Synchronizing IP Mobile Networks

WHITE PAPER
Synchronizing IP Mobile Networks
A Practical Guide to Synchronization For Long Term Evolution Wireless Systems

THE BUSINESS CHALLENGE Technically, the data must be transported over the air, and
then delivered via the backhaul (often referred to as the
As operators tested 3G killer applications a mobile internet
Radio Access Network) to the high capacity core network. Air
generation emerged, taking for granted high-capacity
interface protocols satisfy the current data demand, but the
access irrespective of time or location. Text messaging no
backhaul does not have the transport capacity needed for the
longer satisfies the user needs, being replaced by bandwidth
data transfer.
intensive social networking, picture messaging and streaming
video applications. Unlike voice services, mobile broadband is
“always-on”, and must always be available. IP/Carrier Ethernet meets the mobile business needs
… increased bandwidth and for less money.
An estimated 500 million subscribers will have access to
mobile broadband by the end of 2011, exceeding the number
of wireline subscribers1. Supported by generous data plans,
With the largest part of the operating cost being attributed to
smart handsets and wireless modems, the demand for mobile
the Radio Access Network, the linear relationship between
broadband continues to grow exponentially and is the new
bandwidth and cost for channelized E1/T1 circuits does not
business opportunity for service providers.
meet the financial goals. This is also true for leased circuits,
despite falling prices. Ultimately, mobile operators need to
Mobile broadband has become reality. It has changed transport more data for less money, and TDM backhaul is not
all our paradigms. a viable long term solution.

IP/Carrier Ethernet is a practical and therefore inevitable

choice for the backhaul.
But that opportunity is not without challenges.
When looking at data forecasts, it was also evident that
Fiscally, data demand is growing exponentially, but the
existing radio schemes won’t meet future demands, and
added revenue is incremental, placing pressure on capital
engineers were tasked with designing lower cost, flexible and
and operating budgets. Since tariffs cannot be increased
more efficient wireless systems.
in proportion to demand, the cost per bit delivered must be
reduced!

Traffic

Voice Dominant
Revenues
& Traffic
De-Coupled

Revenues

Data Dominant

Time

FIG 1: Revenue/Traffic Relationship

1 Infonetics Research, Inc., Fixed and Mobile Subscribers annual worldwide market forecasts October 2008

2
Synchronizing IP Mobile Networks

The result… Long Term Evolution (LTE) and IEEE 802.16 WiMAX As we drive for “more performance and lower cost”, our
mobility. Many articles have been published on the relative innovation sometimes depends on secondary, enabling
benefits of LTE and WiMAX, but concisely stated, they offer technologies. This is especially true for IP transport platforms
important benefits for both consumers and operators: in the mobile architecture. Data can be transported through an
IP network without synchronization, but by the same token, IP
• Radio transmission rate increases, with download links up to networks cannot transport synchronization naturally (as was
200 Mbit/second (LTE). the case for SONET and SDH).
• Low round trip latency (10 milliseconds). What does this have to do with the IP backhaul? The answer
lies in the origin of the base station’s frequency reference. The
• Gains in spectral efficiency, allowing more simultaneous
handset (or UE) derives its frequency reference from the base
users for a given bandwidth. This includes Frequency Division
station’s air interface and operates at that recovered frequency.
Duplex (FDD) and Time Division Duplex (TDD) modes.
If the frequency difference between adjacent base stations is
• Underlying all IP transport networks. too big, the handset will not lock to the new BTS as the user
moves into the adjacent cell, and the call will be dropped.

Data rate increases, latency reduction, spectral The air interface stability allows the user equipment to hand off
efficiency and backhaul bandwidth collectively drive calls between cell towers without interruption, and is central to
the need for LTE. the Quality of Service (QoS).

What is important to us is the IP transport network. IP meets Without a stable air-interface frequency reference,
the backhaul investment drivers… increased bandwidth and wireless mobility cannot be supported.
reduced cost. Combining the IP foundation with protocols
that meet tomorrow’s air interface demands makes LTE a
compelling technology.

