+ All Categories
Home > Documents > Synchronisation 120610202311 Phpapp02

Synchronisation 120610202311 Phpapp02

Date post: 26-Oct-2014
Category:
Upload: animshakya
View: 164 times
Download: 16 times
Share this document with a friend
Popular Tags:
247
Synchronisation Training Presenter: Colin Argent CECP On behalf of: Horsebridge Network Systems Ltd.
Transcript
Page 1: Synchronisation 120610202311 Phpapp02

Synchronisation Training

Presenter:

Colin Argent CECP

On behalf of:

Horsebridge

Network Systems Ltd.

Page 2: Synchronisation 120610202311 Phpapp02
Page 3: Synchronisation 120610202311 Phpapp02

Contents

Day 1

Introductions

Chapter 01 - Time, Frequency and Phase

Chapter 02 - Synchronisation

Chapter 03 - Sync Distribution layer

Chapter 04 - Sync Distribution Layer – Clocks

Chapter 05 - Sync Distribution Layer – Equipment

Day 2

Chapter 06 - Sync Distribution Layer – Links

Chapter 07 - SDH Network Topology

Chapter 08 - Synchronisation Network Architecture

Chapter 09 - Synchronisation Standards

Chapter 10 - Sync E (Synchronous Ethernet)

Day 3

Chapter 11 - IEEE 1588v2 - PTP (Precision Timing Protocol)

Chapter 12 - Boundary and Transparent Clocks

Chapter 13 - Clock Measurements

Chapter 14 - CALNEX Testing

On-screen demo

Page 4: Synchronisation 120610202311 Phpapp02
Page 5: Synchronisation 120610202311 Phpapp02

1

Chapter 01

Time, Frequency and Phase

Page 6: Synchronisation 120610202311 Phpapp02

2

Time and Frequency

Page 7: Synchronisation 120610202311 Phpapp02

3

Time, Time Scales & Dates

• There are Two meanings of the word time:

- The date of an event on a time scale

- As a time interval between two events

• Time scales :

- A time scale is defined by :

A time unit

A time origin

• Dates :

A date is a number of units on the time scale

Page 8: Synchronisation 120610202311 Phpapp02

4

The Second

• The second is the time unit of the International

System of Units (SI)

• It just so happens that the second is relative to that of Caesium atom transition

• This means that modern time accuracy is based around time derived from a Caesium device

• This is commonly known as an Atomic Clock

Page 9: Synchronisation 120610202311 Phpapp02

5

Clocks

• A clock consists of:

- A period which can be observed e.g. Secs, Mins & Hrs- A counter which counts the number of periods- A means for setting the counter to a preset value- A display of the registered count

OscillatorOscillator CounterCounter DisplayDisplayu(t)u(t) n(t)n(t) T(t)T(t)

startstartNN00

Page 10: Synchronisation 120610202311 Phpapp02

6

Atomic Time Scales

• Origin of Atomic Time Scales :

1 January 1958, at 0hr 0min 0sec - UT2

• International Atomic Time (TAI) :

Based on atomic time scales and implemented by a network of atomic clocks located all over the earth and operated by the Bureau International de l’Heure (BIH) in Paris

• Coordinated Universal Time (UTC) :

Timescale based on the time unit of TAI, transmitted on air from the GPS system

• In 1967 a new SI definition of a second was created based on the

radiation from the caesium-133 atom

• It is correctly defined as "the duration of 9,192,631,770 periods of the radiation corresponding to the transition between two hyperfine levels of the ground state of the caesium-133 atom"

Page 11: Synchronisation 120610202311 Phpapp02

7

BIPM

• The International Bureau of Weights and Measures

(Bureau International des Poids et Mesures)

• The realization and dissemination of the international time scales is the responsibility of the Time Section of the BIPM. International Atomic Time (TAI) is the uniform time scale; it is kept as close as possible to the second of the SI. Coordinated Universal Time (UTC) is an atomic time scale derived from TAI, to provide a reference scale in step with the irregular rotation of the earth

• Local realizations of UTC exist at the national time laboratories. These laboratories participate in the calculation of the international time scales by sending their clock data to the BIPM. Most of them are equipped with commercial caesium beams that provide a practical realization of the second sufficiently accurate for most applications. More accurate caesium standards exist in a small number of laboratories; for them, the uncertainties are estimated to be a few parts in 1015. New developments in clocks using trapped or cooled atoms or ions are leading to improvements well beyond this

• The atomic time scales TAI and UTC are disseminated monthly through the BIPM Circular T. The Annual Report of the BIPM Time Section provides all relevant information, data and results for the year previous to its publication. Reports on time-transfer techniques are also issued regularly

• Other activities related to the time scales are developed in the section; these contribute to improving the calculation algorithms and increasing knowledge about time transfer techniques

Page 12: Synchronisation 120610202311 Phpapp02

8

Frequency

Definition:• the number of occurrences within a given time period

Unit:• Frequency is expressed in [ Hertz = 1cycle/second ]

0011

00

11

amplitudeamplitude

timetime1x101x10--77 2x102x10--77 3x103x10--77 4x104x10--77 5x105x10--77 6x106x10--77

Page 13: Synchronisation 120610202311 Phpapp02

9

Phase

• Phase is the difference in time relationship between two same frequency waveforms

• It is usually measured in degrees

• It can be a comparison with an earlier instance of the same

waveform - known as Differential Phase

90°

180°

270°

360°

90°PhaseShift 270 Differential Phase Shift°

Page 14: Synchronisation 120610202311 Phpapp02

10

Round Up

• What are the 2 main time standards?

• What is periodic value of a Caesium

atom?

• What are the 3 main components of a

clock?

• What unit is frequency expressed as?

Page 15: Synchronisation 120610202311 Phpapp02

1

Chapter 02

Synchronisation

Page 16: Synchronisation 120610202311 Phpapp02

2

Introduction to Synchronisation

Page 17: Synchronisation 120610202311 Phpapp02

3

What is Synchronisation ?

Definition of Synchronisation

• The timing of all nodes within digital networks to a

common highly accurate and stable clocking source

• To ensure they all use the same data rates to transmit and receive information

Page 18: Synchronisation 120610202311 Phpapp02

4

Why do we need Synchronisation ?

• If synchronisation is not used node clocks operate

asynchronously and the their transmit and receive data

rates would be different.

• This would cause slips or pointer adjustments to occur frequently, seriously degrading the quality of services

transported by the network

Page 19: Synchronisation 120610202311 Phpapp02

5

Types of Synchronisation

3 types of Synchronisation:

1 - Frequency synchronisation

2 - Phase synchronisation

3 - Time synchronisation

Page 20: Synchronisation 120610202311 Phpapp02

6

Frequency Synchronisation

System A

Clock signalof system A

Clock signalof system B

System B

t

t

Page 21: Synchronisation 120610202311 Phpapp02

7

Frequency Synchronisation

If Clock 1 & 2 are at different speeds then slips between equipments will occur at the input buffers

Data

Clock 1 Clock 2

Data

Transmission Link

Data

• Transmission data is ‘loaded’ and ‘recovered’ from transmission lines by data processing circuits

• These use a clock edge to denote the transition of one bit to the next

Page 22: Synchronisation 120610202311 Phpapp02

8

Frequency Sync – Too Fast

• If the frequency of Clock 1 is higher than that of Clock 2 -

– Data will be clocked into the transmission link at a higher rate than it is clocked out

– The result will be that data is lost at the receiving end

Data

Clock 1 Clock 2

Data

Transmission Link

Data

Page 23: Synchronisation 120610202311 Phpapp02

9

Frequency Sync - Too Slow

Data

Clock 1 Clock 2

Data

Transmission Link

Data

• If the frequency of Clock 1 is lower than that of Clock 2 -

– Data will be clocked into the transmission link at a lower rate than it is clocked out

