the
SDH/SONET network. As shown in Figure 5, the
Re-timing function imprints the synchronization
derived from the egress ADMs External Timing
Output port onto the dropped tributary. More
about this subject can be found in the
Oscilloquartz Application Note Re-timing: Cost-
effective Synchronization via Re-timed E1 and
DS1 Signals.

1
called TSG (Timing Supply Unit) or BITS Clock (Building
Integrated Timing Supply) in North America
5.1.2. UMTS-FDD, TRUSTED PDH
TRANSPORT NETWORK
Refer to Figure 6. The network elements of a
PDH network are not synchronous. The higher
order tributaries are not synchronized to a PRC;
instead their data rate is derived from the free
running equipment clocks contained in the
network elements (PDH multiplexer). The
frequency accuracies of these oscillators are in
the range of 15E-6 to 50E-6 (fractional frequency).
This mode of operation is called plesiochronous
mode. Given these frequency accuracies, signals
generated by the PDH equipment clocks cannot
be used for synchronizing UMTS equipment. On
the other hand, PDH transport networks are
perfectly transparent for the timing of the
transported traffic signals (tributaries). This
property is exploited in order to build
synchronization distribution networks. In Figure 6
a CN node (e.g. an MSC) is synchronized by the
co-located PRC. Traffic signals generated by this
CN node are therefore synchronous. In the figure,
one of these traffic signals is used for carrying
synchronization from the CN node to the RNC
over the PDH network. In the RNC site,
synchronization is extracted from the traffic signal
and fed to an SSU. The SSUs output signal is
then connected to the RNCs External Timing
Input port. This principle is applied to all RNC and
also to all other CN nodes (those which are not
co-located with the PRC). The SSU plays the
same role as in the previous scenario: it provides
holdover protection in case traceability to the PRC
gets lost because of some failure condition. The
distribution of synchronization from the RNC to
the Node B is done in the same way (except that
there is no SSU associated with the Node B): a
synchronous traffic signal transports
synchronization and user data from the RNC to
the Nodes B over the PDH network. The
compelling simplicity of this solution is a
consequence of the nearly perfect timing
transparency of PDH networks.

Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 11
5.1.3. UMTS-FDD, LEASED LINES
Refer to Figure 7. This section presents a
reference solution for the case where the
transport network is not sufficiently transparent for
the timing of the transported traffic signals, and
where the transport network cannot deliver valid
synchronization signals derived from its own
synchronization distribution system. In this case
there is only one solution: each site containing a
UMTS network element must be equipped with a
GPS-receiver. As shown in Figure 7, the GPS-
receivers deliver synchronization signals which
are connected to the External Timing Input ports
of the UMTS network elements. Frequency
accuracy as well as jitter & wander levels of
synchronization-grade GPS-receivers comply
easily with PRC
1
standards; hence these GPS-
receivers are suitable for the synchronization of all
UMTS network elements.

A few comments on leased lines are in order at
this place. There are two categories of leased
lines in use: structured and unstructured leased
lines. Unstructured leased lines do not require the
transported client signal to comply with any
particular frame structure, as long as the data rate
is within the specified range (e.g. +50 ppm for
2048 kbit/s unstructured leased lines).
Unstructured leased lines are only semi-
transparent for the timing of the transported client
signal: they preserve the long-term average data
rate, but they inject jitter & wander. However,
service definitions usually do not include any
guaranteed limits on the injected jitter & wander.
Real world experience shows that the wander
levels at the leased line egress points can be very
high and often violate the Network Limits of
Figures 2 and 3. Structured leased lines, on the
other hand, require the transported client signals
to have a standard frame structure (e.g. according
to ITU-T Rec. G.704, etc.). Structured leased lines
come with several timing options. The common
ones are called User Timing, Network Timing
and Loop Timing. A full discussion of the
synchronization properties of these timing modes
is outside the scope of this document. It is
important, however, to summarize the main
points: 1) none of the three options provides
adequate timing transparency, and 2) none of the
three options can deliver good synchronization
signals derived from the transport network itself
(with the Network Timing option the data rates of
the client signals are in fact imposed by the
transport network timing, but service definitions
usually do not include any guaranteed limits on
frequency accuracy, jitter and wander). From the
above it follows that leased lines are neither good
synchronization carriers nor good synchronization
sources, hence they should never be trusted for
synchronization.

1
ITU-T Rec. G.811, ANSI T1.101, Telcordia GR-2830, ETSI
EN 300 462-2
5.1.4. UMTS-TDD SCENARIOS
Refer to Figures 8 to 10. With UMTS-TDD,
Nodes B require accurate phase synchronization.
The only practical way to make available a phase
reference is the GPS. In Figures 8 to 10 Nodes B
are equipped with synchronization-grade GPS-
receivers. They deliver a synchronization signal
acting both as a frequency reference and as a
phase reference. The 3GPP Technical
Specification TS 25.402 proposes a standard
signal format consisting of a 100 Hz pulse train
with special markers every 2.56 s and every 40.96
s. Due to the GPS, the markers of all Nodes B in
the network are phase-aligned with sub-
microsecond accuracy. The way the CN nodes
and the RNC are synchronized is exactly the
same as in the three UMTS-FDD scenarios.

