E-BOOK

Synchronization
Distribution in 5G
Transport Networks

The world is moving to 5G, which offers a wide range of new services beyond the voice
and data combination that was the primary service offering in the first four generations
of mobile technology. This latest generation of mobile networks will expand service
offerings into highly reliable and low-latency services that will potentially revolutionize
many areas of industrialization and our day-to-day lives. In order to deliver the higher
performance that these new services will require, all aspects of the mobile network will
require modernization. This includes the DWDM-based mobile transport network that
underpins the end-to-end mobile network.

5G is driving discussion around advances in optical network architecture, such as the

move to front/mid/backhaul-based xHaul networks, network slicing, and multi-access
edge compute architectures. It is also driving a need to improve performance in many
areas of basic transport network performance, such as low latency and synchronization
performance. 5G synchronization is a complex topic with many moving parts that all need
to come together harmoniously across all aspects of the transport network to provide
the right quality synchronization to the cell tower without overengineering the network
and driving up cost. This e-book explains the challenges involved in delivering 5G-quality
synchronization and the toolbox required to create end-to-end synchronization strategies
to meet 5G performance demands now and in the future.
E-BOOK

Table of Contents
INTRODUCTION SECTION 2
The Importance of Synchronization in 5G Networks __ 3 Infinera’s Sync Distribution Solution for End-to-End
Synchronization Delivery ______________________ 17
SECTION 1 Synchronization in the IP Layer _________________ 18
Understanding Synchronization and Synchronization in Packet Optical Transport
Synchronization Distribution ____________________ 4 in Metro and Regional Networks ________________ 20
Synchronization Basics ________________________ 5 Packet Optical Transport with Layer 2 Ethernet/
Evolution of Synchronization Requirements ________ 6 eCPRI Switching ____________________________ 20
Synchronization Delivery Mechanisms ____________ 7 DWDM Transport ____________________________ 21
Frequency Synchronization Standards ____________ 8 Synchronization in DWDM Transport over Regional,
Long-haul, and Legacy Networks _______________ 23
Phase Synchronization Standards ________________ 8
End-to-End Sync Planning and Management ______ 25
ITU-T G.8271.1 Network Limits ___________________ 9
Summary _________________________________ 26
ITU-T G.8273.2 PTP T-BC Classes _______________ 12
Further Reading ____________________________ 27
ITU-T G.8275.1 Full On-path Support and G.8275.2
Partial On-path Support _______________________ 13
3GPP TS 38.104 Time Alignment Error ____________ 14
Pulling It All Together to Provide 5G-quality ________ 15
Synchronization
Getting Synchronization Right __________________ 16

Synchronization Distribution in 5G Transport Networks 2

E-BOOK

The Importance of Synchronization in 5G Networks

Network synchronization is a very specialized topic that has seen its relevance to network operators come and go over
time as technology trends have changed. In the era of synchronous TDM (SDH and SONET) networks, synchronization was
critical, but in recent times the availability of “good enough” synchronization for Ethernet-based transport has pushed the
topic to more of a niche in many network operators’ networks. The need for a step change in synchronization performance
in 5G networks is reversing this trend, bringing synchronization back into the top group of challenges that need to be
addressed within transport networks.

The new Phase 2 5G services, especially ultra-reliable low-latency communications (uRLLC) services, will drive significant
changes into overall mobile network architecture, as well as into the mobile transport network that connects the cell tower
to core processing resources. These architectural changes include lower latency through multi-access edge compute (MEC),
new network slicing capabilities, and better synchronization performance to support new 5G RAN functionality like carrier
aggregation (CA) and previous 4G/LTE-A functionality that is now being rolled out in 4G/5G networks, such as coordinated
multipoint (CoMP).

This e-book intends to give an overview of synchronization distribution and of Infinera’s approach to this challenging
environment. It is split into two major sections to enable readers to quickly navigate to the most relevant sections, or the
complete e-book can be read sequentially if preferred. The first section covers the background to network synchronization
in mobile networks, why synchronization is needed, and how it works. The second section outlines Infinera’s end-to-end
Sync Distribution Solution and the benefits that the breadth and enhanced performance capabilities of this synchronization
distribution solution are bringing to mobile operators across the globe as they build out 5G networks.

Synchronization Distribution in 5G Transport Networks 3

SECTION 1

Understanding
Synchronization
and Synchronization
Distribution
E-BOOK

Synchronization Basics another without the user or connected device noticing

There are many reasons that a network might require any drop or interference in the connection performance.
some form of synchronization within its capabilities, All cell sites require good synchronization back to a
either to support its underlying transport mechanisms centralized primary reference time clock (PRTC) so that
or because the end service or end devices require essentially all the RAN portions of the network are in
some form of timestamping or tight coordination with synchronization with each other, as shown in Figure 1.
other locations within the network. From an underlying
Depending on the synchronization capabilities that are
transport mechanism perspective, Synchronous Digital
required, there are three main components to network
Hierarchy (SDH) and Synchronous Optical Networking
synchronization that can be implemented in the network:
(SONET) are excellent examples of networking
protocols that rely on all the nodes in the network being ■ Frequency synchronization – ensuring the
synchronized so that specific timeslots in the data frequency of the local clock in the cell site is the
stream can be accessed without the need to completely same as that of the PRTC, so that the time interval
demultiplex the data stream at each node. Examples between timing pulses is exactly the same but the
of applications/devices that require synchronization timing signals do not necessarily occur at exactly
include financial trading applications that require the same time
accurate timestamping of financial transactions, electrical ■ Phase synchronization – ensuring that the timing
power distribution management where synchronization
pulses occur at exactly the same time
is needed for some power line failure management
systems, TDM circuit emulation applications, and of
■ Time of day – synchronization messages that have
course mobile networks. the ability to contain information and include the
exact time of the clock signal
In the simplest form, mobile networks require good
synchronization in the radio access network (RAN) to
ensure that devices can connect seamlessly to the cell
tower and to enable smooth handover from one cell to

Synchronization Distribution Network

GNSS-enabled Network Hardware

Global Navigation
Satellite System
(GNSS)

RU
RU

RU DU CU
Grandmaster
(T-GM) Clock Primary Reference
RU
Timing Clock (PRTC)
Fronthaul Network Midhaul Network Backhaul Network

Cell Street Cabinet Regional Office Core Data Center

Tower

Synchronization Distribution from PRTC to Cell Towers

Figure 1: Simplified synchronization network

Synchronization Distribution in 5G Transport Networks 5

E-BOOK

SYNC SIGNALS – FREQUENCY SYNC SYNC SIGNALS – PHASE SYNC SYNC SIGNALS – TIME SYNC

Time 8:24:45 8:24:46 8:24:47 8:24:48

interval

Clock A Clock A Clock A

8:24:45 8:24:46 8:24:47 8:24:48

Clock B Same time interval Clock B Same time interval and Clock B Pulses aligned and equal
pulses aligned timestamps on each sync signal

Figure 2: The three basic components of network synchronization

Evolution of Synchronization performance that was required in the FDD domain and
add phase and time synchronization, as outlined earlier
Requirements
in Figure 2.
2G, 3G, and initial releases of 4G all use frequency-
division duplex (FDD) as the underlying transmission Phase synchronization quality is measured by the time
within the RAN. FDD uses two separate frequencies difference between the timing signals and is typically
for upstream and downstream communication, and represented in microseconds (µs) or even nanoseconds
these networks require tight frequency synchronization (ns). Initial use of TDD in 4G LTE networks drove a
to ensure the correct frequencies are used and that requirement for phase synchronization accuracy of 1.5
these frequencies can be tightly packed to achieve µs. As will be discussed in the following sections, 5G has
efficient use of the available spectrum. Tight alignment tightened this requirement, especially for the relative
to the planned frequencies for a cell also ensures that difference between adjacent cell sites, where it can be
regulatory commitments are met in terms of spectrum as low as just 60 ns.
licenses and enables smooth handover of calls to
adjacent cells. FDD Operation