BTS2 drifts outside 50ppb window

F1 +∆F
} +/- 50 ppb

BTS1

Mobile cannot lock to BTS2

and call is dropped

F1
} +/- 50 ppb

BTS2

T1 T2 Time

FIG 2: Mobility Requirements

3
Synchronizing IP Mobile Networks

From the table below, it is apparent that all base stations The means to deliver the required time accuracy has
must support an air interface stability of 50ppb2 or better, traditionally been to install GPS receivers at every BTS, and
irrespective of the mobile protocols or technology generation. this will be discussed further as we explore next generation
At first glance, this may seem a daunting task, but base synchronization distribution.
stations traditionally sourced their frequency from the E1/
T1 backhaul (assuming they meet the synchronization masks
IEEE 1588 and SyncE are the two standards based
defined by the ITU-T G.823 or Telcordia GR.253). To meet the
methods for distributing frequency through packet
50ppb air interface requirement, the span line stability must
networks.
be approximately 15ppb.

Mobility Standard Frequency Time/Phase REBUILDING THE SYNCHRONIZATION CHAINS

CDMA2000 50 ppb Range: ‹3µs to ‹10µs While the transport vendors built packet network elements,
GSM 50 ppb the timing community worked on methods to deliver Timing
WCDMA 50 ppb over Packet cost-effectively. The obvious goals were to keep it
TD-SCDMA 50 ppb ±3µs inter-cell phase ∆ simple, cost-effective and predictable. Simple suggests using
LTE (FDD) 50 ppb the network to deliver time and frequency (at the physical
LTE (TDD) 50 ppb ±3µs inter-cell phase ∆
+
layer or in-band). There were many resultant methods to
LTE SFN-MBMS 50 ppb ±6.4µs inter-cell phase ∆
+
distribute precise time and frequency through the network, but
WiMAX (TDD) 50 ppb inter BTS Typically 1-1.5µs those selected by standards organizations were Synchronous
Femtocell (FDD) ~250 ppb Ethernet (SyncE) and IEEE 1588-2008 (also referred to as PTP
+ Standards being consolidated v2). Although not a packet technology, the use of GPS is also
considered.
TABLE 1: Mobility Air Interface Stability Needs
Global Positioning System (GPS)
To get to the heart of the matter, the E1/T1 links traditionally
Technology advances and the widespread adoption of GPS
provided the frequency reference to base stations. When
have resulted in cost reductions, allowing GPS timing receivers
replaced with Ethernet/IP, the link to the frequency source
to be deployed without the cost penalties of former versions.
is lost, and the timing chain is broken. This is the design
GPS is a high-performance solution, providing time, frequency
challenge for mobile backhaul planners.
and location information independently of the network,
and has traditionally been used to deliver microsecond
IP interrupts the frequency distribution needed requirements at base stations.
by base stations. Synchronization must now be
engineered into the network. One obstacle to GPS is service availability in metropolitan and
indoor installations resulting from weak and reflected signals.
The second concern is that GPS is not autonomous, and many
The air interface frequency requirement is ever-present, but
wireless operators prefer not to be dependent on widespread
there are cases where precise Time of Day is also required at
GPS. Finally, the deployment and maintenance costs of GPS
every base station.
are not trivial, particularly in urban applications.
•T
 he first is driven by technology. Time Division Duplex (TDD)
mode improves the spectral efficiency of the allocated
Political barriers to GPS are declining, but widespread
bandwidth. Phase synchronizing base stations eliminate
deployment demands a recovery contingency for GPS
inter-cell interference by ensuring that adjacent uplink and
outages.
downlink transmissions are coordinated. This can only be
achieved by synchronizing all the base stations to the same
Time of Day reference.
New Global Navigation Satellite Systems (GNSS), such
•T
 he second is driven by application. Applications such as SFN- as Galileo, are being deployed and will satisfy the same
MBMS transmit video frames from multiple base stations, application objectives as GPS. The political barriers to these
and accurately timing them allows the handset to rebuild systems may be lower in different geographies, but the urban
the video stream coherently. In addition, Location Based canyon, autonomy and cost challenges are the same.
Services that use Time of Arrival (TOA) to triangulate position
also need precise time synchronization.