In the UMTS-TDD cases based on PDH or
SDH/SONET transport, there is the possibility of
using a (synchronous) traffic signal going from the
RNC to the Node B as a 2
nd
priority frequency
reference. The idea is to have some protection
against GPS-receiver problems: if the output
signal of the GPS-receiver fails (e.g. because of a
hardware failure or bad radio reception), then the
Node B uses the traffic signal coming from the
RNC as a frequency reference. This back-up
solution pre-supposes that the Node Bs internal
clock is capable of maintaining phase alignment
while being steered by the frequency (data rate) of
the traffic signal; this is the case with all known
implementations. Provided that the 2
nd
priority
frequency reference is accurate and stable, the
accuracy of the phase-alignment in this operation
mode will be slightly degraded, but still good
enough for acceptable performance.
5.2. Real World Solutions
In the reference solutions presented here, one
and the same transport network type
(SDH/SONET or PDH or leased lines) is used in
the entire Core Network and for the
interconnections of all RNC. In reality there is
often a mix of several transport network types. In
such cases a synchronization distribution concept
must be worked out by combining the techniques
from the six reference solutions.


Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 12






Figure 4: UMTS-FDD, trusted SDH/SONET transport network, without re-timing











Figure 5: UMTS-FDD, trusted SDH/SONET transport network, with re-timing

Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 13





Figure 6: UMTS-FDD, trusted PDH transport network






Figure 7: UMTS-FDD, leased lines

Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 14


Figure 8: UMTS-TDD, trusted SDH/SONET transport network, without re-timing




Figure 9: UMTS-TDD, trusted PDH transport network

Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 15



Figure 10: UMTS-TDD, leased lines


BIBLIOGRAPHY
[1] St. Bregni; Synchronization of Digital
Telecommunications Networks; J ohn Wiley
& Sons, Chichester, England; 2002.
[2] D. Schneuwly; Application Note: Re-timing:
Cost-effective Synchronization via Re-timed
E1 and DS1 Signals; Oscilloquartz,
Neuchtel, Switzerland; J anuary 2004.
[3] 3
rd
Generation Partnership Project;
Technical Specification TS 25.104 V6.5.0:
Base Station (BS) radio transmission and
reception (FDD); Valbonne, France; March
2004.
[3] 3
rd
Generation Partnership Project;
Technical Specification TS 25.105 V6.0.0:
Base Station (BS) radio transmission and
reception (TDD); Valbonne, France;
December 2003.
[3] 3
rd
Generation Partnership Project;
Technical Specification TS 25.401 V6.2.0:
UTRAN overall description; Valbonne,
France; December 2003.
[4] 3
rd
Generation Partnership Project;
Technical Specification TS 25.402 V6.0.0:
Synchronisation in UTRAN Stage 2;
Valbonne, France; December 2003.
[5] 3
rd
Generation Partnership Project;
Technical Specification TS 25.411 V6.0.0:
UTRAN Iu interface layer 1; Valbonne,
France; December 2003.
[5] 3
rd
Generation Partnership Project;
Technical Specification TS 25.421 V6.0.0:
UTRAN Iur interface layer 1; Valbonne,
France; December 2003.
[5] 3
rd
Generation Partnership Project;
Technical Specification TS 25.431 V6.0.0:
UTRAN Iub interface layer 1; Valbonne,
France; December 2003.
[7] 3
rd
Generation Partnership Project;
Technical Specification TS 25.836 V4.1.0 :
NodeB Synchronisation for TDD; Valbonne,
France; March 2001.
[8] 3
rd
Generation Partnership Project;
Technical Specification TS 25.838 V4.1.0:
Node B Synchronisation for TDD (Iub/Iur
aspects; Valbonne, France; December 2001.
[9] 3
rd
Generation Partnership Project;
Technical Specification TS 25.868 V5.0.1:
Node B Synchronisation for 1.28 Mcps TDD;
Valbonne, France; March 2002.
[10] International Telecommunication Union; ITU-
T Recommendation G.803: Architecture of
transport networks based on the synchronous
Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]
The Synchronization of 3G UMTS Networks 16
[13] International Telecommunication Union; ITU-
T Recommendation G.824: The control of
jitter and wander within digital networks which
are based on the 1544 kbit/s hierarchy;
Geneva, Switzerland; March 2000.
digital hierarchy (SDH); Geneva,
Switzerland; March 2000.
[11] International Telecommunication Union;
ITU-T Recommendation G.811: Timing
characteristics of primary reference clocks;
Geneva, Switzerland; September 1997. [14] International Telecommunication Union; ITU-
T Recommendation G.825: The control of
jitter and wander within digital networks which
are based on the synchronous digital
hierarchy; Geneva, Switzerland; March
2000.
[12] International Telecommunication Union;
ITU-T Recommendation G.823: The
control of jitter and wander within digital
networks which are based on the 2048
kbit/s hierarchy; Geneva, Switzerland;
March 2000.

Oscilloquartz S.A. / CH-2002 Neuchtel 2 / Switzerland / Tel. +41 32 722.5555 / Fax +41 32 722.5556 / e-mail: [email protected]