Frequency synchronization quality is represented by a Upstream Traffic

measurement of the difference in frequency between the Downstream Traffic

actual and desired frequencies and is represented as a
Time
figure showing the difference in parts per billion (ppb).
Mobile networks are specified to require 50 ppb at the TDD Operation
air interface of the RAN, and to ensure this requirement
is met, the backhaul interface from the 4G baseband unit
(BBU) back to the transport network is 16 ppb.
Time
As networks move to more advanced 4G Long-Term
Evolution (LTE) and 5G mobile networks with more Figure 3: FDD and TDD operation
complex functionality in the RAN, such as CoMP,
TDD operation should not be confused with dynamic
multiple-input multiple-output (MIMO) antennas, and
spectrum sharing (DSS), where a frequency band within
higher frequencies above 2 GHz, the underlying RAN
the spectrum is simultaneously shared between 4G and
mechanism has to migrate from FDD to time-division
5G radios. DSS operates with differing timeslots that are
duplex (TDD). TDD uses the same frequency for both
allocated in increments of 1 millisecond (ms). During a
upstream and downstream communications with
timeslot, the appropriate 4G or 5G RAN technology is
specific timeslots allocated to each, as shown in Figure
used for the defined period of time before stopping to
3. To enable the cell site (4G BBU or 5G distributed
allow the alternative RAN technology to take over for its
unit [DU]) and the devices connected to the RAN to
timeslot.
correctly transmit and receive at the right time, networks
need to maintain the same frequency synchronization
Synchronization Distribution in 5G Transport Networks 6
E-BOOK

Synchronization Delivery Mechanisms 5G in dense urban environments will require millimeter-

The delivery of synchronization information in mobile wave small cells that provide high-bandwidth
networks is achievable through several different connectivity over a shorter range, and operators are
mechanisms and strategies. The uptake of these various planning deployments of these in those tough-to-reach
options has varied across the geographic regions of the locations such as deep inside shopping malls, cells per
globe due to technical and geopolitical reasons. The floor in high-rise office buildings, etc.
main synchronization delivery options are:
It should be stressed that while GNSS networks do
■ Synchronization/timing signals from a global occasionally suffer from interference and downtime
navigation satellite system (GNSS), such as the caused by natural effects or deliberate jamming/
U.S.’s Global Positioning System (GPS), Europe’s spoofing, they are still highly reliable and form a key
Galileo, Russia’s Global’naya Navigatsionnaya component of most synchronization networks. There are
Sputnikovaya Sistema (GLONASS), or China’s solutions to protect GNSS and deliver GNSS signals into
BeiDou Navigation Satellite System, directly to tough locations, but overall, these factors are causing
every location requiring synchronization in the more and more operators that were previously GNSS-
network focused to plan to utilize network-based synchronization
as a backup to GNSS at every node. In some cases,
■ Synchronization/timing signals delivered from key
these operators plan to migrate fully to network-based
centralized GNSS-enabled locations in the network
synchronization, with GNSS limited to key centralized
through the backhaul/transport network to all other
locations in the network that use these protection and
locations requiring synchronization
resiliency methods to harden GNSS against attacks.
■ Synchronization/timing signals delivered through a
totally separate synchronization delivery network Network-based synchronization can take the form
of either synchronization delivery through the
Each approach has its own strengths and transport network or through a totally separate
weaknesses, and operators across the globe have dedicated synchronization delivery network. Both
built synchronization strategies to best suit their own approaches provide the operator with the right level of
environments. For example, historically GNSS using synchronization performance, and backhaul network-
GPS to every location has been the primary mechanism based synchronization offers the opportunity for
in North America, whereas Europe predominantly uses significantly better overall network economics as it
synchronization through the backhaul network with avoids a complete overlay network for synchronization.
GNSS limited to key timing locations. Wherever possible, mobile network operators typically
utilize backhaul-based synchronization delivery, but it
However, in recent years there has been an increase
should be noted that this is not always possible, and
in the incidence of both deliberate and inadvertent
therefore, synchronization overlay networks cannot be
hacking and jamming of GNSS as the use of cheap illegal
discounted from the discussion.
GNSS jammers has increased and as some countries
have even tested GNSS jamming and/or spoofing as Overall, there will always be a mix of strategies deployed
part of military strategies. Due to the importance of across the globe, but the trend is moving more and more
network synchronization, these factors are leading some toward network-based synchronization delivery, and due
countries to introduce legislation to force protection and to better economics, the option to transport this over the
reliability into synchronization networks. It is possible to backhaul network is nearly always the primary option.
protect GNSS receivers from some of this jamming, but
this greatly increases the cost per node. The rest of this e-book will primarily focus on network-
based synchronization delivery approaches and will
Another consideration that mobile network operators consider how best to deliver synchronization over mobile
must take into account as they move to 5G is the transport optical networks.
proliferation of cell sites, especially those in locations
that are tough to reach from a GNSS perspective.

Synchronization Distribution in 5G Transport Networks 7

E-BOOK

Frequency Synchronization Standards also include an upgraded local oscillator capable of

When network-based synchronization is used, there free running at a much tighter specification of 4.6 ppm
is a requirement to deliver the synchronization clock to improve performance. However, in normal operation,
frequency from the centralized PRTC clock to the BBU a SyncE device will operate in locked mode, where
with less than 16 ppb, as outlined previously. When this the distributed synchronization is locked to one of the
is delivered through modern Ethernet-based networks, incoming frequencies. In this mode, accuracy improves
multiple standards exist to ensure the required level of again to ±1 ppb, so around 100,000 times better than the
performance is achieved: original Ethernet clock. Furthermore, if a SyncE device
loses incoming synchronization and goes into holdover
■ Synchronous Ethernet (SyncE) – A collection mode, then the SyncE specifications only allow for the
of specifications (ITU-T: G.8261, G.8262, and clock to drift by 0.01 ppm per day, meaning this is still
G.8264) together specify the architecture, 10,000 times better than a standard Ethernet clock after
wander performance, clock performance, and one day. eEEC tightens the clock frequency specification
synchronization signaling messages of an Ethernet- in areas such as jitter and wander performance and
based network synchronization method that is ensures the device is never in free running mode with
comparable to synchronization of SDH and SONET lower clock stability.
networks.
Generally speaking, all IP routers and Layer 2 Ethernet
■ G.8262.1 enhanced Ethernet equipment clock
switches in the network will need SyncE/eEEC
(eEEC) – Within SyncE, the G.8262 specification
compliance and any Layer 1 DWDM transponders or
covered clock performance, and this has been
muxponders must be transparent to synchronization, i.e.,
enhanced to support more demanding 5G
they must ensure that any synchronization clocks on a
specifications in G.8262.1. Often 5G-quality SyncE is
client input are replicated exactly on the corresponding
referred to by either G.8262.1 or eEEC.
far-end output. Typically, network-level testing is also
Standard Ethernet does not have any built-in network- required to ensure a particular network provides the right
level synchronization capability or synchronization level of frequency synchronization performance.
signaling mechanism. Ethernet devices include a
free-running oscillator for local use with a specified
Phase Synchronization Standards
Standardization becomes much more complex as
clock accuracy of within ±100 parts per million (ppm).
networks move into phase synchronization to support
Synchronization is used on each link on a hop-by-hop
4G LTE and 5G. As previously mentioned, phase
basis, but it is never passed from one link to the next to
synchronization requires delivery within time budgets
create a broader network-level synchronization chain.
for both absolute time error between the PRTC and
The additional SyncE and eEEC frequency the cell tower and relative time error between adjacent
synchronization specifications use the physical towers. The main phase synchronization standard is the
layer transmission of Ethernet frames to deliver IEEE’s 1588-2008 and 1588-2019, commonly known as
synchronization, add the necessary mechanisms to 1588v2 and 1588v2.1, specifications for a Precision Time
transfer an input frequency to output ports, and add the Protocol (PTP) for packet-based networks in a range
necessary synchronization status messages to ensure of applications, including mobile networks within the
traceability of the synchronization source. Overall, this “Telecoms Profile” definitions of the spec.
creates a network-wide SyncE/eEEC synchronization
1588v2 provides a mechanism for using specific timing
domain. To ensure that the network can deliver SyncE-
packets to deliver frequency information, and by adding
compliant frequency synchronization, any device in the
very accurate time stamping to the packets, it can also
optical network path between the clock source and
deliver time of day and accurate phase information.
destination must support SyncE/eEEC compliance or be
1588v2 is designed to interwork with existing frequency
totally transparent to synchronization. In addition to the
synchronization mechanisms where they exist, such as
mechanisms to deliver synchronization, SyncE devices
SyncE or eEEC.