2 Fractional frequency offset (∆f/f) of 5 x 10-8

4
Synchronizing IP Mobile Networks

Synchronous Ethernet (SyncE) There are many similarities to SDH/SONET distribution, and
this was by design. SyncE distributes frequency from a source
Synchronous Ethernet is a scheme that distributes
to a destination, and Primary Reference Clocks are needed
synchronization over the Ethernet physical layer without
at the source. SyncE accumulates jitter and wander over the
compromising the asynchronous data switching functions.
path just like SDH/SONET, and SSUs must be used to filter the
Based on the IEEE 802.3 standard for Ethernet, it is
jitter and wander per the ITU-T G.823 and Telcordia GR.253
synchronous at layer one with the higher layers being
guidelines.
asynchronous. There is no difference between a Synchronous
and Asynchronous switch in the way the data is handled. The
only difference is at the clock distribution and recovery layer. Planning for SyncE Jitter and Wander filtering follows
the ITU-T G.823 guidelines.
Asynchronous
TX TX switches receive data
at the incoming line
The 4.6ppm SyncE Equipment Clock (EEC) is the same one
RX RX rate, and in adherence
defined for SDH/SONET Equipment Clocks (SEC/SMC).
to IEEE 802.3, transmit
100 ppm data using a free-
running clock of TX Accurate TX TX Inaccurate TX

100ppm (poor stability RX RX RX RX

TX TX
in synchronization
terms, but suitable 100 ppm

RX RX for the switching

function). Ext. Sync 4.6 ppm

SyncE Switch Asynchronous Switch

4.6 ppm
SyncE switches, by
contrast, use a more FIG 4: Broken Synchronization Chain
accurate 4.6ppm
TX TX oscillator disciplined And just like SDH/SONET, SyncE only distributes frequency.
to the RX (incoming) SyncE does not distribute a Time of Day or phase reference.
RX RX
line rate. There is
In summary, SyncE has the advantage of being a deterministic
a Sync relationship
frequency distribution method that is independent of the data
between the RX and
flow. However, all the switches in a path must support SyncE,
4.6 ppm TX, allowing the
Ext. sync
and widespread adoption will be governed by the cost and ease
incoming clock to be
with which installed asynchronous switches can be upgraded.
propagated.

By adding an external sync port to the SyncE switch, a Stratum IEEE 1588-2008 (PTP v2)
1 reference can be introduced to, and distributed through, a The IEEE 1588-2008 protocol (also called Precision Time
packet network independently of the traffic. Protocol or PTP) is a standardized method to distribute
Unfortunately cascading synchronous and asynchronous accurate time and frequency over IP networks. The basis
(traditional) switches will interrupt the originating (and of operation is that packets carry timestamp information
accurate) sync flow, making SyncE a point-to-point frequency between a master (sometime called a server) and slave
delivery method. This means that cascaded synchronous and (sometimes called a client), and the slaves use the timestamps
asynchronous switches will not deliver the source frequency to synchronize to the master. Bidirectional flows eliminate the
through the network. The 100ppm reference will be substituted round trip delay to enhance the accuracy. Frequency can, in
in the asynchronous switch, breaking the timing chain. turn, be recovered from the disciplined Time of Day clock.

IEEE 1588 is an in-band distribution system,

eliminating the need for a dedicated timing plane.
Cascaded SyncE and asynchronous switches will
transport data, but cannot not distribute the source
frequency. The 1588 timing and management messages are transported
in-band with the mainstream traffic, eliminating the need for a
dedicated timing plane.