– The result will be that data is lost at the receiving end

Page 24: Synchronisation 120610202311 Phpapp02

10

Phase Synchronisation

System A System B

Clock signalof system A

Clock signalof system B

t

t

Page 25: Synchronisation 120610202311 Phpapp02

11

Time synchronisation

System A

t

t

Time signal of system A

Time signal of system B

System B

14/01/0008:34:56

14/01/0008:34:57

14/01/0008:34:55

14/01/0008:34:55

14/01/0008:34:56

14/01/0008:34:57

Page 26: Synchronisation 120610202311 Phpapp02

12

N'wks/Services that require Sync

• Public Switched Telephone Networks

• SONET and SDH transport networks

• Cellular mobile telecom networks - GSM, UMTS etc

• Location Services over Mobile Networks - E911, GSM 03.71, etc

• Ground stations of satellite networks

• Digital Audio Broadcasting (DAB)

• Digital Video Broadcasting (DVB)

• Time distribution for charging & event time stamping

• Next Generation Networks – Wi-Max, MPLS

Page 27: Synchronisation 120610202311 Phpapp02

13

Slips

• Slips are the main consequence of poor synchronisation within SDH / SONET networks

• Slips can have a detrimental effect on quality of service

Page 28: Synchronisation 120610202311 Phpapp02

14

What Are Slips ?

• A slip occurs when an equipment input buffer

over or underflows due to differences in timing

Incoming data rate

Outgoing data rate

SlipSlip

• This results in information being lost

Page 29: Synchronisation 120610202311 Phpapp02

15

Mobile Networks – 2G & 3G

BTS

BTS

BTS

BTS

Successful handover requires synchronisation

between base transceiver stations (BTS)

Page 30: Synchronisation 120610202311 Phpapp02

16

Cellular Mobile Telecom Networks

Radio carrier frequencies must be synchronised

precisely in order to prevent cross-talk

Radio spectrum

Frequency

Page 31: Synchronisation 120610202311 Phpapp02

17

Effects of Frequency Errors

Slip – PDH EnvironmentFIFO buffer overflows and dumps its contents

Pointer Adjustment – SDH Environment

2Mbit/s transmission VC12 pointer adjustments cause phase hits of 3.47µs 34 & 140 Mbit/s transmission

Pointer Movement at SDH-PDH boundaries PDH takes sync from 2 Mbit/s and needs excellent phase performance

Pointer adjustments will create phase hits

Page 32: Synchronisation 120610202311 Phpapp02

18

Services effected by slips

Voice

Uncompressed - only 5% of slips lead to clicks

Compressed - a slip will cause an audible click

Fax

A slip can wipe out several lines

Modem

A slip can cause several seconds of drop out

Compressed video

A slip can wipe out several lines

More slips can freeze frames for several seconds

Encrypted/compressed data protocol

Slips will reduce transmission throughput

Cellular

Dropped calls and poor cell handover

Page 33: Synchronisation 120610202311 Phpapp02

19

Implications on Performance

• Poor network synchronisation means that network performance is not optimized, quality of service is reducedand customers are lost

• Implementation of synchronisation in network design enhances Quality of Service for your customers

• Protection of your customers traffic means confidence, loyalty and ultimately improved business relationships

Page 34: Synchronisation 120610202311 Phpapp02

20

Round Up

• So why do we need synchronisation?

• What are the 3 main types of synchronisation?

• What is a slip?

• What can slips cause?

Page 35: Synchronisation 120610202311 Phpapp02

1

Chapter 03

Sync Distribution

Layer

Page 36: Synchronisation 120610202311 Phpapp02

2

Network synchronisation

• The objective of network synchronisation is to ensure that all the telecommunication systems use the same transmit

and receive data rates to avoid slips

• So the clocks in the telecommunication systems must be

synchronised to the same master network clock, or synchronised to a number of very closely matched master

clocks

Page 37: Synchronisation 120610202311 Phpapp02

3

Distribution Layer function

• To generate a primary reference timing signal

• To distribute timing signals from the primary reference source to nodes and equipment

• To provide protection against failures in the generation

and distribution of timing

Page 38: Synchronisation 120610202311 Phpapp02

4

Logical Synchronisation Network

MasterClock

MasterClock

Telecom equipment clocks

Page 39: Synchronisation 120610202311 Phpapp02

5

Master-Slave Mechanism

MasterMaster SlaveSlave

PRC 1

Data

Transmission Link

Data + Clock

• The clock is injected into the master unit

• The slave unit locks to the incoming clock rate and is now synchronised to the master

• No slips occur between these elements

Page 40: Synchronisation 120610202311 Phpapp02

6

Physical synchronisation network

• Not every system in the network can have a direct

connection to the master network clock

• Therefore the telecommunication systems are

synchronised in chains or trees

• Each system clock is the master clock of the subordinate

system clocks slaved to it

• The slave system continually adjusts its own clock to the

incoming signal

• Therefore both the master and slave systems have the

same transmit and receive rates

• There are no slips (WE HOPE!!!!)

Page 41: Synchronisation 120610202311 Phpapp02

7

Clock types

The clock elements of the synchronisation distribution layer are categorised in three ways:

1. Primary Reference Clock (PRC)

- This is defined by ITU-T recommendation G.811

2. Synchronisation Supply Unit (SSU)

- The purpose of these elements is to provide filtering and regeneration

- Defined by ITU-T recommendation G.812

3. SDH Equipment Clock (SEC)- These devices have an internal SDH Equipment

Clock (SEC) that is normally synchronised to a traffic or an external timing input signal

- Defined by ITU-T recommendation G.813

Page 42: Synchronisation 120610202311 Phpapp02

8

Master-slave principle

= slave

PRC

SEC

SSU

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

= master

= slave

= slave

= slave

= slave

= slave

SSU

SSU SSU SSU

Page 43: Synchronisation 120610202311 Phpapp02

9

Sync Distribution (SD) Trails - 1

• The clock frequency along an SD trail is the SAME as the head-end, i.e. PRC, SSU or SEC

• SD trails can be very long or very short

• There can be hundreds of SD trails in a synchronisation network

Page 44: Synchronisation 120610202311 Phpapp02

10

Sync Distribution (SD) Trails - 2

• There are several types of SD trails:

– PRC SD trail - when the head-end is a PRC

– SSU SD trail - when the head-end is a SSU

– SEC SD trail - when the head-end is a SEC

• SSU and SEC SD trails are created only when the PRC

SD trail is broken

Page 45: Synchronisation 120610202311 Phpapp02

11

Physical synchronisation network

PRC SD trailPRC SD trail clock qualityclock quality

traceable backtraceable back

to the PRCto the PRC

PRCPRCPRC

SSU

clock qualityclock quality

traceable backtraceable back

to the SECto the SEC

SEC SD trailSEC SD trail SSU SD trailSSU SD trail

clock qualityclock quality

traceable backtraceable back

to the SSUto the SSU

SEC

Physical synchronisation network

Page 46: Synchronisation 120610202311 Phpapp02

12

Slip Rate due to Freq Deviation

For 2 Mbit/s signals, frame duration = 125

microseconds:

– 10-11 = 1 slip in 4.8 months PRC G.811

– 10-10 = 1 slip in 14.5 days SSU G.812

– 10-9 = 1 slip in 1.45 days SSU G.812

– 10-8 = 6.9 slips per day SEC G.813

– 10-7 = 2.9 slips per hour SEC G.813

– 10-6 = 28.8 slips per hour SEC G.813

– 10-5 = 4.8 slips per minute SEC G.813

Page 47: Synchronisation 120610202311 Phpapp02

13

Calculating Slip Rates

Normal Calculation for Slip Rates per observation period

is:

Slip Rate = F x T x D

L

F= Frequency Offset

T= Observation Time

D= Data Rate

L= Frame Length

Page 48: Synchronisation 120610202311 Phpapp02

14

Calculating Slip Rates – Exercise!