Synchronization Distribution in 5G Transport Networks 8

E-BOOK

In theory it is possible to build networks with only or routing functions and correcting the timestamp
1588v2, using the G.8265.1 PTP frequency specification, in outgoing PTP packets to compensate for this
but the vast majority of modern 1588v2-capable internal delay. This gives the effect of reducing the
equipment also supports SyncE or eEEC, and most impact of the node on the PTP stream at a lower
networks today will use a “SyncE assist” mode, which cost than T-BC capabilities. However, it should be
will improve PTP performance to varying levels. IEEE noted that T-TC performance also is significantly
1588v2 contains a range of standard definitions of lower than that achieved by T-BC devices with a
differing classes of devices with varying capabilities lower level of compensation, issues with longer
and performance levels that together can build a chains of nodes, and a more restricted range of
synchronization delivery network. The most common of network architectures than T-BC-enabled networks.
these are: ■ Time slave clock (TSC), also called a telecom time
■ Grandmaster (GM), also called a telecom slave clock (T-TSC) in ITU-T specifications – The
grandmaster (T-GM) in ITU-T specifications – A GM end device that receives the clock information,
clock is typically located in the core of the mobile typically a BBU in a 4G LTE network or a DU or RU
network. In the core, the T-GM is typically the PRTC, in a 5G network. Also called a clock client.
but in synchronization networks built over different To enable 1588v2 to be deployed in telecom networks,
synchronization domains, the T-GM is the master the ITU-T has defined a range of specifications that
clock at the start of a PTP domain. ensure that the mechanism defined in 1588v2 can meet
■ Boundary clock (BC), also called a telecom the demanding requirements, especially those of TDD-
boundary clock (T-BC) in ITU-T specifications – A based mobile networks. These specifications outline
device with a built-in PTP clock client and PTP the available time error budget, or in other words, the
master interconnected with a local clock. This maximum allowed phase error in µs or ns, and how this
enables a network node (typically a router or budget is allocated across the network elements and the
Ethernet switch) to synchronize the local clock to performance specifications of specific devices. All these
the upstream T-GM/T-BC and act as a master to any specifications are important, and the most significant of
downstream client clock. Many modern Ethernet them are as follows:
switching devices now contain T-BC functionality,
whereas earlier implementations often had T-BC ITU-T G.8271.1 Network Limits
capabilities via an external “sync box” that added As previously mentioned, TDD networks, either 4G LTE
this capability to the node. or 5G, require 1.5 µs maximum time error at the cell site
to ensure compliant operation. The maximum absolute
■ Transparent clock (TC), also called a telecom
time error (Max|TE|) is subdivided into smaller error
transparent clock (T-TC) in ITU-T specifications – A
budgets for differing segments of the network, as shown
device with the capability to measure any delay
in Figure 4 for an example 10-hop network.
created internally within the device by switching

G.8271.1 Network Reference Points

A,B C D
±100 ns
(PRTC/T-GM) ±200 ns dTE
(random network
variation)
link asymmetry
node asymmetry compensation
±250 ns cTE ±250 ns
±550 ns cTE (11 nodes at ±50 ns per node) (short-term holdover)
±150 ns
(end application)

±1.1 µs network equipment budget

±1.5 µs end-to-end budget

Figure 4: G.8271.1 time error budget and network reference points

Synchronization Distribution in 5G Transport Networks 9

E-BOOK

This allocation of time error allows for a total of 1,000 dTE is also represented in ns, although as it varies over
ns for the transport network between the T-GM and time, it is usually specified as maximum time interval error
the T-TSC at the cell site, as shown between reference (MTIE) over the observation period, as shown in blue
points B and C in Figure 4. This time error budget is in Figure 5. Max(dTE) is the maximum dTE measured
largely taken up by asymmetry in the nodes and the from the cTE, and Min(dTE) is the minimum dTE, again
links (fibers). Managing this asymmetry is of paramount measured from the cTE, giving a negative value.
importance in building a 5G-quality mobile transport
Looking at the mobile transport network, the main
network. The remaining budget includes ±100 ns for
consideration in synchronization-friendly network design
the PRTC/T-GM; ±150 ns for the end application, which
is managing both the constant and dynamic time errors
is essentially the base station in a mobile network; and
throughout all network components, paying particular
±250 ns for short-term holdover in the base station to
attention to the asymmetry.
allow for switching to an alternative PRTC/T-GM in failure
scenarios, etc. The main contributors to time error in optical transport
The primary reason that asymmetry management is so networks can be summarized as follows:
important is that 1588v2 fundamentally assumes that the ■ Fiber asymmetry within the network. DWDM is
network is symmetrical, with exactly the same delay in typically unidirectional, with each fiber being used
both directions. Understanding the transit time from the for transmission in one direction only and a fiber
T-GM to the T-TSC is a critical part of 1588v2 operation, pair being used for a bidirectional transmission
and this is determined by measuring then halving the channel. Differences in the lengths of the fibers
time for a PTP packet to go from the T-GM to the T-TSC over the route will create a constant time error.
and back again. In a totally symmetrical world, this Differences occur in outside plant fiber, patch cable
method would give an accurate calculation, but in reality, length, repair splicing, etc. Each meter of fiber
as we will discuss at length later in the e-book, there are length asymmetry creates 5 ns of additional latency
lots of sources of asymmetry in transmission networks with a corresponding 2.5 ns of cTE. This asymmetry
that impact this measurement and need to be managed is predominantly static but will change when fibers
to enable 1588v2 operation in telecom networks. The are repaired following fiber cuts or when patch
situation is further complicated as these time/phase cables are changed during network maintenance or
errors are not static over time itself. Therefore, the reconfiguration.
Max|TE| is calculated from understanding both the
constant time error (cTE) of a node, link, or network and
the corresponding dynamic time error (dTE), as shown in
Figure 5. Time Error

Time error (TE) at any given time is the sum of cTE Max|TE|

Max(dTE)
and dTE, as shown in green in Figure 5. Max|TE| is the
dTE Time Error (TE)
maximum observed absolute value of TE in the network TE MTIE
cTE
measured from zero, is represented as a time, usually in cTE Min(dTE)

Time
ns, and is always a positive figure. cTE, shown in orange,
MTIE = Max(dTE) – Min(dTE)
is constant time error, which again is represented as a
time figure in ns and can be either a positive or negative
figure. For network components with a static error, such Figure 5: Maximum time error relationship to constant
and dynamic time errors
as optical fiber, cTE is the same at any instance in time.
For network components with a more dynamic nature,
such as an IP router, then the standards define that cTE is
calculated using an average measurement of time error
over a 1,000-second period.

Synchronization Distribution in 5G Transport Networks 10

E-BOOK

■ Dispersion compensation for non-coherent DWDM. or unintentional network instances such as fiber
Many access networks either are not yet using cuts or power grid failures. These events are not
coherent optics or mix coherent with 10 or 25 a common occurrence on an individual link in an
Gb/s on/off-keyed optics that require dispersion operational network, but the size of the random cTE
compensation. Dispersion compensation based that can be created on initial startup and in restarts
on compensating fiber (DCF) is most common can be significant.
and uses lengths of fiber cut to meet a dispersion ■ DWDM transponders and muxponders based on
requirement rather than of constant length. OTN mapping. OTN mapping chips also utilize FIFO
Variable length creates variable cTE issues in buffers, which have a latency that varies on initial
synchronization networks. Dispersion compensation startup and restarts. These deep FIFO buffers are
modules (DCM) based on fiber Bragg gratings used in OTN mapping to enable the devices to
rather than fiber remove this issue, but these are accommodate a wide range of service types and
less common in brownfield networks due to the can cause an even larger latency/delay than those in
higher cost. coherent optics. As with the FIFO buffers in coherent
■ First-in first-out (FIFO) buffers in coherent optics. optics, the figures here do not vary once the network
DWDM optics operating at 100 Gb/s and above use is up and running, but the size of this error is random
coherent optics that contain FIFO buffers within the across a large range, created on initial startup and
digital signal processor (DSP). These buffers have every restart, and differs in each direction.
a random latency/delay upon initial startup, which ■ Time error in IP routers and Ethernet switches.
varies in each optical interface and therefore varies Asymmetry within the router/switch can be created
in each direction, creating asymmetry. This creates through inaccuracies in timestamping. There are
a random time error that is constant (cTE) over strict T-BC requirements on the specification for
the shorter term but can sometimes be dynamic these devices for all aspects of time error, which are
(dTE) over the longer term if there are restarts on covered below in the ITU-T G.8273.2 section.
a link due to intentional network maintenance
Overall, these elements can be summarized as follows:

Contributor Fiber Dispersion Coherent Optics OTN Mapping IP Routing and Ethernet Switching
Compensation

Source Asymmetry in fiber Random asymmetry FIFO buffers in Deep FIFO buffers Timestamping inaccuracy.
lengths, jumper in DCF used in each DSP. Varies on in OTN mapping.
cables, etc. cTE of 2.5 direction. restart. Varies on restart.
ns/m.

Impact Large but Very large but Varying and Large and random. Tight requirements to control
predominantly static. predominantly static. random. impact.