5
Synchronizing IP Mobile Networks

IEEE 1588 was initially developed for industrial automation Network Time Protocol (NTP)
over Local Area Networks, but a second version, tailored for
NTP is the most popular protocol for distributing time
constrained telecommunication environments, was published
over LANs and WANs, and when looking at IEEE 1588, the
in 2008. The pending ITU-T G.8265 Telecom Profile simplifies
similarities cannot be ignored. The reasons NTP is not used in
the diversity of configurable parameters needed to support
high performance applications are the low transaction rate (1
WAN’s, improving the protocol’s interoperability.
per 64 seconds) and the software nature of the solution. NTP
What makes IEEE 1588 very attractive is the microsecond packets pass through the Ethernet PHY and Media Access
accuracy (and associated 1ppb frequency stability) that can be Control (MAC) layers like any other packet. CPU processing for
realized over managed Ethernet. This allows PTP platforms to the timing requirements is not addressed until the software
support a wider range of applications than any other solution, stack is fully processed. The NTP packets are therefore delayed
addressing both the FDD and TDD modes of LTE. by an indefinite time depending on the operating system
latency, limiting the assured accuracy of the solution.

NTP will continue to be used in LTE for Call Detail Records and
IEEE 1588 over Ethernet designed for QoS can deliver
billing, just as it was in earlier mobile generations. NTP time
microsecond accuracy.
synchronization is also used by switch and router elements
in IP networks to monitor performance and optimize routing
Being packet based, IEEE 1588 is sensitive to the network tables (with one way latency methods being notable). Carrier-
behavior and the accuracy depends on the clock recovery class NTP servers distributed at mobile switching offices
algorithm and the packet jitter (also called Packet Delay remains a fundamental component of any wireless system.
Variation or PDV). In general, meeting the frequency
requirement is moderately easy, but phase synchronization is Call Data Record management and routing tables are
more sensitive to PDV and requires added planning. still dependent on NTP services.
Fortunately the protocol designers foresaw this, and included
on-path support in the specification. On-path support consists
of Transparent and Boundary clocks that reduce the packet It is worth mentioning that a Femtocell assumes the user
jitter, improving performance over long hop counts. On-path to be essentially stationery, and the 50ppb air interface for
support will be discussed in more detail later. this technology was relaxed to ~250ppb. As a result, some
Femtocell FDD solutions have embedded NTP clients that
recover frequency from the Time of Day.

Embedded
Slave

Grandmaster
1588
(Server) 1588 Packet Flow

External Slave

FIG 5: IEEE 1588 Architecture

6
Synchronizing IP Mobile Networks

IEEE 1588-2008 (PTP) & SyncE In Perspective ENGINEERING FOR COST, PERFORMANCE & SIMPLICITY
How are IEEE 1588 (PTP) and SyncE similar? How are they Selecting A Synchronization Strategy
different? And what does this mean to you?
Assuming a standards-based timing solution is chosen,
Both SyncE and IEEE 1588 are standards-based methods to synchronization will be based on either Synchronous Ethernet
transport frequency (and time in the case of PTP) through the (SyncE) or IEEE 1588 (PTP), or the two together. Both these
network to heal the broken synchronization chains. Where they technologies have distinct advantages and disadvantages over
differ is in implementation and function. each other. SyncE is deterministic and the performance is
independent of the traffic. PTP can function over asynchronous
switches, and distributes frequency and time with the data
SyncE and IEEE 1588 are complementary technologies traffic.
and can co-exist in the network.
Selecting one over the other is an economic reality as much
as a technical reality, but setting aside the cost of upgrading
The table below summarizes key differences between the the switches to SyncE, deployments with the following
two, but in essence SyncE is a conscious decision to add the characteristics are likely to be based on IEEE 1588:
feature to every switch between the source and the destination.
• Need time/phase such as LTE TDD and SFN-MBMS
To recap, cascaded synchronous and asynchronous switches
will not transport synchronization (even though they can • Do not have end-to-end SyncE switches (Ethernet)
route data). IEEE 1588 is largely independent of the transport • Function over diverse transports (microwave, Ethernet,
elements; largely because boundary and transparent clocks SHDSL..)
may be embedded in switching elements, but this is not a pre-
requisite. This allows PTP networks to be built independently • Share leased network sections, unless SyncE or a SDH/SyncE
over diverse transport systems. hybrid can be assured over the full path