Calculate the Slip Rate for the following conditions:

F= Frequency Offset = 1 x 10E-08

T= Observation Time = 86400 Seconds (1 day)

D= Data Rate = 2.048Mbps x 10E+06

L= Frame Length = 256

What clock quality would this slip rate equal?

Page 49: Synchronisation 120610202311 Phpapp02

15

Calculating Slip Rates – Result

The Solution:

F (1x10E-08) x T(86400) x D(2.048x10E+06)

L(256)

= 6.912 or 7 slips per day!

F= Frequency Offset = 1 x 10E-08

T= Observation Time = 86400 Seconds

D= Data Rate = 2.048E+06

L= Frame Length = 256

So 7 slips per day is G.813 or SEC in HOLDOVER

Page 50: Synchronisation 120610202311 Phpapp02

16

Sync Distribution (SD) Trails - 3

• The synchronisation distribution trails are not perfect:

what comes out is not exactly what went in

• This is due to:

- Equipment and cables generate jitter and wander

- Excessive jitter or wander causes slips

- Equipment or cables can fail

Page 51: Synchronisation 120610202311 Phpapp02

17

Jitter

• Pattern, or pattern-dependent, jitter is sometimes called "flanging". This

type of jitter is not random; it generally results from sub-harmonics

• Viewed in the time domain, this type of jitter appears as multiple modes.

Pattern jitter is deterministic jitter that can be attributed to a unique

source. All other jitter is stochastic (random) in nature

Jitter can be quantitatively expressed in the following ways:

– In unit intervals (UIs). One UI is one cycle of the clock frequency. Jitter

expressed in UIs describes the magnitude of the jitter as a decimal fraction of

one UI

– In degrees. Jitter expressed in degrees describes the magnitude of the jitter in units of degree for which one cycle equals 360°

– In absolute time. Jitter expressed in units of time describes the magnitude of

the jitter in appropriate orders of magnitude, usually picoseconds.

– As a power measurement in units of radians or unit intervals squared, which

is often expressed in decibels relative to one cycle squared

Page 52: Synchronisation 120610202311 Phpapp02

18

Wander

• Because it involves low frequencies for long periods, wander data can consist of hours of phase information.

• Because phase transients are of importance, high temporal resolution is also needed. So to provide a

concise measure of synchronisation quality, three wander

parameters have been defined and are used to specify

performance limits:

- TIE Time Interval Error (wander in ns)

- MTIE Maximum Time Interval Error (related to Peak-to-Peak wander)

- TDEV Time Deviation (related to RMS wander)

Page 53: Synchronisation 120610202311 Phpapp02

19

• The short term variations of the significant instances of a

digital signal from their reference positions in time

• Greater than 10Hz in modulation frequency

• Jitter is caused by the sync trail equipment

Ideal

Jittered

Sampling (reading) points

Definition of Jitter : ITU- Rec G.810Definition of Jitter

Page 54: Synchronisation 120610202311 Phpapp02

20

Definition of Wander

Ideal

Wander

Sampling pointsSampling points

• The long term variations of the significant instances of a digital

signal from their reference positions in time

• Less than 10Hz in modulation frequency

• Wander is caused by the interaction of technologies in a network

Page 55: Synchronisation 120610202311 Phpapp02

21

Main Causes of Jitter/Wander

Jitter/Wander is caused by the sync trail equipment

For Example:

– MUX / Switch equipment PLL

– Poor equipment component quality– Proximity of components to EMI

– Microprocessor noise

– Equipment Transfer functions

– Length of transmission paths due to cable expansion and contractions

– Inter-reaction of different technologies e.g. SDH, PDH, ATM

Page 56: Synchronisation 120610202311 Phpapp02

22

PLL Effects on Reference Signals

• PLL can overcompensate and oscillate above and below the reference

• PLL can under-compensate and take too long to get to the reference

• Changes in Temperature effect the stability of the OCXO

• Jitter is generated by granularity (steps) in correction voltage applied to the OCXO

Page 57: Synchronisation 120610202311 Phpapp02

23

Jitter/Wander summary

Page 58: Synchronisation 120610202311 Phpapp02

24

Cable Expansion/Contraction

�Wc = 80ps/Km/oC, for fibre optical cable

�Wc = 725ps/Km/oC, for copper cable

10101110010101010000011001010001000101010111001010101000001100101000100010

2020ooCC

10101110010101010000011001010001000101010111001010101000001100101000100010

4040ooCC Cable has expanded Cable has expanded -- the bits come out laterthe bits come out later

10101110010101010000011001010001000101010111001010101000001100101000100010

00ooCC Cable has contracted Cable has contracted -- the bits come out earlierthe bits come out earlier

Page 59: Synchronisation 120610202311 Phpapp02

25

ITU-T G.823 Model of Wander

• Specifies the maximum network limits for jitter and wander that should not be exceeded

• Specifies the minimum equipment tolerance to jitter and wander based on the 2048 kbit/s hierarchy

• The jitter control philosophy is based on the need:

- to recommend a maximum network limit that should not be exceeded at any hierarchical interface

- to recommend a consistent framework for the specification of individual digital equipments

- to provide sufficient information and guidelines for organizations to measure and study jitter accumulation in any network configuration

Page 60: Synchronisation 120610202311 Phpapp02

26

ITU-T G.823 Wander formula

• The maximum relative wander between the slave clock and the data input at a node is:

Dwpk = WequipPk + WconnectionsPk + < / = 18µs

• The formula allows the planner to calculate the

accumulated Wander by simply adding the sum of each element within the network

• This total must be less than 18µs to adhere to G.823

Page 61: Synchronisation 120610202311 Phpapp02

27

PRCPRC

SlaveSlave

Cable C

Wc = 6 Microseconds

Cable A

Wa = 6 Microseconds

Cable B

Wb = 4 Microseconds

Equipment Wander =

1 microsecond

Equipment Wander =

1 microsecond

Total Wander =

18microseconds

SlaveSlave

ITU-T G.823 Network Wander

This diagram demonstrates the effect of accumulative wander

Page 62: Synchronisation 120610202311 Phpapp02

28

The control of jitter and wander

• SDH requires that jitter and wander be kept below tight network limits.

• This is achieved by inserting narrow-bandwidth SSUsin the synchronisation chain (SEC bandwidth is

relatively wide).

• Narrow-bandwidth SSUs attenuate jitter and wander

components that lie outside the SSU bandwidth.

Page 63: Synchronisation 120610202311 Phpapp02

29

Sync Distribution in SDH - Rules

SASE

SASESASE

SASESASE

SASE

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SECSEC

SEC

SEC

SEC

SEC

SEC SEC

SEC

SEC

N x SECs

N x SECs

N x SECs

Level 1

Level 2

N = 20 Max

Level = 10 Max

Maximum 60 SECs in a single trail

PRC

SEC

Page 64: Synchronisation 120610202311 Phpapp02

30

Synchronisation reference chain

• See ITU-T G.803 or ETS 300 462-2

• The ITU-T/ETS synchronisation reference chain meets

the network limits on jitter and wander:

– Not more than 60 SECs in a chain

– Not more than 20 SECs between two SSUs

– Not more than 10 SSUs in the chain

Page 65: Synchronisation 120610202311 Phpapp02

31

Distribution Layer composition

The SDH Synchronisation distribution layer

is comprised of 3 parts:

1. Clocks

2. Equipment

3. Links

Page 66: Synchronisation 120610202311 Phpapp02

32

Round Up

• What are the two main causes of

synchronisation problems?