Range Fixed cTE of ±5 to Fixed cTE of ±5 to Random cTE per Random cTE per Class A/B/C specifications. Max|TE|
1,000+ ns. 20,000 ns. device/interface device/interface of of 30 to 100 ns. cTE of 10 to 50
of ±20 to 130 ns ±20 to 1,000 ns on ns. dTE (low-pass-filtered) noise
on restart. restart. generation (MTIE) of 10 to 40 ns.

Synchronization Distribution in 5G Transport Networks 11

E-BOOK

OTN Coherent Dispersion

Mapping Optics Comp.

Fiber
Dispersion Coherent OTN
Comp. Optics Mapping
Class C T-BC Class C T-BC
Ethernet Switch Ethernet Switch

Figure 6: Sources of asymmetry in optical transport networks

Returning to the network limits outlined in G.8271.1 and the allocation within this for node and link asymmetry, it is clear
that careful design of the underlying DWDM-based transport network is required. The dTE elements of Max|TE| are largely
generated by switching/routing devices that can be managed through the use of G.8273.2-compliant devices. The cTE
elements of Max|TE| are either large static figures that can potentially be compensated for within boundary clocks or random
elements from coherent optics and OTN mapping. These random cTE elements can be managed through the careful
selection of optimized packet optical and DWDM devices with a significantly lower, and more acceptable, level of random
cTE, or through optical timing channel techniques that can bypass these elements totally. Without the careful management
of dTE and both static and random cTE across the complete end-to-end 5G transport network, these time error limits can be
costly and very hard, if not impossible, to achieve.

G.8271.1 defines ±200 ns of dTE for random network variation and ±800 ns of cTE asymmetry error, split between ±550 ns for
nodes and ±250 ns for the overall end-to-end link for a Type A network with Class A boundary clocks.

ITU-T G.8273.2 PTP T-BC Classes

G.8273.2 defines the performance of T-BC PTP boundary clocks and defines differing performance classes that devices are
required to achieve. Max|TE| and cTE are given in ns as outlined previously. dTE is specified by MTIE figures, as outlined
earlier, which are also given in ns. When measuring MTIE, the error is normally low-pass-filtered by the measurement device
using the same bandwidth as would be expected to be applied by the next clock in the chain, which is 0.1 hertz (Hz).

The table below compares the G.8273.2 T-BC Classes against various parameters. Due to the more dynamic nature of dTE,
multiple parameters are defined in G.8273.2 and multiple measurements are required to classify dTE performance. MTIE, as
outlined earlier, is the maximum error measured against the reference clock for the specified time interval. Time deviation
(TDEV) is a measurement of the phase stability of a signal over a given period of time. MTIE and TDEV are used together to
give a measurement of dTE requirements and performance.

The original G.8273 specification included Class A and B T-BC specifications and G.8273.2 added new Class C and Class
D to support tighter 5G requirements, especially to support mobile fronthaul networks. Note that Classes A, B, and C have
an unfiltered value for Max|TE|, whereas Class D uses a low-pass-filtered value. Class D also does not contain cTE and dTE
specifications as the overall low-pass-filtered Max|TE| of 5 ns is such a tight requirement that any combination of cTE and dTE
is permissible as long as the overall Max|TE| specification is met.

Parameter Conditions Class A Class B Class C Class D

Max|TE| Unfiltered 1,000 second measurement 100 ns 70 ns 30 ns -

Max|TEL| 0.1 Hz low-pass filter, 1,000 s - - - 5ns

cTE Averaged over 1,000 s ±50 ns ±20 ns ±10 ns -

dTEL MTIE 0.1 Hz low-pass filter, constant temp, 1,000 s 40 ns 40 ns 10 ns -

0.1 Hz low-pass filter, variable temp, 10,000 s 40 ns 40 ns - -

dTEL TDEV 0.1 Hz low-pass filter, constant temp, 1,000 s 4 ns 4ns 2 ns -

dTEH peak-to-peak 0.1 Hz high-pass filter, constant temp, 1,000 s 70 ns 70 ns - -

Synchronization Distribution in 5G Transport Networks 12

E-BOOK

G.8271.1 Network Reference Points

A,B C D
±100 ns
(PRTC/T-GM) ±200 ns dTE
(random network
variation)
link asymmetry
node asymmetry compensation
Class A T-BCs: ±250 ns cTE
±550 ns cTE (11 nodes at ±50 ns per node)
Class B T-BCs: ±380 ns cTE
±420 ns cTE (21 nodes at ±20 ns per node)
Class C T-BCs: ±590 ns cTE (link asymmetry compensation) ±250 ns
±210 ns cTE (21 nodes at ±10 ns per node) (short-term holdover)
±150 ns
(end application)

±1.1 µs network equipment budget

±1.5 µs end-to-end budget

Figure 7: G.8271.1 time error budget including type A, B, and C networks

As discussed in the previous section, G.8271.1 ITU-T G.8275.1 Full On-path Support
specifies ±550 ns for node asymmetry and ±250 ns
and G.8275.2 Partial On-path Support
for link asymmetry (±800 ns in total) in what it calls
The ITU-T has defined two phase profiles for PTP
Type A networks. Type A networks can contain up
networks. The first is G.8275.1, which provides full
to 11 nodes, so 10 links, with ±50 ns cTE per node,
on-path support for PTP with boundary clocks at
hence ±550 ns total node asymmetry. However, the
each IP routing or Ethernet switching node in the
G.8271.1 specification allows network operators to take
network. G.8275.1 full on-path support uses Layer 2
advantage of T-BC network nodes with better cTE
multicast Ethernet as the main delivery mechanism and
performance with Type B and Type C networks. Type
recommends the use of SyncE to assist in locking T-BC
B and Type C networks support 21 nodes, so 20 links,
nodes to a stable frequency. This profile is intended for
with ±20 ns and ±10 ns node cTE, which reduces the
all networks where new hardware is being deployed in
total node asymmetry to ±420 ns and ±210 ns. In turn,
the network, including both greenfield new deployments
this increases the possible link asymmetry to ±380 ns
and cases where new routing or switching hardware is
(Type B) and ±590 ns (Type C) while maintaining the
being added to a network to support higher capacity
overall ±800 ns total node and link cTE. dTE must still be
or performance for 5G. Overall, this approach provides
managed within the available ±200 ns random network
superior PTP performance and is recommended for any
error budget. Therefore, it is highly desirable to take
new network buildout where mobile traffic is planned.
advantage of T-BC nodes with Class B or ideally Class
C performance and lower cTE within a mobile transport To allow for the fact that not all upgrades involve
network when they are available so as to enable a Type complete upgrades to all routing and switching hardware
C G.8271.1 network with its higher allocation of cTE to in the network, the ITU-T also developed the G.8275.2
the link elements of the transport network, such as fiber partial on-path support profile. Partial on-path support
asymmetry and asymmetry within DWDM devices. uses Layer 3 unicast IP as the main delivery mechanism.

Synchronization Distribution in 5G Transport Networks 13

E-BOOK

This profile uses T-BC functionality at intermediate From a synchronization perspective, fronthaul adds
nodes and, wherever possible, T-TC functionality at another level of complexity, with the need to not only
other nodes to reduce noise and improve performance deliver high-quality frequency and phase synchronization
over generic Ethernet clock-enabled devices. Some but also to manage relative phase synchronization error
operators are unable to utilize G.8275.1, with its between adjacent cell towers within the cluster of cells
better performance, and therefore need to utilize under the DU. The specific level of relative phase error
G.8275.2 partial on-path support when they upgrade budget is highly dependent on the functionality being
older networks. Generally speaking, G.8275.2 is not utilized within the RAN and within the specifications
recommended for 5G synchronization distribution due defined by the 3GPP’s technical specification (TS) 38.104.
to its limitations. It is possible to use assisted partial TS 38.104 defines the capabilities of eCPRI-based
timing support (APTS) mode to help mitigate the impact fronthaul networks and includes the maximum relative
of issues such as network islands without PTP support, phase error that is allowed when specific functions are
but this is complicated and potentially unreliable. In order used within the RAN.
to minimize the impact of the poorer synchronization
The most demanding functionality, such as inter- and
performance, these operators often need to introduce
intra-band carrier aggregation and the use of MIMO
more T-GM clocks into the core network to reduce the
antennas, have very demanding relative phase error
distance between the T-GM and the cell tower.
budgets in 5G networks. This is specified as relative
The specifications of both G.8275.1 and G.8275.2 |TE| measured at the UNI of the RU of as low as 190 ns
contain a range of features that are required to support to just 60 ns. The corresponding time alignment error
the profile across a network domain. The level of (TAE) specification, which is defined as the largest timing
support for these features by the products deployed difference between any two signals, of 260 ns to just
within the network will determine the overall level of 130 ns as measured at the antenna, is shown in Figure
support for either of the profiles and the corresponding 8. Of course, this should not be confused with the 1.5 µs
synchronization delivery performance. absolute phase error requirement, which is still required
at every cell site – this is an additional requirement to
3GPP TS 38.104 Time Alignment Error control the relative phase error between all the cell sites
The ITU-T specifications that have been described within a 4G BBU or 5G DU cluster. These relative |TE|
so far in this e-book provide the specifications for and TAE specifications also apply to cooperating cell site
frequency and phase synchronization delivery through clusters that are subtended from multiple DUs.
a transport network to meet the requirements for both
4G LTE and 5G TDD mobile networks, with frequency
synchronization of 50 ppb at the air interface of the
RAN and 16 ppb from the backhaul network, and phase TAE
RU

synchronization of 1.5 µs with a budget of 1 µs allocated (130/260 ns)