Attribute IEEE 1588 SyncE Selecting a Sync strategy is an economic and technical
Capability Frequency, Phase, Time Frequency reality, determined in part by the mobility protocols.
Layer Ethernet/UDP Physical
Distribution In-band 1588 Packets Physical layer
Keeping these limitations in mind, an operator planning to
Schema Point to multi-point Point to point deploy LTE (TDD mode), or SFN-MBMS would typically select
Transport Media Ethernet, SHDSL, Native Ethernet, IEEE 1588 as the primary synchronization method. SyncE would
Microwave Other in development only address frequency synchronization requirements. This
Interoperability Standards-based Standards-based SyncE does not preclude using a combination of SyncE or E1/T1 to
Grandmaster & slave switches only synchronize legacy base stations.
Sensitivity Packet Jitter / Bandwidth Asynchronous switches
utilization For LTE specifically an IEEE 1588 strategy requires:
Standards IEEE 1588, ITU G.8261/3/5 G.8261/2/4
• Carrier-class Grandmaster clocks installed at strategic
locations
TABLE 2: IEEE 1588 / SyncE Comparison
• A managed Ethernet network designed for QoS
It is important to note that SyncE and 1588 are not mutually
exclusive. SyncE functions at layer one independently of the • Base stations with integrated 1588 slaves or with external
traffic. IEEE 1588 functions at higher layers (UDP/Ethernet) sync ports, including 1PPS/ ToD for phase synchronization.
independently of the transmission rate. It is entirely possible By definition LTE base stations will not have external sync
that both technologies could be used on the same path - SyncE ports, but these may be available for early deployments.
for frequency and PTP for time/phase. In reality, slaves that
support both can converge on an accurate time very quickly by
using the SyncE frequency to discipline the local oscillator.

SyncE in it’s current form cannot fulfill the phase

requirements of LTE TDD or SFN-MBMS.

7
Synchronizing IP Mobile Networks

Defining the Sync Objectives some operators are more cautious, and have selected the
synchronization mask as the minimum performance to be met
The air interface requirement defines the stability that must be
at their BTS interfaces.
maintained between the base station and the handset, typically
over 1 time-slot. For LTE TDD, the frequency stability must be How do we reconcile IEEE 1588 to the TDM mask? Packet
50ppb, and the phase alignment ±3μS. The air interface needs networks don’t transport synchronization, but the TDM
must not be confused with the synchronization that needs to be interfaces are still synchronous. Ultimately, the 1588 packets
delivered to the base station (on the backhaul interface). must be converted to an analog source, and the quality of
the recovered clock at the BTS must pass the MTIE test (as
measured with a traditional SDH analyzer). If the measured
Sync Masks are used to define goals and determine result is under the mask, the 1588 system has worked. If not, it
whether those objectives are being met. has failed and some level of re-engineering is required.

We also know that when TDM network synchronization Selecting the short and long Sync terms goals is the
objectives are met, 50ppb or better can be maintained on the first step in engineering an all IP network solution.
air interface. The TDM network performance objectives are
defined in ITU-T G.823 and G.824 in the form of masks.
Because TDM masks only define the performance over ~28
Masks define the MTIE limits (quality of synchronization) that a
hours, it is not uncommon for operators to define a long term
network should conform to over 100,000 seconds (~28 hours).
stability requirement. A long term stability of 15ppb is typically
Depending on where the clock signal is measured, there are
specified and adequate for mobile base stations.
two masks, the traffic and the synchronization mask. If the
clock signal is being measured at an “end point”, it must Finally, for TDD applications, the phase alignment (usually
comply with the traffic mask. measured on a rising or falling edge) of the source
(Grandmaster Clock) and slave must be less than the specified
If it is measured along the synchronization chain, it should
value, ±3μS for LTE with TDD support. This is also more
comply with the more demanding synchronization mask.
accurate than the ±6.4μS for SFN-MBMS, addressing that
The BTS is an end point, and by definition, a carrier-grade application well.
service can be delivered if the network jitter is lower than
(below) the traffic mask. This is also true in practice, but