• What equipment is normally at the top of

a sync tree?

• What is the maximum allowed wander under G.823?

Page 67: Synchronisation 120610202311 Phpapp02

1

Chapter 04

Sync Distribution

Layer - Clocks

Page 68: Synchronisation 120610202311 Phpapp02

2

Distribution Layer Clocks

• The clock element of the synchronisation

distribution layer is categorised in three ways:

1. Primary Reference Clock (PRC)

2. Synchronisation Supply Unit (SSU)

3. SDH Equipment Clock (SEC)

Page 69: Synchronisation 120610202311 Phpapp02

3

PRC Systems

• The PRC System is a master clock used to synchronise the entire network with a frequency accuracy of < 1 x 10-11

• This is defined by ITU-T recommendation G.811.

• A complete Primary Reference Clock consists of the following:

– 3 separate Primary Reference Sources

– Reference Selector

– Tracking Unit

– Output Distribution Unit

Page 70: Synchronisation 120610202311 Phpapp02

4

Typical PRC

Caesium 2

GPS 1

Caesium 1

Input stage & reference Selector

Tracking Oscillator Section

Output Section

SASE Unit

Page 71: Synchronisation 120610202311 Phpapp02

5

Primary Reference Source

• This is a single source of G.811 Reference which is used to feed a PRC or can be used as a

standalone reference

• This can take two forms:

1. Caesium reference source

2. Off air source i.e. GPS, Loran, GLONASS

Page 72: Synchronisation 120610202311 Phpapp02

6

Caesium PRS

• These elements are usually expensive items but provide the network with its own source of synchronisation

– The beam can last up to 10 years before requiring a re-tube

– They provide a stability of 1x10-11

over 20 years

– Where two beams are compared the maximum difference between the two beams equates to 2x10-11 which will result in 1 slip every 72 days on E1 links

– This is the standard set by ITU and can be found in ITU-T G.811

Page 73: Synchronisation 120610202311 Phpapp02

7

Off-air PRS

• Off-air Primary Reference Sources provide an excellent alternative or additional reference to Caesium

• This is a single source of G.811 Reference which is used to feed a PRC or can be used as a standalone reference

• Off-air Primary Reference Sources include:

– Global Positioning Systems (GPS)

– Loran

– GLONASS

GPS is the most commonly used

Page 74: Synchronisation 120610202311 Phpapp02

8

GPS as a PRS

• The raw GPS signal contains various information including UTC time and identification data for navigation purposes

• The GPS satellites contain Caesium beams which are used to provide the accuracy needed for time and location information

• In Telecoms it is the stability of the Satellites Caesium that is of interest. Therefore receivers used in telecoms are specific to the requirement

• The signal is received by the GPS receiver and is converted into a useable source for slaving a local oscillator. The resulting output is G.811 compliant and therefore is suitable to be used within a PRC

Page 75: Synchronisation 120610202311 Phpapp02

9

PRS Operation

• Three PRS sources are fed into an SASE (Stand Alone Synchronisation Equipment). This forms the hub of the PRC

• The SASE is normally configured to be completely redundant. It will contain an input stage, a prioritising and selection mechanism to determine which source it will utilise, a tracking oscillator unit and an output stage

Page 76: Synchronisation 120610202311 Phpapp02

10

Which PRS is Best?

• Technically both are perfectly acceptable sources. The differing factor is mainly down to cost

• On the surface GPS looks like a cheap option, but beware, it is not always easy to obtain permission from landlords to fit antenna systems and the cost of installation can be more than the cost of the units

• When deciding on a solution look at the cost of each of the options and consult your vendor as to what is most cost effective

• Most common forms of PRC use a combination of different types of PRS

Page 77: Synchronisation 120610202311 Phpapp02

11

Oscillators

• PLL – Phase Lock Loop

– The Most common form of slaving an oscillator

– This function employs a feedback mechanism which

feeds the Oscillator output into a comparator where

the signal is compared with the input and the difference is sent as correction voltages to the

oscillator

Page 78: Synchronisation 120610202311 Phpapp02

12

Phase Lock Loop

Simple Block Diagram of a Phase Lock Loop

Phase Comparator

Error Pulses

Error Correction Voltage

OutputS1 +

-

S2

Input Referenc

e

Low Pass Filter

Voltage Controlled Oscillator

Frequency Divider(optional) Feedback Path

Page 79: Synchronisation 120610202311 Phpapp02

13

Slaved Oscillator Mechanisms

• DDS – Direct Digital Synthesis

– A mechanism in which the Oscillator is free

running and the output signal is synthesised in software with the incoming reference signal to

produce a stable output

Page 80: Synchronisation 120610202311 Phpapp02
Page 81: Synchronisation 120610202311 Phpapp02

1

Chapter 05

Sync Distribution

Layer - Equipment

Page 82: Synchronisation 120610202311 Phpapp02

2

Synchronisation Elements

• SASE - Stand Alone Synchronisation Equipment

• SSU - Synchronisation Supply Unit• TNC - Transit Node Clock

• CTO - Compact Tracking Oscillator

• BITS - Building Integrated Timing Supply

– All refer to the same class of equipment

– Used with PRCs or standalone for regenerating timing signals

– Modular by design and configured to be redundant to allow for single or multiple internal and/or external failures

Page 83: Synchronisation 120610202311 Phpapp02

3

The SASE

• SASE stands for Stand Alone Synchronisation Equipment (Element)

• The purpose of these elements is to provide filtering, regeneration and distribution of a primary reference signal

• This is achieved by simple yet highly accurate equipment

• The equipment is designed to be very resilient to internal and external failurestherefore all components are at least duplicated

Page 84: Synchronisation 120610202311 Phpapp02

4

SASE Architecture

Reference

Selector

Input

Interface

Input

Interface

Input

Interface

Output

Interface

Output

Interface

Output

InterfaceJitter/Wander

Low-Pass

Filter

Holdover

Memory

Output

Interface

Page 85: Synchronisation 120610202311 Phpapp02

5

SASE – Input Section

• Inputs

– Most SASEs have multiple input capability, allowing the unit to select from a number of references

– The amount and type of inputs are dependent upon

the make and model. For regeneration purposes, 2

or 3 inputs are sufficient

Page 86: Synchronisation 120610202311 Phpapp02

6

SASE – Reference Selectors

• Reference Selectors

– Within each SASE will be the reference selector. Normally this is duplicated for redundancy. This section contains the priority table and selection criteria for the units inputs

– Selection can be made by the following methods

• Automatic

• Manual

• Forced

• Synchronisation Status Message (SSM)

Page 87: Synchronisation 120610202311 Phpapp02

7

SASE - Selection Modes

Selection Modes

Automatic – This mode will choose the highest available priority source set within the priority table

Manual – This mode will only switch to another available input by manual intervention

Forced – This mode will stay fixed to its intended sync source at all times

SSM – The selection is dictated by the incoming SSM information

Page 88: Synchronisation 120610202311 Phpapp02

8

SASE - Fltering & Holdover

• Filtering & Holdover section

– This section of the unit provides the filtering of the reference signal. This can be achieved using two

techniques:

- Phase Lock Loop - PLL

- Direct Digital Synthesis - DDS

– This section should be duplicated for redundancy

– All SASEs employ a mechanism which prevents phase and frequency jumps when switching between channels

Page 89: Synchronisation 120610202311 Phpapp02

9

SASE – Output Section

• Output Stages

– Most SASE output arrays allow for different output frequencies to

be used. 2.048Mhz is the most common for SDH but 1MHz, 10MHz, E1 Framed, etc. can all be generated by the SASE.