Relative |TE|
to the transport network, as outlined above. (60/190 ns) Fronthaul DU

These networks operate with either a distributed RAN RU

(DRAN) architecture with a 4G BBU or 5G DU located
at the cell site or a centralized RAN (CRAN) architecture
Figure 8: Relative |TE| and TAE in fronthaul networks
with the BBU/DU a short distance (15-20 km max) from
the cell site and a fronthaul network connecting the BBU/
DU to the corresponding 4G remote radio head (RRH)
or 5G radio unit (RU). Most mobile operators started
5G deployments following their existing 4G LTE DRAN/
CRAN architectures, and as they start to migrate to more
advanced 5G services and architectures, more will move
to CRAN and eCPRI-based fronthaul.

Synchronization Distribution in 5G Transport Networks 14

E-BOOK

Putting It All Together to Provide cTE must also be managed and within budget for the
worst-case links, which are often longer routing paths
5G-Quality Synchronization
around networks in protection scenarios where the
Providing 5G-quality synchronization is a complex
shortest path has failed. cTE is mainly created within
problem that needs careful consideration early in the
the DWDM layer, and therefore, careful consideration
network design process. Many factors that impact
of the cTE performance of these devices is required.
synchronization quality are fundamentally linked to the
For long DWDM links that interconnect T-BC Class B/C
operational performance of networking hardware and
routers or switches, it may well be the case that even
cannot simply be solved for at a later stage without
Class D T-BC devices are needed at intermediate DWDM
replacing substandard networking hardware.
nodes in order to maintain the required synchronization
Networking hardware needs to meet the strict performance within the underlying DWDM network.
requirements outlined in the preceding sections of this
Fronthaul networks: eCPRI/Ethernet-centric domains
e-book and summarized below in Figure 9. Every aspect
where Ethernet switches must support the necessary
of the transport network needs to be considered from
synchronization standards and meet T-BC Class C to
a synchronization point of view. Each domain within the
support both the absolute and relative phase error
mobile network must be optimized for synchronization
budgets. dTE is largely from eCPRI/Ethernet switching
performance as follows:
devices that will typically also use Time-Sensitive
Midhaul and backhaul networks: IP-centric domains Networking (TSN) capabilities, such as preemption,
where any IP routers must support the necessary to prioritize latency-sensitive packets such as eCPRI
synchronization standards and meet T-BC Class B, or fronthaul traffic and PTP packets over other traffic within
ideally Class C, to enable type C networks with a larger the network. Again, cTE is largely from the underlying
allocation of cTE budget to support the links connecting DWDM layer, and as cTE must be considered from the
the T-BC clocks. Most of the dTE within the network will PRTC to the cell tower, fronthaul cTE must be managed
come from these IP devices, and any Layer 2 Ethernet from end to end along with midhaul and backhaul
devices in the network and dTE must be managed and domains.
within budget across the complete network.

3GPP TS 38.104 Time Alignment Error ITU-T G.8271.1 Network Limits IP Router or Ethernet
Switch with inbuilt T-BC
ITU-T G.8273.2 PTP T-BC Class C ITU-T G.8273.2 PTP T-BC Class B/C
ITU-T G.8275.1 Full On-path Support ITU-T G.8275.1 Full On-path Support
ITU-T G.8262.1 eEEC Synchronous Ethernet ITU-T G.8262.1 eEEC Synchronous Ethernet

Fronthaul Midhaul CU Backhaul

4G Network DU BBU Network MEC
MEC Network
CPRI MEC
eCPRI
5G T-TSC CPRI PRTC/T-GM

5G T-TSC
Ethernet
eCPRI

4G

5G
IP/MPLS Access DWDM IP/MPLS Metro DWDM IP/MPLS Core DWDM

Network Budget Fronthaul Midhaul and Backhaul

Max|TE|
ePRT/PRTC
eRE eCPRI Transport 60 ns - 190 ns eREC Transport Network 1000 ns 15 ns - 100 ns

Inter antenna relative phase alignment TDD/LTE-A and 5G macro phase services traceable to UTC

Figure 9: Delivering 5G-quality synchronization across the transport network

Synchronization Distribution in 5G Transport Networks 15

E-BOOK

Getting Synchronization Right In summary, delivering high-quality synchronization

Hitting the required synchronization performance is a must for 5G networks, and it is not simply a case
within a network requires a combination of features/ of meeting the minimum possible standard. Superior
functionality and measurable performance synchronization performance can bring improved
characteristics, such as cTE and dTE of networking network performance and user experience. It is always
devices. The specifications enable the transport network a balancing act over the economics of chasing ever-
to deliver a suitable clock to the cell towers in the mobile improving synchronization performance, but the goal
network of the right quality to support the demanding should always be to get the best performance that meets
features required in the 5G RAN. Meeting the basic or exceeds the required level for 5G without breaking
synchronization performance levels enables network the bank.
operators to fully utilize their most valuable asset, their
spectrum. Lower synchronization performance can mean
that frequency management within the RAN isn’t tight
enough and the spectrum can’t be fully utilized or that
advanced functionality such as carrier aggregation or
MIMO antennas cannot be fully utilized or even activated
at all. Overall, getting synchronization right is mandatory
in mobile networks.

However, synchronization performance is not simply a

pass/fail test. It is possible to exceed the synchronization
specifications and have a higher-performance network.
It is too early to see the impact of higher synchronization
performance that exceeds the specifications in full
5G standalone (SA) networks yet, although we can
be sure this certainly will not be a negative factor. But
we can look back at 4G networks to see how above-
standard synchronization helped network operators
improve network efficiency and user experience. These
performance improvements are hard to quantify in
most cases – sometimes it is simply a case of network
engineers being able to tell which backhaul network
a cell tower is supported by just from looking at the
performance metrics for the cell site, for example, one
backhaul network with OK synchronization versus one
with much better synchronization performance. In other
cases, network operators have used before-and-after
user experience data such as upload speed, download
speed, and latency and seen over 40% improvement in
all metrics once the backhaul network was upgraded
to one with much better synchronization performance.
In this instance, there was no increase in backhaul
capacity to the cell site, just better synchronization over
the backhaul network and the capacity for growth in the
future.

Synchronization Distribution in 5G Transport Networks 16

SECTION 2

Infinera’s Sync
Distribution Solution
for End-to-End
Synchronization
Delivery
E-BOOK

In order to meet the complex transport requirements in Synchronization in the IP Layer

5G mobile networks, Infinera has developed a complete Infinera’s IP portfolio for mobile applications embraces
end-to-end mobile transport solution. This includes the open and disaggregated approach as promoted by
synchronization delivery for all aspects of the optical industry organizations such as the Telecom Infra Project
network from the mobile core to the cell tower. The (TIP). This approach takes the traditional closed router
following section of this e-book will outline the Infinera architecture, with a complete software and hardware
synchronization delivery solution by looking at its major package provided as a single entity from one vendor,
building blocks. These building blocks can be deployed and breaks it into network operating system (NOS)
in a standalone manner for a particular segment, layered software and open white box hardware from potentially
in the 5G transport network, or deployed together as a different vendors. Synchronization features in this
complete solution. environment rely on both software capabilities within the
NOS and hardware features within the specific white box
Infinera has a full portfolio of DWDM, Layer 2/2.5 packet
hardware selected for the various network domains by
optical, and IP products that are widely deployed
the network operator.
across fixed and mobile networks around the globe.
The synchronization distribution solution outlined Infinera’s Converged NOS (CNOS) builds on 15 years
in this section focuses on the products within the of experience in IP networking, especially in mobile
portfolio that are most commonly positioned today for networks with Infinera’s 8600 Series of IP routers. With
mobile networks where synchronization is a critical a deployed base of over 200,000 routers supporting
consideration. Within the DWDM and Layer 2 packet over 300,000 cell towers in leading mobile operators’
optical layers of the network, this comprises the XTM networks across the globe, the 8600 Series was
Series within the access and aggregation packet optical optimized for IP in mobile environments, and CNOS
domains at the edge of the network and the OTC2.0 takes that heritage into the open disaggregated age.
solution deeper in the network across regional and long- CNOS includes the broad range of synchronization
haul networks. Other products within Infinera’s portfolio, features that are required in order to provide 5G
such as the 7090 and 7100 packet optical platforms, are synchronization, including:
also widely deployed in mobile networks, providing high-
quality synchronization delivery.
■ ITU-T G.8262 and G.8262.1 eEEC
■ Station clock input and output ports
■ Pulse-per-second (PPS) input and output
■ Time-of-day input