8
Synchronizing IP Mobile Networks

It should be apparent that the Grandmaster has not been

The 1588 slave’s PDV rejection capability governs the mentioned, and for good reason. The slave design determines
quality of the recovered clock. the achievable performance. High-cost oscillators are not
an option in base stations already under pricing pressure, so
PDV (or the elimination of PDV) becomes the overwhelming
The quality of the synchronization delivered to the BTS for an consideration.
LTE TDD service would typically be defined as:

• Meet the Traffic mask as defined in ITU-T G.823 over the first
Eliminating Packet Delay Variation (PDV) has the
100,000 seconds
largest effect on 1588 clock recovery.
• Maintain a long term average frequency stability of 15ppb
(1.5x10-8) or better
PDV is governed by the bandwidth utilization, upstream/
• Maintain the phase difference between the source and the
downstream path symmetry, and the number of hops (store
slave of less than/equal to ±3 microseconds.
and forward buffer devices) between the Grandmaster and the
What Affects IEEE 1588 Accuracy? slave. There is not much that can be done about variations in
bandwidth use, but the 1588 packet PDV can be reduced in two
What affects IEEE 1588 clock recovery, and how it is mitigated? ways. First, by filtering PDV in the clock recovery algorithm
First, let’s set aside the attributes that don’t affect 1588. Packet and second, by reducing the number of switch and routing
latency, packet loss and packet errors do not significantly elements between the Grandmaster and the Slave during the
impact the clock recovery protocols. network planning stage.
The quality of the recovered time and frequency does depend On-Path Support: Pushing the Limits
on:
As described earlier, one method used to reduce PDV is to
1) The stability of the local (slave) oscillator limit the number of switching and routing elements between
the Grandmaster and the slave. Where this is not possible or
2) T
 he Packet Delay Variation3 (PDV) of the IEEE 1588 packets
desirable, a second method is to minimize the effect of switch
3) The quality of the servo-loop algorithm in the slave. transit delays (the time that 1588 packets spend in the switch).
To do that, the IEEE 1588 standard includes two new clock
types, the Transparent Clock and the Boundary Clock.

FIG 6: Performance Masks

3 Packet Delay Variation is similar to packet jitter, but includes the low frequency wander components.

9
Synchronizing IP Mobile Networks

Boundary Clock (BC) Transparent Clocks (TC)

The Boundary Clock was defined in both version 1 and 2 of the Transparent Clocks were added to the second version of the
IEEE 1588 standards, and is a special form of network switch 1588 standard to support cascaded topologies. Like boundary
or router. From the network perspective, a boundary clock is a clocks, transparent clocks are a special form of network switch
normal switch element, passing network traffic seamlessly. or router. Unlike boundary clocks, timing is not recreated, but
the residence time (time that 1588 packets spend in a switch)
is measured and reported to the slave. The transparent clock
The burden of holdover and traceability belongs to the does this by inserting
Boundary Clock when it is used. the residence time
Residence Time =
into the correction Packet Egress — Packet Arrival
field in the sync,
The clock plane is very different. The boundary clock has an delay-request and
embedded 1588 slave that synchronizes the internal clock follow-up messages.
PTP Packet PTP Packet