Consult your vendor if specific frequencies are required

– Again these can be configured to provide protection in the event

of a hardware failure

– Cards can be fitted with differing protocols. In some units the framing and bit pattern of E1 or T1 signals can be altered to

generate AIS and other states. Units may also have the ability to

change output protocol by means of software, e.g. 2.048Mbps to

2.048Mhz

Page 90: Synchronisation 120610202311 Phpapp02

10

Which SASE Configuration?

• When deciding which equipment is right for your network, it is important to look at the application first

• SASEs can be fitted with an array of different cards and clock types

• Review the importance of the location and the equipment the SASE is to be connected to, before deciding on oscillator types and configuration

• Typically, SASEs should be used within the core of the transmission network

Page 91: Synchronisation 120610202311 Phpapp02

11

SDU – Sync Distribution Unit

• The purpose of the SDU is to expand the capacity of an SASE O/P

• The SDU typically has two I/Ps and no Hold over capability

• The SDU will have a large O/P capacity - consider it as an amplifier for Synchronisation signals

• An SDU will typically be referenced from an SASE or SSU

Page 92: Synchronisation 120610202311 Phpapp02

12

SDU – Layout

Input

Interface

Output

Interface

Output

Interface

Output

Interface

Output

Interface

Low

Loss S

plit

ter

Input

Interface

Page 93: Synchronisation 120610202311 Phpapp02

13

SDU – Inputs

• SDU Inputs

– The SDU will normally have two input Interface Units, these are typically 2Mhz. Jitter can be filtered on these units

– The SDU I/P reference source are often derived from an associated SASE / SSU

– SDUs can be daisy-chained: however the lack of holdover and wander filtering make this undesirable

– Some manufacturers have incorporated a HOU (Hold Over Unit) capability - this is normally for a single channel

Page 94: Synchronisation 120610202311 Phpapp02

14

SDU – Outputs

• SDU Outputs

– The SDU is designed as a low cost, high O/P capacity option

– Most SDU output arrays allow for different output frequencies to be used. 2.048Mhz is the most common for SDH but 1MHz, 10MHz, E1 Framed, etc. can all be generated by the SASE. Consult your vendor if specific frequencies are required

– Again, these can be configured to provide protection in the event of a hardware failure

Page 95: Synchronisation 120610202311 Phpapp02

15

SDH Elements

• With the possible exception (depending on manufacturer) of the

Optical Line Amplifiers, all the following equipments contain SECs(SDH Equipment Clocks) and should be counted within the trail

count for SDH design:

– Add Drop Multiplexers -- Microwave Systems

– Cross-Connects -- Regenerators

– Optical Line Amplifiers

This internal SEC is normally synchronised to a traffic or external

timing input signal

SECSECTraffic &timinginput

Traffic &timingoutput

Externaltiminginput

Externaltimingoutput

11

nn

Page 96: Synchronisation 120610202311 Phpapp02

16

Synchronous Equipment Timing Source (SETS)

External timinginput(2 MHz or1.5 Mbit/s or 2Mbit/s)

STM-Ninput

Externaltiming output(2MHz or1.5 Mbit/s, or 2Mbit/s)

SynchronousEquipment

TimingGenerator

NEinternaltiming

Selector C

PDHinput

SDH Equipment

Clock

SDH Equipment Clock (SEC)

Page 97: Synchronisation 120610202311 Phpapp02

17

SDH SEC features

• Input synchronisation signals are:

• STM-N aggregates and tributaries

• 2Mbit/s tributaries

• 2MHz and 2Mbit/s (non traffic) timing inputs

• Input selection is determined by:

• a priority table, that is user definable

• Synchronisation Status Message (SSM) on the STM-

N and 2Mbit/s interfaces

• Output synchronisation signals are:

• All STM-N aggregates and tributaries

• 2MHz and 2Mbit/s (non traffic) timing outputs

Page 98: Synchronisation 120610202311 Phpapp02

18

SEC Timing options

• Line timing: in nodes not equipped with a node clock (SSU)

• External timing: in nodes equipped with a node clock

• Tributary timing: only in exceptional cases, e.g. during the evolution from PDH to SDH

• Internal timing: when all synchronisation reference signals are lost (= holdover mode)

Page 99: Synchronisation 120610202311 Phpapp02

19

Line timingLine timingSTMSTM--NNSTMSTM--NN

SEC Line Timing

• Line timing: in nodes not equipped with a node clock

(SSU)

• Timing is extracted from the STM-N (optical overhead)

• In normal operating conditions timing is traceable to a G.811 primary reference

Page 100: Synchronisation 120610202311 Phpapp02

20

SEC External Timing

External timingExternal timing

STMSTM--NNSTMSTM--NN

2 MHz or 2Mbit/s2 MHz or 2Mbit/s

• External timing: in nodes equipped with a node clock

• Used for synchronisation injection points within the network

• Normally either a 2 MHz or 2 Mbit/s reference signal

• Node clock traceable to G.811 Primary RefenceSource

• Node Clock has G.812 holdover capability

Page 101: Synchronisation 120610202311 Phpapp02

21

SEC Internal Timing

Internal timingInternal timing

STMSTM--NNSTMSTM--NN

• Internal timing: when all external timing references

are lost

• Intended for failure conditions

• SEC Clock has G.813 holdover capability

Page 102: Synchronisation 120610202311 Phpapp02

22

• If a 1.5Mbit/s, 2MHz, 2Mbit/s synchronisation output is

derived from the Synchronous Equipment Timing Generator (SETG), then it is called a SETG locked output

• If a 1.5Mbit/s, 2MHz, 2Mbit/s synchronisation output is

directly derived from the OC-N or STM-N input, then it is called a non-SETG locked output

• The 2MHz timing output can be squelched when :

• The SEC enters hold-over or free-run mode

• The input SSM falls below the set threshold

SEC Timing Outputs

Page 103: Synchronisation 120610202311 Phpapp02

23

SEC - Selection Modes

• Automatic – This mode will choose the highest available priority source set within the priority table

• Manual – This mode will only switch to another available input by manual intervention

• Forced – This mode will stay fixed to its intended sync source at all times

• SSM – The selection is dictated by the incoming SSM information

Page 104: Synchronisation 120610202311 Phpapp02

24

Inter-working: SDH NE & SASE

SDH NEcleaned

traffic &

timing

outputs

SASE

Noisy

traffic &

timing

inputExternal

timing

input

External

timing output

Page 105: Synchronisation 120610202311 Phpapp02

25

Round Up

• What are the 3 main equipment types found in the sync distribution layer?

• What is a SEC?

• What provides the higher clock order - PRS or SEC?

• What is the difference between a SASE and an SEC?