5G Mobile Orchestration (Includes RAN and Core)

5G Transport and Intelligent Orchestrator

RU DU CU 5G
Core

Access Aggregation Core

Fronthaul Midhaul Backhaul

4G
EPC
4G

Autotuneable Ethernet-based Point-to-Multipoint Optical Disaggregated

DWDM Pluggables Fronthaul with TSN XR Optics Transport Routing

Autotuneable and point-to-multipoint DWDM optics, Ethernet-based fronthaul, and

disaggregated routing expand the toolbox and reduce costs for 5G transport

Figure 10: Infinera’s end-to-end 5G transport solution

Synchronization Distribution in 5G Transport Networks 18

E-BOOK

■ Synchronous Ethernet standard white box hardware and includes a range of

■ SSM over Ethernet [G.8264] devices optimized for mobile networks from the cell site
■ SyncE Master (GNSS input as PRC for SyncE- through aggregation nodes to the core. The extended
unaware island) capabilities found in the DRX-30 and DRX-90 include the
■ IEEE 1588v2 Boundary Clock for phase broad range of synchronization features needed for 5G
synchronization (G.8275.1, G.8273.2) mobile networks, such as:
■ Fully G.8275.1 profile compliant, including priority 2
and local priority attributes ■ GNSS receiver ports for customers that use
■ G.8275.1 topology control via port configuration GNSS-based synchronization strategies
■ Auto (the PTP state is determined by the ■ Timing port input/output options, including
BMCA rules) 1PPS/ToD ports and PTP ports
■ Master (the PTP state is determined by the ■ T-BC Class C performance
BMCA rules, with the “notSlave” attribute of
G.8275.1 enabled)
Class C performance at every IP node is an important
■ Disabled (all frames with PTP EtherType tool in enabling the wider 5G transport network to
0x88F7 are dropped)
achieve the G.8271.1 network limits with tight control
■ IEEE 1588v2 PTP telecom profile for frequency
synchronization (G.8265.1) over dTE budgets. Furthermore, the CNOS software and
DRX platforms also support a comprehensive range of
■ 1588v2 HW base timestamping
features within G.8275.1, beyond standard features such
■ SyncE Assist
as T-GM/T-BC/T-TC/T-TSC clock options, alternative best
■ GNSS input (built-in GNSS chip) including the latest master clock algorithm, and manual configuration of
technology:
various topology preference options.
■ Multi-constellation
■ Synchronization by only single satellite
visibility
■ Multi-band (solution to mitigate ionosphere
delay)
■ Ionosphere is one of the biggest error
sources in GNSS timing Figure 11: Infinera’s DRX-30 DCSG router with T-BC Class C
performance for demanding 5G networks
■ Monitoring and measurement capabilities
Of particular interest in the wider 5G sense is the push
In synchronization hardware matters, and key to
by TIP for an open approach to their Disaggregated
understanding how this broad range of features can be
Cell Site Gateway (DCSG) router, which provides a
deployed is the understanding of the synchronization
standardized hardware specification for 5G networks,
features and relative performance of the underlying
including synchronization features. This open approach
hardware. Today’s white box hardware has evolved from
has been embraced by many hardware and software
its data center origins into carrier-grade hardware where
vendors and has now started commercial deployment
many, but not all, of the options available in the market
with a range of mobile operators, including some of
support the capabilities needed for complex networks
the world’s largest, such as Telefónica and Vodafone,
such as mobile transport networks.
both of which have deployed Infinera’s CNOS software
Infinera’s CNOS is an open NOS designed to run on a over either Infinera DRX or Edgecore DCSG hardware.
variety of commercially available hardware platforms Synchronization capabilities are key selection criteria
from both Infinera and third-party vendors, with in 5G networks, and real-world deployments using
commercial deployments to date running on both network-based synchronization delivery are an important
Infinera and Edgecore hardware. validation of the broad synchronization feature set and
the high performance in IP networks.
Infinera’s disaggregated IP hardware, the DRX Series,
brings additional carrier-grade functionality over

Synchronization Distribution in 5G Transport Networks 19

E-BOOK

Synchronization in Packet Optical the EMXP range with the EMXP-XH800, which is a
hardened 800 Gb/s device supporting a broad range of
Transport in Metro and Regional
functions required from fronthaul networks and hybrid
Networks xHaul networks, which encompass fronthaul, midhaul,
To interconnect the devices within the IP layer, DWDM is
and potentially backhaul traffic flows over the same
typically used for reach and fiber capacity or availability
infrastructure, such as TSN.
reasons. As outlined in part one of this e-book, the main
challenge in delivering synchronization over fiber and
DWDM is controlling asymmetry and corresponding
cTE, although any elements in the network that extend
into Layer 2 Ethernet switching will introduce a dTE
factor that also needs management. To understand how Figure 12: XTM Series EMXP-XH800 for fronthaul
and xHaul networks
cTE and dTE can be managed across the network, the
packet optical domain will be subdivided further into the From a synchronization perspective, the EMXP-XH800
Ethernet switching layer and the optical DWDM layer for brings the range of synchronization features needed
access and aggregation networks and legacy/long-haul to support 5G fronthaul and xHaul environments, such
networks. as fiber asymmetry compensation, SyncE/eEEC, and
nanosecond-level timestamping for very accurate T-TC
Packet Optical Transport with Layer 2 operation coupled with T-BC Class C performance that
Ethernet/eCPRI Switching significantly exceeds the required performance for Class
Fronthaul networks and those that support a combination C certification, as shown in Figure 13. Along with the
of front/mid/backhaul over an xHaul infrastructure rest of the EMXP range, the EMXP-XH800 also utilizes
utilize Ethernet switching capabilities that from a a hardware design with a highly accurate SyncE assist
synchronization perspective need to be considered in a mode for 1588v2 PTP operation that is optimized for
similar manner to the IP layer due to the predominately demanding 5G fronthaul applications.
dTE implications on synchronization. Due to the
extremely tight relative phase error budgets within
fronthaul networks, which can be as low as 60-190 ns,
synchronization performance is critically important within
these networks.

Infinera’s XTM Series is widely deployed in packet

optical networks with a broad range of EMXP devices.
The EMXP range utilizes a switch-on-a-blade architecture
that provides significantly better synchronization
performance than comparable packet optical transport
devices. In wholesale environments, the switch-on-a-
blade architecture enables the XTM Series to support
multiple synchronization domains within a single chassis,
which enables wholesale operators to support multiple
mobile operators over the same chassis and network,
each with its own independent synchronization domain.
Figure 13: EMXP-XH800 MTIE performance
To address the new eCPRI-based transport requirements
within fronthaul networks, Infinera has extended

Synchronization Distribution in 5G Transport Networks 20

E-BOOK

Most xHaul networks require tight control of both DWDM Transport

cTE and dTE in order to meet tough synchronization In addition to widespread deployments within 4G and
requirements. In the most demanding networks, the 5G Layer 2 packet optical networks, the XTM Series also
EMXP-XH800 can be coupled with an optical timing supports many network operators with Layer 1 DWDM-
channel (OTC) approach that bypasses coherent optics based mobile transport networks for both 4G and 5G
and other optical components that can add further transport.
elements of fixed and random cTE. The fixed and random
cTE of these elements may be small, but over a complete One of the most challenging aspects of building
optical network, they can add up to considerable levels synchronization distribution networks is controlling fixed
that need management. dTE is kept low through the and random cTE within the DWDM links that interconnect
hardware design of the device and is a key factor in the Ethernet or IP devices. In metro and regional mobile
T-BC performance that significantly exceeds Class C. transport networks, Infinera uses the XTM Series as
the platform is highly optimized for this application.
The optimization includes a wide range of factors such
as TSN support for mobile fronthaul and hardened
hardware options. Of particular importance from a
100/200G GbE High-density CWDM synchronization perspective are:
DWDM (Synchronization OTC)