from an upstream IEEE 1588 flow. It also has a Grandmaster

function providing 1588 flows to downstream slaves within the Two transparent
Arrival Egress
network. One port of clock modes are Time Time
the BC is a designated defined, peer-to-
Slave
1588 slave port, and the peer and the more
remaining ports are 1588 widely advocated
master ports. end-to-end version, Transparent Clock
accumulating the
The master/slave residence time of all FIG 8: Transparent Clock
functions in the boundary the Transparent Clocks
clock terminate a sync GMC (switches) in the correction field. This allows the highly variable
flow4, and generate new residence periods to be removed from the round trip delay
flow(s), thus reducing Boundary Clock
calculation, reducing the PDV considerably.
the PDV that has been
accumulated. Reducing FIG 7: Boundary Clock We know that boundary and transparent clocks reduce the 1588
the PDV is important, message PDV, and can increase the achievable performance for
but in telecommunication applications, the Grandmaster is a a fixed path length, or increase the path length for a fixed goal.
carrier-class device with high stability holdover oscillators, But is it necessary?
management and resilience. The ability of the boundary
clock to provide the same capability will depend on the 1588 As our frequency and phase results show, boundary and
implementation by switch vendors. transparent clocks are not necessary for FDD mobile systems
(assuming the number of links between the MME and eNodeB
is typical). Depending on the slave clock performance, TDD
Transparent clocks preserve the integrity of the sync applications with phase requirements can also be met without
flow, but are limited in encrypted networks. on-path support.

On-path support is not always necessary for IEEE

1588 applications, particularly for frequency
distribution.

4 Sync, Follow_Up, Delay_Req, and Delay_Resp messages.

10
Synchronizing IP Mobile Networks

When functioning over a managed, symmetric metro Gigabit The Grandmaster capacity can be checked by dividing
network, engineered for QoS, with 5 to 10 switching elements the maximum transaction rate by the number of served
and controlled load, the Symmetricom slave can typically slaves and the number of transactions per second. If the
deliver 1ppb frequency stability and ±1 microsecond of phase demand exceeds the capacity, additional modules (or even
accuracy5. These values will vary as a function of traffic load, Grandmasters) can be added to share the load. Served slaves
switching jitter and path asymmetry. Having said that, a well- is the sum of the slaves that are expected to obtain sync flows
designed slave will benefit from boundary and transparent from a Grandmaster routinely, and those that could request
clocks. the service if their designated GMC clock fails.

High quality clock recovery algorithms have

Probing a representative sample of backhaul links
demonstrated the extent to which PDV filtering is
provides insight into the PTP capability of path.
effective.

In a world characterized by rapid change, the ITU-T working • Second, select a good quality 1588 slave that has been
group’s decision to exclude on-path support from the Telecom designed to work over the transport being used (e.g. Native
profile is a welcome step in reducing early deployment risk and Ethernet, Ethernet over SDH, Microwave and SHDSL).
complexity. But to ensure that your investment is protected, the Embedded slaves will become commonplace, but stand-
slaves that you buy today must process the correction field in the alone slaves allow early deployment risks to be reduced, as
PTP messages. This will ensure that the 1588 slaves can take well as serving the large installed base of legacy devices that
advantage of on-path support elements if added in the future. can only accept the DS1/E1 Sync references.

Deployment Made Easy • The next step is locating the Grandmasters based on the
Simple deployment guidelines ensure the best 1588 results: message rate of the Grandmaster and the PDV that each
slave will experience. Cost considerations encourage
• First select a high performance, high-reliability Grandmaster fewer Grandmasters functioning through more switches
clock that has capacity for the number of slaves expected, and routers. Robustness calls for more Grandmasters,
now and in the future. A system that is scalable to 1,000 with fewer links between the Grandmaster and slave.
slaves, at a transaction rate of 64 per slave is typical. Telecommunication networks favor robustness for Quality of
Service.

Validation of the slave’s ability to use transparent Experience has shown that the best place to locate the
switch delays is key to a future-proof solution. Grandmaster is at the RNC. LTE does not have a dedicated
RNC, but it is reasonable to assume that the MME will be