Page 106: Synchronisation 120610202311 Phpapp02
Page 107: Synchronisation 120610202311 Phpapp02

1

Chapter 06

Sync Distribution

Layer - Links

Page 108: Synchronisation 120610202311 Phpapp02

2

SDH Sync Link connections - 1

• Supported by an SDH multiplex section trail

i.e. the timing information is carried by the STM-N data rate

(N x 155 Mbit/s) by a retimed PDH E1 connection

• SDH regenerator timing generators are not counted as

elements of the synchronisation distribution layer, they

belong to the synchronisation link connection

• The SDH multiplex section trail may be supported by an

optical transport layer such as DWDM (dense Wavelength Division Multiplexing) or OTN (Optical Transport Network)

Page 109: Synchronisation 120610202311 Phpapp02

3

SDH Sync Link connections - 2

OTN

SDH

PRC

OTN

SDH

SSU

Synchronisation link connection

SDH Multiplex section trail

Optical Trail

Synchronisation

SDH/SONET

OTN/DWDM

Page 110: Synchronisation 120610202311 Phpapp02

4

Types of Oscillators in Links

• Oscillators are a common section of all of the three clock types described

• Various types of oscillators exist and their selection is based upon the application, i.e. PRC, SASE or SEC,that they are to be utilised within

• The main types of crystal oscillators are:

– Caesium

– High Quality Rubidium

– Low Quality Rubidium– High Stability Double Oven OCXO

– Single Oven OCXO

– Temperature Controlled TCXO

Page 111: Synchronisation 120610202311 Phpapp02

5

Rubidium Oscillators

• Atomic Standard Tracking Oscillators are extremely stable and thus

are very suitable for using within a telecom network

• Typically these oscillators are used within a Primary Reference

System as Slaved units to Caesium or GPS

• These are usually the most expensive tracking oscillators available

• They have a life span varying from 6 to 12 years, depending on

Manufacturer

• The longer life span of Rubidium oscillators is achieved using DDS

rather than PLL techniques within the clock element

• Usually Maintenance Free

• Holdover Quality dependent on Manufacturer, typically 1x10-11/mth

Page 112: Synchronisation 120610202311 Phpapp02

6

Single Oven OCXO

• Relatively inexpensive OCXO

• Provides Holdover of 1x10-9 to 1x10-10 per day depending

on Manufacturer

• Usually suited for Standby oscillators within SASEs or as

Local Node Clock Systems

• They will also be found in Large transit Switches and X-

Connects

Page 113: Synchronisation 120610202311 Phpapp02

7

High Stab. Double Oven OCXO

• Life span of 20+ years

• Maintenance free

• Lower cost than Rubidium

• Holdover is typically 1x10-11/day

• Can be used in a Primary Reference System or as a SASE Oscillator

Page 114: Synchronisation 120610202311 Phpapp02

8

Temperature Controlled TCXO

• Inexpensive Oscillators

• Maintenance free

• Used in a majority of Telecom equipment

such as Multiplexers and Radio Systems

• Holdover of 1x10-6 per day or less

Page 115: Synchronisation 120610202311 Phpapp02

9

Typical Oscillator Performance

1400 per day2x10-6/dTCXO

1 after 1 day2x10-10/dSingle Oven OCXO

1 after 3.5 days1x10-11/dHigh Stability Double Oven OCXO

1 after week5x10-11/mLow cost Rubidium

1 after month1x10-11/mRubidium

Slip Rate (Worst Case)

Holdover Quality (Typical)

Oscillator Type

Page 116: Synchronisation 120610202311 Phpapp02

10

Oscillator Characteristics

• Oscillators are susceptible to changes in

temperature and stability of rectified

power

• Variations in these conditions can affectthe performance of the oscillator

Page 117: Synchronisation 120610202311 Phpapp02

11

Retimed PDH Sync Link - 1

• PDH path layers supported by SDH path layers are not suitable for transporting synchronisation

• Retiming is used when E1 traffic signals

transported over SDH are used as synchronisation links (e.g. to synchronise distant

PABXs or GSM BTSs)

• Retiming is applied on E1 traffic signals affected

by excessive wander (e.g. from pointer adjustments)

• Retiming buffers can be integrated in the SDH

network element or the SASE

Page 118: Synchronisation 120610202311 Phpapp02

12

Retimed PDH Sync Link - 2

SDH

network element

STM-

NSECSEC

Retiming

Buffer

PDHtributaryoutput rere--timedtimed

PDHPDH

signalsignal

SECtiming signal

PDHtributaryoutput rere--timedtimed

PDHPDH

signalsignal

Retiming

Buffer

PDHtributaryoutput rere--timedtimed

PDHPDH

signalsignal

Retiming

Buffer

Page 119: Synchronisation 120610202311 Phpapp02

13

Retiming

• The retiming buffer transmits the incoming traffic at

the data rate of the SEC timing signal, thus removing

the excessive wander

• The long-term frequency (data rate) of the E1 traffic

signal must be synchronized to the network PRC

• Slips will occur if the SEC has lost its

synchronisation to the PRC

Page 120: Synchronisation 120610202311 Phpapp02

14

Round Up

• What is the synchronisation link layer?

• How is synchronisation transported from site to site?

• How is synchronisation delivered within the node?

• Name 3 Oscillator types

Page 121: Synchronisation 120610202311 Phpapp02

1

Chapter 07

SDH

Network Topology

Page 122: Synchronisation 120610202311 Phpapp02

2

Master-Slave Principle

• A designated master clock is used as a

reference frequency generator

• The frequency generated by the master clock

is disseminated to all other clocks which are

slaved to the master clock

Page 123: Synchronisation 120610202311 Phpapp02

3

Master-Slave Mechanism

MasterMaster SlaveSlave

PRC 1

Data

Transmission Link

Data + Clock

• The clock is injected into the master unit

• The slave unit locks to the incoming clock rate and is synchronised to the master

• No slips occur between these elements

Page 124: Synchronisation 120610202311 Phpapp02

4

= slave

PRC

SEC

SSU

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

= master

= slave

= slave

= slave

= slave

= slave

SSU

SSU SSU SSU

Master-slave principle

Page 125: Synchronisation 120610202311 Phpapp02

5

Principle of trail redundancy

• Each slave clock should get at least two reference signals from the master clock via geographically separate trails

• Sometimes it is not possible to fulfil this principle for all nodes of the network (depending on connectivity)

Page 126: Synchronisation 120610202311 Phpapp02

6

Hierarchy of Quality Levels

• There is a hierarchy of clock quality levels

• The higher the clock quality level, the higher the frequency accuracy of the clock

• Frequency accuracy =

– either overall free-run accuracy or holdover accuracy

over a limited time period

Page 127: Synchronisation 120610202311 Phpapp02

7

Clock Quality Levels

2048 kbit/s based:

PRC: 1E-11

SSU I: 2E-10/d

SEC 1: 4.6E-6

1544 kbit/s based:

PRC: 1E-11

SSU II: 1.6E-8/1yr

SSU III/IV: 4.6E-6

SEC 2: 20E-6

Page 128: Synchronisation 120610202311 Phpapp02

8

Weak Hierarchical Distribution

RULE

• A clock of a given quality level must always (even

under failure conditions) take timing (directly or indirectly) from a source clock with the same or

higher quality level

Page 129: Synchronisation 120610202311 Phpapp02

9

Implementing the Rule

• How can we implement the Weak Hierarchical

Distribution Rule? ..........

• By implementing the Strong Hierarchical Distribution

Rule:

« A clock of a given quality level must take timing (directly) from a clock with the same

or higher quality level »

• Or by the use of SSM signalling

Question:

Answer:

Page 130: Synchronisation 120610202311 Phpapp02

10

Strict Hierarchical Layering

PRC

SSU SSU

SSUSSU SSU

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC SEC

Page 131: Synchronisation 120610202311 Phpapp02

11

Failure Scenario

PRC

SEC

SSU

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

Link failure!

Holdover mode!