■ Single platform for packet optical networks – when

Layer 2/2.5 capabilities such as those described
Optical
Amplifiers
ROADM Optical
Amplifiers
above are required, the XTM Series provides these
in a single platform
Figure 14: EMXP-XH800 optical timing channel in action
■ Optical timing channel – HDCWDM single fiber
working optical timing channel supports PTP in
The EMXP-XH800 optical timing channel implementation higher layers either through the XTM Series EMXP-
uses a high-density CWDM (HDCWDM) channel that XH800, Infinera’s OTC2.0 solution (outlined below),
provides bidirectional single fiber working (SFW) or other third-party timing solutions
operation over a single CWDM wavelength to remove
■ DWDM transport options with very low random cTE
the majority of the fixed and random cTE elements of
the underlying DWDM and fiber components. The use At the DWDM layer, this third item is probably the
of Gigabit Ethernet CWDM optics enables longer optical most critical as without it, 5G-quality synchronization
reach, which means the configuration can use CWDM distribution networks can be very difficult to build
filters to bypass the inline optical amplifiers and high- and can rapidly become very expensive to build and
speed 100G/200G coherent optics due to the longer maintain. Mobile operators often design, build, and
reach of the lower-speed CWDM optics. SFW requires manage their DWDM transport and IP layers as separate
a different wavelength in each direction over the fiber domains, which means that ideally the DWDM layer
and HDCWDM uses two tightly spaced CWDM channels needs to have very low cTE, including those components
within a single standard CWDM channel. Using just one that create random cTE, to enable the DWDM layer to
of the fibers removes the asymmetry due to differing support PTP packets within the IP data plane without the
fiber lengths in DWDM networks and limits overall fiber need for special management of these IP flows.
asymmetry to the very small level of asymmetry from the
differing speeds of the two wavelengths. The biggest challenge with this underlying DWDM layer
is cTE, or random cTE from OTN mapping chips used
Overall, the XTM Series EMXP range and the EMXP- in transponders and muxponders. These devices use
XH800 in particular enable network operators to build deep FIFO buffers to enable support for a broad range
packet optical networks for mobile xHaul that meet or of service types, which of course is an advantage in
exceed the tight requirements for 5G transport. general networking terms but a serious challenge from a
synchronization perspective.

Synchronization Distribution in 5G Transport Networks 21

E-BOOK

The XTM Series provides a range of DWDM restart there will be a random cTE within ± the quoted
transponders and muxponders that are OTN-based and figure. Devices with larger random cTE may well initially
use the same commercial off-the-shelf (COTS) OTN chips start up with a lower acceptable level of cTE but in a
as the rest of the industry. In addition, the XTM Series restart situation this may change to a much larger and
also contains devices that are optimized for applications unsupportable level of cTE.
such as mobile transport with a very tight focus on
By careful network design, it is therefore possible
optimal performance for a more limited set of services.
to build a DWDM transport layer that is capable of
These devices avoid the COTS OTN mapping chips and
supporting 1588v2 PTP in higher networking layers with
focus on providing a low-latency, low-power, and high-
a low enough cTE within DWDM links that the overall
density offering with the very positive side effect of a
G.8271.1 network limits can be achieved.
very low cTE on restart.
This challenge is compounded by the fact that optical
To put this into perspective, Infinera has tested a wide
layer design is built around fiber availability and routing,
range of transponders, muxponders, and packet optical
switches from the EMXP range for random cTE and dTE and while the normal working path may be a relatively
direct route between two T-BC-enabled routers or
performance, and the results are summarized below.
switches, the protection route may be substantially
The cTE figures quoted are maximum random cTE figures longer and involve a lot more DWDM components that
for cTE on restart of the device, and therefore on each will potentially have a substantial impact.

XTM Series Function Client Line Maximum Maximum dTEL 5G Phase Sync
Device Random cTE (Low-pass- Support?
filtered) MTIE

Dual 100G/200G flexponders on 100G 100G or ±20 ns 0 ns Yes

FXP400G
a single card. 1 or 2 x 100G clients 200G
mapped into each of the 100G or 200G
lines.

200G multi-service muxponder. Various 10G, 32G 100G or ±670 ns 0 ns No

MXP200G
lower-speed services (OTN, Ethernet, FC, 100G, 200G
Fibre Channel, etc.) at rates from 10G to etc.
100G mapped into 100G or 200G line.

6 x 10G transponders on a single card 10G 10G ±10 ns <1 ns Yes

FHAU/1
or hardened pizza box option for street
cabinet deployments. Non-OTN-based
mapping.

6 x 10G transponders on a single card. 10G 10G ±372 ns <5 ns No

TPHEX10GOTN
OTN-based mapping.

440G packet optical transport 10G 100G or ±37 ns 2 ns Yes

EMXP440
switching card. 200G

800G hardened pizza box packet 10G or 100G or ±28 ns 2 ns Yes

EMXP-XH800
optical transport switch. 25G 200G

Synchronization Distribution in 5G Transport Networks 22

E-BOOK

L2 Switch or Grandmaster
IP Router

T-BC T-BC Optical Timing Channel T-BC

ROADM ILA ROADM

Node Node Node

Figure 15: Infinera’s OTC2.0 Solution

Synchronization in DWDM Transport The TimeProvider 4100 supports the very broad range of
synchronization features required for 5G synchronization
over Regional, Long-haul, and Legacy
and timing distribution, such as a high-performance
Networks boundary clock operational mode, GNSS and network
Outside of the metro access, metro aggregation, and
inputs, multiple output options, and the full range of
regional footprint that is addressed with the XTM Series,
frequency and phase synchronization standards. The
Infinera has developed a very high-performance OTC2.0
device also couples T-BC Class D performance with a
solution in conjunction with Microchip, a market leader in
range of rubidium and OCXO local oscillator options to
network synchronization technology.
provide a very high-quality timing source for downstream
The OTC2.0 solution builds on the combined networking nodes. A summary of the TimeProvider 4100
synchronization and optical networking strengths of the features that are utilized in the OTC2.0 solution includes:
two companies to provide network operators with highly
■ IEEE 1588v2 PTP grandmaster
optimized and highly reliable synchronization distribution
solutions. OTC2.0 provides synchronization distribution ■ Timing distribution over T-BC 1588/SyncE via
over Infinera’s full portfolio of DWDM platforms, such overlay optical timing channel
as the 7100, 7300, FlexILS, and GX Series platforms, or ■ GNSS (GPS, GLONASS, BeiDou, QZSS, and Galileo)
even over third-party DWDM networks. The solution can and SBAS support
also be deployed over the XTM Series when extreme
■ PRTC Class A and Class B
performance and an enhanced synchronization/timing
feature set is required. OTC2.0 essentially couples ■ Enhanced PRTC (ePRTC) that meets 30 ns
Microchip’s industry-leading TimeProvider® 4100 with a performance and uses a combination of Cesium
broad range of DWDM optical timing channel capabilities and GNSS time sources
and a deep understanding of how the two systems can ■ Oscillator options – SuperOCXO (future support),
be optimized to meet the toughest synchronization OCXO, and rubidium (Rb)
requirements for mobile networks. ■ Standard base unit with 8 Ethernet ports, 4 E1/T1
ports, 1 craft port, 2 × 1PPS/ToD ports, 2 × 1PPS/10
MHz ports

Figure 16: Microchip TimeProvider 4100

Synchronization Distribution in 5G Transport Networks 23

E-BOOK

■ Optional internal expansion module with 4 SFP and From an overall solution perspective, OTC2.0 provides
4 SFP+ for 10G support, 100M Fast Ethernet, and 1G network operators with a high-performance timing
fanout solution that is also highly robust and very scalable.
■ Support for multiple IEEE 1588v2 profiles per unit The solution is decoupled from the underlying DWDM
layer, which enables timing resiliency during network
■ Support for high-performance single domain/multi- upgrade and reconfiguration activities. The broad
domain boundary clock with Class C and D accuracy range of synchronization and optical timing channel
■ Fully supports ITU-T profiles for phase features, coupled with support for any transport network
synchronization: G.8275.1 and G.8275.2 topology, including meshed, ring, tree, and point-to-
■ Fully supports ITU-T profiles for frequency point architectures, enables network operators to bring
synchronization: G.8265.1, Telecom 2008, and 5G-quality synchronization to even the most demanding
default networks, including those with high levels of asymmetry.