5 The solution referred to is the Symmetricom slave. Not all slaves have the same performance and results will vary.

11
Synchronizing IP Mobile Networks

located at the same locations as the RNC or aggregation

nodes in the foreseeable future. The number of hops between Standards based solutions are interoperable and will
the MME and eNodeB will typically be about five. In addition, grow with tomorrow’s needs.
the Ethernet path is likely to be managed, and a reasonably
low PDV can therefore be expected. deterministic result. The second is the IEEE 1588 precision
time protocol that functions over higher layers. IEEE 1588
• A representative sample of links should be probed to validate
is more versatile, but is sensitive to network behavior. It is
the Grandmaster location assumptions. Probing the network
important to remember that SyncE and IEEE 1588 are not
in this case refers to characterization of the PDV using a 1588
mutually exclusive and can be used together in some networks.
tool, and not the measurement of data packet latency or loss.
Both methods are valuable to network planners for FDD
• And finally, the 1588 elements should be managed for
mobility protocols, but IEEE 1588 is the preferred method for
faults, availability and synchronization performance. This is
TDD and SFN-MBMS applications that require frequency and
especially true for boundary and transparent clocks if they
phase.
are used. If working correctly, on-path support can improve
the result, but if it does not function correctly, the impact on On-path support elements (boundary and transparent clocks)
clock recovery can be devastating. are not necessary in most FDD applications, and depending on
the client performance and network design, on-path support
Refer to the Symmetricom document “Deployment of Precision
may not be necessary for TDD applications. An improvement
Time Protocol” for a more detailed IEEE 1588 planning guide.
in the quality of the extracted clock can be expected if on-path
elements are used and managed from a time perspective.
CONCLUSION Minimizing the use of on-path support in the short term helps
Although revenue is derived from voice services today, mobile reduce early deployment risks.
broadband is the new business for service providers. To be
Ultimately, deploying a standards-based synchronization
viable, high bandwidth must be offered for less revenue, and
solution for next generation mobile networks is key to
be augmented with value added services. LTE and IP mobile
investment protection, as well as meeting long term
backhaul are two significant steps in that direction.
requirements. A universal synchronization platform is most
The economics for IP backhaul is compelling, but IP creates cost-effective, more robust and plays a large part in the
synchronization discontinuities that must be consciously optimization of the network.
addressed in the network. The international community has
To find out more about next generation synchronization,
accepted two new methods to distribute synchronization
contact your local Symmetricom representative, or e-mail us at
through packet switched networks. The first is Synchronous
[email protected]
Ethernet, that transports frequency at Layer 1 for a

Probing a representative sample of backhaul links

provides insight into the PTP capability of the path.

12
Synchronizing IP Mobile Networks

ABBREVIATIONS USED NTP......... Network Time Protocol

BC............ Boundary Clock PDV.......... Packet delay Variation (synonymous with Packet
Jitter)
BTS.......... Base Station Transceiver
PTP.......... Precision Time Protocol
CES.......... Circuit Emulation Service
PWE......... Pseudo-Wire Emulation
DSL.......... Digital Subscriber Line
SEC.......... SDH Equipment Clock
EEC.......... Ethernet Equipment Clock
SDH......... Synchronous Digital Hierarchy
ETSI......... European Telecommunications Standards Institute
SFN......... Single Frequency Network
FDD......... Frequency Division Duplex
SONET..... Synchronous Optical Networking
GigE......... Gigabit Ethernet
SMC......... SONET Minimum Clock
GMC........ Grandmaster Clock
SyncE...... Synchronous Ethernet
GNSS....... Global Navigation Satellite System
TC............ Transparent Clock
GPS......... Global Positioning System
TDD......... Time Division Duplex
IEEE......... Institute of Electrical and Electronic Engineers
TDM......... Time Division Multiplex
IP............. Internet Protocol
TOA.......... Time Of Arrival
ITU-T....... International Telecommunication Union
UDP......... User Datagram Protocol
LBS.......... Location Based Service
UE............ User Equipment
LTE.......... Long Term Evolution
UTC......... Universal Coordinated Time
MBMS...... Multimedia Broadcast Multicast Service
WAN........ Wide Area Network
MTIE........ Maximum Time Interval

NGN........ Next Generation Network

© Copyright 2010 Symmetricom, Inc. All rights reserved. Symmetricom and the Symmetricom logo are registered trademarks of Symmetricom, Inc.
All other trademarks are the property of their respective companies. All specifications subject to change without notice. 04-27-10

www.symmetricom.com Phone: +1 408 428 7907; toll-free: +1 888 FOR SYMM | Fax: 408-516-9597 | E-mail: [email protected]

13