SEC

SEC

SSU

SSU

SSUSSU

Page 132: Synchronisation 120610202311 Phpapp02

12

Sync network with SSM

• There is a link failure within a chain of SECs

• The SSM signalling prevents the downstream SSU from

following a SEC in holdover mode

• Instead, the downstream SSU enters holdover mode and

becomes the source clock for the cut off sub-network

Page 133: Synchronisation 120610202311 Phpapp02

13

The control of jitter and wander

• SDH requires that jitter and wander be kept below tight network limits

• This is achieved by inserting narrow-bandwidth SSUsin the synchronisation chain (SEC bandwidth is

relatively wide)

• Narrow-bandwidth SSUs attenuate jitter and wander

components that lie outside the SSU bandwidth

Page 134: Synchronisation 120610202311 Phpapp02

14

Sync Distribution in SDH

SASE

SASESASE

SASESASE

SASE

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SECSEC

SEC

SEC

SEC

SEC

SEC SEC

SEC

SEC

N x SECs

N x SECs

N x SECs

Level 1

Level 2

N = 20 Max

Maximum 60 SECs in a single trail

PRC

SEC

Maximum 10 SASEs in a single trail

Page 135: Synchronisation 120610202311 Phpapp02

15

Synchronisation reference chain

• See ITU-T G.803 or ETS 300 462-2

• The ITU-T/ETS synchronisation reference chain meets the network limits on jitter and wander:

– Not more than 60 SECs in a chain

– Not more than 20 SECs between two SSUs

– Not more than 10 SSUs in the chain

Page 136: Synchronisation 120610202311 Phpapp02

16

Summary

OTN

SDH

OTN

SDH

Synchronisation

SDH/SONET

OTN

PRC

PSTN

SSU

PSTN

Page 137: Synchronisation 120610202311 Phpapp02

17

Synchronisation Signalling layer

• To provide the source clock quality level from clock to clock down the synchronisation chains, in order

to:

– Enable clocks to select the best available reference timing signal

– Enable clocks to go into holdover mode if

reference timing signals are of low quality

– Prevent timing loops in SDH chains and rings

Function:

Page 138: Synchronisation 120610202311 Phpapp02

18

• The clock source quality level is indicated by the Synchronisation Status Message (SSM)

• In SDH, the message set is:

• QL-PRC = PRC, G.811

• QL-SSU-A = SSU, G.812 Type I or V

• QL-SSU-B = SSU, G.812 Type VI

• QL-SEC = SEC, G.813 Option 1

• QL-DNU = Do not use

SDH Sync Status Messages

Page 139: Synchronisation 120610202311 Phpapp02

19

SSM Transmission Channels

• The timing quality level carried by STM-N signals (SDH)

is indicated by the S1 byte in the STM-N Multiplex

Section Over Head (MSOH)

• The timing quality level carried over 2048 kbit/s

synchronisation signals is indicated in one of the bits Sa4

to Sa8 in Time Slot Zero (TS0)

• 1544 kbit/s T1 signals: see ITU-T Rec. G.704

• 34 Mbit/s E3 and 140 Mbit/s E4 signals: see ITU-T Rec.

G-832

Page 140: Synchronisation 120610202311 Phpapp02

20

SSM Algorithm

• Always select the highest quality input and if a number

of equal quality timing inputs are available, then select

the highest priority timing input

• In locked mode, the output SSMs are set to the

selected input SSM e.g. G.811 in = G.811 out

• The SSM in the return direction of the selected input is automatically set to Do Not Use (DNU)

Page 141: Synchronisation 120610202311 Phpapp02

21

If all inputs are bad ...

• The SEC enters holdover mode…

• The SEC memorises the phase and frequency values of the last known good input, but quickly drifts toward 4.6 x 10-6

• The SEC will be in free-run mode if it has never locked to a higher level reference signal

• Ext Clock Out signals should be squelched

• Output SSM STM-n value is set to G.813 (unless manually set)

• Unfortunately, SECs inject jitter on to the PRC signal and accumulative jitter can cause slips

Page 142: Synchronisation 120610202311 Phpapp02

22

SDH ring sync protection

• Automatic SSM correction and automatic

synchronisation distribution trail reconfiguration under

failure conditions

– Using the SSM algorithm

• Revertive operation

– SD trails returns to the original paths when the

failed section or the failed network element has been repaired

• No operator action is needed

Page 143: Synchronisation 120610202311 Phpapp02

23

Revertive/Non-Rev Switching

• Revertive switching will allow previously disqualified inputs to be re-qualified and re-selected as the selected source

• Non-Revertive switching will not allow previously disqualified inputs to be re-selected if they return to a useable reference

• Pro’s & Con’s are associated with either option. The operator must decide which method to use as a standard for the whole network

Page 144: Synchronisation 120610202311 Phpapp02
Page 145: Synchronisation 120610202311 Phpapp02

1

Chapter 08

Synchronisation

Network Architecture

Page 146: Synchronisation 120610202311 Phpapp02

2

Centralised or Distributed PRC

• Two methods of deploying PRCs are utilised

in today’s networks:

– Centralised

– Distributed

• A combination of both methods is also a valid

strategy for Synchronisation

Page 147: Synchronisation 120610202311 Phpapp02

3

Centralised PRCs - 1

• This method was typically used by operators when PDH systems were used as the main transmission media.

• This method, utilised 2 or 3 fully equipped PRC’s located separately. The clock was embedded within the E1 bearer on the Primary Multiplexers and distributed over the PDH to the Exchanges.

• Each PRC system typically employed three Caesium beams for redundancy.

• These systems were expensive to purchase and required maintenance to be carried out.

Page 148: Synchronisation 120610202311 Phpapp02

4

Centralised PRCs - 2

• In today’s networks the Centralised PRC is still supported and is still valid for timing SDH.

• With the onslaught of new technologies and transport mechanisms, new operators have approached the issue of network synchronisation differently.

• Preferring to have multiple low cost PRS clocks situated all around the network – Distributed PRCs.

Page 149: Synchronisation 120610202311 Phpapp02

5

Physical View

SEC

PRC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC

SEC SEC SECSEC

Equipment Clocks

PRC = Primary Reference Clock

The PRC distributes timing through master slave Synchronous Equipment Clocks.All elements are traceable to the PRC

Page 150: Synchronisation 120610202311 Phpapp02

6

• There are many primary reference sources distributed in the

network

• The most common form of PRS utilised for this purpose is GPS

• The GPS satellite system distributes USNO-UTC-derived time and timing to all GPS-clocks

• Each GPS-clock is the master of a synchronization subnetwork

• This is also referred to as decentralised PRCs

Distributed PRS - 1

Page 151: Synchronisation 120610202311 Phpapp02

7

Distributed PRS - 2

• This mode of clocking is favoured by the new operators. Mainly due to having isolated sites or regions which are linked by lines, leased from the incumbent carrier

• In this instance, trace-ability is lost when transported over another operator's network. Therefore, installing primary reference sources at all sites ensured synchronisation quality is maintained

Page 152: Synchronisation 120610202311 Phpapp02

8

Distributed PRS - 3

MasterMaster MasterMaster

PRS 1

Data

Leased Transmission

Link

Data + Clock

PRS 2

• Each element or node is timed by a separate high stability clock

• These are virtually identical speeds (accuracy of 1x1011)

• Will cause one slip every 72 days - perfectly acceptable

Page 153: Synchronisation 120610202311 Phpapp02

9

Physical View

Local Equipment Clocks

PRC… Primary Reference Clock

SEC…SDH Element Clock

Region 1

Region 2

Region 3

Timing FeedsTraffic Links

PRC

SEC

SEC

SEC

SEC

SEC

PRC

SEC

SEC

SEC

SEC

SEC

PRC

SEC

SEC

SEC

SEC

SEC

Page 154: Synchronisation 120610202311 Phpapp02

10

• Mixed synchronisation network architectures offer a combination of benefits from centralised and distributed PRS solutions

• Less wander due to short synchronisation chains

• Simplified network design

• Easy to modify the network (evolution)