■ ITU-T G.8273.4 APTS with enhanced automatic Field deployments of OTC2.0 using TimeProvider
asymmetry compensation over multiple network 4100 have shown that networks can exceed G.8273.2
variations Class D cTE performance over long-distance DWDM
■ Supports timing in DWDM networks with up to 6 networks. Figure 17 shows one-week live traffic test
DWDM degrees, optional extension to 14 DWDM data of the OTC2.0 solution in action over a 500-km,
degrees 96-channel DWDM network. The network comprises
six DWDM spans connecting a combination of ROADM
■ PTP timing path protection (bidirectional timing
and in-line amplifier (ILA) sites using a mix of EDFA-only
service for resiliency)
and hybrid EDFA/Raman amplification options. With
■ Monitoring and measurement capabilities TimeProvider 4100-based T-BC timing at the end nodes
■ Multiple management options – Microchip and the five intermediate sites, this is therefore seven
TimePictra® synchronization management system hops from a synchronization perspective. The results
support, Microchip Web GUI, CLI, SNMP, and Open show an impressive end-to-end cTE performance of 11 ns
API support planned (NETCONF) throughout the one-week monitoring period. To put this
performance into perspective, Class D performance over
Looking at the optical layer, OTC2.0 uses two very seven timing hops would result in a time error of 35 ns at
tightly spaced WDM channels, often bidirectionally 5 ns per hop.
over a single fiber, for transmit and receive channels
for PTP messages to minimize network asymmetry and
the corresponding impact on PTP operation. OTC2.0
provides a broad range of WDM options such as O-, E-,
and L-band timing channel options to optimize these
timing channels to the specific characteristics of the
DWDM network. Furthermore, the solution utilizes both
PTP T-BC and 3R DWDM regeneration options to ensure
a high-performance, robust, and economical network.

Critically, Infinera’s OTC2.0 also brings a detailed

understanding of how DWDM and synchronization
interact over complex and varied DWDM networks, such
as those involving Raman amplification over very long
distances. Furthermore, the capabilities and expertise
Figure 17: OTC2.0 time error of 11 ns after 500 km over six
in Infinera’s Synchronization Verification Lab in Munich DWDM spans and seven timing hops
enables Infinera and Microchip to collaborate to verify
and optimize OTC2.0-based synchronization network
designs for customers.
Synchronization Distribution in 5G Transport Networks 24
E-BOOK

36.2 km 53.7 km
vPRTC (Timing Cloud)
9 km 13 km 56.6 km
with every DWDM node 5G Fronthaul/Midhaul
capable of delivering 16 km 39 km
timing/synchronization 24.3 km 6 km
9 km
10 km
48.3 km
58.2 km 56.2 km
5G Fronthaul 53.3 km 54 km

/Midhaul
100 ns 72 km

53.6 km 54.5 km
81 km 38.1 km
50 km
65 km 60 km
35.3 km Redundant GNSS
spoofing hardened
71 km GM/Cesium source

53 km
Range of 100 ns vPRTC 44 km 51 km 58 km 72 km 60 km Other timing-critical applications
(Timing Cloud) resilient (power distribution, scientific)
timing service

Figure 18: OTC2.0-enabled vPRTC regional/core network preserves G.8271.1 network limits for
fronthaul and midhaul packet optical networks and other timing-critical applications

The combination of synchronization and optical End-to-End Sync Planning and

networking capabilities within OTC2.0 also enables
Management Interconnection
network operators to create a highly flexible and resilient
Having high-quality synchronization requires more
virtual PRTC (vPRTC) or “Timing Cloud” architecture
than simply high-performance hardware and software
over long-haul and regional networks. This essentially
capabilities. Planning and management capabilities are
converts the complete regional/long-haul network into a
critical in building real-world networks and managing
distributed PRTC within the G.8271.1 PRTC budget of ±100
synchronization through the life cycle of a network.
ns, or even potentially the enhanced ePRTC budget of
±30 ns, with timing source redundancy. This architecture During the network planning and design stages,
pushes the PRTC from a few core node locations out Infinera’s planning tools enable network designers to
across the long-haul and regional network, preserving model fixed and random cTE and dTE across Infinera’s
critical G.8271.1 budget for metro access and aggregation synchronization portfolio to ensure that optical layer
DWDM networks and further last mile access for non- designs support the required cTE and dTE budgets, as
fiber-connected cell sites, as shown in Figure 18. well as from an optical design perspective.

Figure 19: Infinera’s Transcend synchronization management in action

Synchronization Distribution in 5G Transport Networks 25

E-BOOK

Figure 20: Infinera’s Transcend 3D synchronization view with frequency and phase planes

Once installed, Infinera’s Transcend NMS (TNMS) Summary

provides synchronization management features to 5G synchronization is a complex topic with many moving
provide visibility, management, and traceability of parts that all need to come together harmoniously
synchronization signals and sources, such as: across all aspects of the transport network to provide
the right quality synchronization to the cell tower without
■ Support for configuring all synchronization-related
overengineering the network and driving up cost.
functionalities in node manager
Infinera’s toolbox of high-performance synchronization
■ IEEE 1588v2, phase/time and frequency
capabilities is enabling both mobile network operators
■ SyncE
and wholesale carriers that provide mobile transport
■ GNSS synchronization services to deliver network-based synchronization
■ IEEE 1588v2 clock monitoring and troubleshooting with industry-leading performance. In some cases,
this Infinera solution simply supports high-quality
■ Third-party network element integration in common
synchronization within a particular network layer or
sync network map
geographic domain, and in others, operators are able to
■ Synchronization network view combine the solutions outlined in this e-book to create
■ Phase/time and frequency end-to-end synchronization strategies to meet their 5G
■ 3D network view performance demands now and in the future.

GNSS has a critical role to play in synchronization

distribution, but operators are moving away from “GNSS
everywhere” strategies to those that utilize resilient
GNSS at key network locations and network-based
synchronization distribution. This approach provides
better holdover performance and removes the risk of
GNSS jamming and interference at cell sites.

Synchronization Distribution in 5G Transport Networks 26

E-BOOK

In-band delivery of synchronization

T-BC T-BC T-BC
• Transponder synchronization performance
1. In-band T-GM
• Coherent synchronization performance
Delivery • High-performance PTP 1588 and SyncE delivery
(XTM Series)
Out-of-band delivery of synchronization
• Very high-performance PTP 1588 and SyncE
T-BC T-BC T-BC • Single-fiber CWDM and O/E/L-band overlay
2. Optical T-BC/3R T-BC/3R
T-GM • OTC network elements:
Timing
• T-BC Class D boundary clocks
Channel
• Optical 3R regeneration
(OTC2.0)

Hybrid use of in-band and OTC mechanisms

• Interoperable and interchangeable
T-BC T-BC T-BC
3. Hybrid Sync T-BC/3R • All high-performance 1588 PTP, not proprietary
T-GM
Distribution • Use the appropriate solution for the best fit
• In-band delivery perfect for metro-access
• OTC widely used as core distribution

Figure 21: Infinera’s synchronization options in DWDM transport networks

It also removes the challenges of providing GNSS signals This e-book has focused on synchronization in mobile
into hard-to-reach cell sites planned for 5G, such as networks as this is a large focus area currently within the
those deep inside buildings or in underground metro telecom industry. But the benefits of high-performance
railway stations. synchronization are not limited to mobile networks.
Network operators are also benefiting from Infinera’s
The benefits that Infinera’s solution brings to high-quality synchronization in a broad range of
those operators that are building network-based applications such as power utility networks, financial
synchronization strategies over alternative approaches trading networks, TDM circuit emulation, and video/DAB
include: distribution networks.
■ Better overall synchronization performance
Further Reading
leading to potentially better RAN performance and
Infinera has a range of more detailed product-specific
spectrum utilization
synchronization documentation, such as product data
■ Better overall network economics with optimized sheets and sync performance testing documentation for
solutions for in-band synchronization delivery for the solutions outlined in this e-book. Please contact your
metro access and aggregation networks, OTC2.0 Infinera sales representative for more details.
for long-haul/core/legacy networks, and the ability
to blend the two solutions within the same network All Infinera and Microchip product feature lists and
specifications referenced in this e-book are subject
■ More resilient synchronization distribution
to change over time. Please refer to the appropriate
■ More stable synchronization environments requiring product data sheets and detailed documentation for the
less ongoing maintenance and support most up-to-date features lists and specifications.

© 2021 Infinera Corporation. All Rights Reserved. Infinera and logos that contain Infinera are trademarks or registered trademarks of Infinera Corporation in the United States and other
countries. All other trademarks are the property of their respective owners. Statements herein may contain projections regarding future products, features, or technology and resulting
commercial or technical benefits, which are subject to risk and may or may not occur. This publication is subject to change without notice and does not constitute legal obligation to deliver
any material, code, or functionality and is not intended to modify or supplement any product specifications or warranties. 0282-EB-RevA-0321