• Low risk of creating timing loops

Mixed sync distribution - 1

Page 155: Synchronisation 120610202311 Phpapp02

11

nnnn

nn nn

GG

nnnn

nn nn

GG

nnnn

nn nn

GG

nnnn

nn nn

GG

nnnn

nn nn

GG

nnnn

nn nn

GG

MM

Sub-networkSub-networkGPS-clockGPS-clock equipmentclock

equipmentclock

G nodeclock

nodeclocknn

central master clock

central master clock

M

Mixed sync distribution - 2

Page 156: Synchronisation 120610202311 Phpapp02

12

• It is technically feasible to deploy one GPS-clock per node

• Less wander due to very short synchronization chains

• Very simple network design

• Easy to modify the network (evolution)

• Very low risk of creating timing loops

• Not robust enough, since there is only one synchronizationreference signal available (dual GPS receivers do not provide

protection against interference and jamming!) - no trail redundancy

• Not economical for very large networks with many nodes, because

of total equipment cost

Sync entirely based on GPS

Page 157: Synchronisation 120610202311 Phpapp02

13

Which Strategy is Best? - 1

• There are benefits and drawbacks to both options. Both methods are perfectly acceptable

• With distributed GPS there is a cost impact for installation and technically speaking there are multiple boundaries inside the network

• Plan for a failure. If one GPS fails the network must be able to recover from a standby GPS or if possible from a GPS at an adjacent site

Page 158: Synchronisation 120610202311 Phpapp02

14

Which Strategy is Best? - 2

• With a centralised PRC careful planning is required to prevent timing loops

• Inter oscillator chain length can be large, so regenerator units such as SASE’s are required

• If planned correctly this can be a more cost effective way of synchronising the network

• All elements will be traceable to one Master Clock

Page 159: Synchronisation 120610202311 Phpapp02

15

Totalequipment

cost

number of nodes or synchronisation sub networks

60-80

GPS-BASED SYNCHRONISATION

DISTRIBUTION:

(single GPS + SASE) per node

WIRED (PDH- OR SDH-BASED)

SYNCHRONIZATION

DISTRIBUTION: one central PRC with 3 Cs clocks,

and one SASE per node

Equipment cost model

Page 160: Synchronisation 120610202311 Phpapp02

16

• Most common when there is no PRC in the network

• All clocks in the network are slaved to synchronization signals from a co-operating network

• Under normal operating conditions all slave clocks operate at the same frequency as the PRC in the co-operating network

• There are normally no slip for on-net and off-net traffic to the co-operating network

Sync from a co-operating N'wk - 1

Page 161: Synchronisation 120610202311 Phpapp02

17

• The clock signals from the co-operating network may be received at only a few synchronisation gateway nodes

• The clock signals from the co-operating network may also be received at every node, or at every sub-network

Sync from a co-operating N'wk - 2

Page 162: Synchronisation 120610202311 Phpapp02

18

• The network’s synchronisation performance is dependent on the quality of the synchronisation signals from the co-operating network

• There must be an agreement with the co-operating operator on service level

• The cost to lease the synchronisation signals can be high

Sync from a co-op N'wk - Issues

Page 163: Synchronisation 120610202311 Phpapp02

19

• Physical interface specification (e.g. 2 Mbit/s, G.703)

• SSM configuration

• Guaranteed synchronization quality (e.g. G.823 Network Limit)

• Upstream synchronisation chain length (number of clocks)

• Guaranteed availability of agreed quality (e.g. 0.9999)

• Mean Time to Repair in case of failure

• Worst case quality degradation in case of failure (e.g. max.

frequency error, max. frequency drift, max. jitter & wander)

• Alarming method in case of failure (e.g. SSM)

• Quality monitoring criteria

Agreement on sync interfaces

Page 164: Synchronisation 120610202311 Phpapp02

20

Round Up

• What is a master slave clock arrangement?

• What is a centralised PRC system?

• What is a de-centralised PRC system?

• Which system is best?

• Give two examples of a PRS

Page 165: Synchronisation 120610202311 Phpapp02

1

Chapter 09

Synchronisation

Standards

Page 166: Synchronisation 120610202311 Phpapp02

2

Standardisation Bodies

• International level : ITU Recommendations

• Regional level, Europe: ETSI Legally binding standards

• USA: ANSI Legally binding standards

• Industry level: e.g. TIA Industry standards

• Company level: e.g. Bellcore Internal standards

ITU : International Telecommunication Union

ETSI : European Telecommunications Standards Institute

ANSI : American National Standards Institute

TIA: Telecommunication Industry Association

Page 167: Synchronisation 120610202311 Phpapp02

3

ITU-T Recommendations

Page 168: Synchronisation 120610202311 Phpapp02

4

ITU-T Rec. G.810

• Definitions for synchronisation networks

• Includes the definitions of time error, MTIE, TDEV, etc.

Page 169: Synchronisation 120610202311 Phpapp02

5

ITU-T Rec. G.803

• Architecture of SDH transport networks

• Section 8.2: architecture of SDH-based synchronisation networks

• Section 8.2.4: synchronization network reference chain

Page 170: Synchronisation 120610202311 Phpapp02

6

ITU-T Rec. G.823 (11/98)

• The control of jitter and wander in PDH networks based on the 2048 kbit/s hierarchy:

1.\ Maximum network limits on jitter and wander

2.\ Minimum equipment tolerance to jitter and wander

Page 171: Synchronisation 120610202311 Phpapp02

7

G.823 - What does it specify?

• Network limits for traffic interfaces

• Network limits for synchronization interfaces

• Jitter and wander tolerance of traffic interfaces

Page 172: Synchronisation 120610202311 Phpapp02

8

ITU-T Rec. G.824

• The control of jitter and wander in PDH networks based on the 1544 kbit/s hierarchy

• Similar to ITU-T Rec. G.823, but for the 1544 kbit/s based PDH network

Page 173: Synchronisation 120610202311 Phpapp02

9

ITU-T Rec. G.825

• The control of jitter and wander in SDH networks based on the 2048 kbit/s hierarchy

• Similar to ITU-T Rec. G.823, but for SDH networks

Page 174: Synchronisation 120610202311 Phpapp02

10

ITU-T Rec. G.811

• Specification for Primary Reference Clocks (PRC)

• It is an equipment specification

• Specifies only one PRC type

Page 175: Synchronisation 120610202311 Phpapp02

11

G.811 - What does it specify?

• Frequency accuracy

• Noise generation

• Phase discontinuity in case of internalprotection switching

Page 176: Synchronisation 120610202311 Phpapp02

12

ITU-T Rec. G.812 (6/98)

• Specification for Node Clocks (Node Clock is the ITU-T term for SSU)

• It is an equipment specification

• Specifies six SSU types

Page 177: Synchronisation 120610202311 Phpapp02

13

ITU-T Rec. G.812 - SSU Types

SSU Type Primary ApplicationWhich hierarchy? Which case?

Type I 2048 kbit/s Sync. chains as long as G.803 reference chains

Type II 1544 kbit/s Distribution hubs (1); single input reference

Type III 1544 kbit/s End offices (1)

Type IV 1544 kbit/s If used in SDH: must also comply with G.813 option 2

Type V 1544 & 2048 kbit/s Existing (2)transit nodes; same as TNC G.812 - 1988

Type VI 2048 kbit/s Existing (2)local nodes; same as LNC G.812 - 1988

Note (1): see Bellcore terminology

Note (2): prior to introduction of SDH

Page 178: Synchronisation 120610202311 Phpapp02

14

G.812 - What does it specify?

• Frequency Accuracy

• Pull-in, hold-in, and pull-out ranges

• Noise tolerance

• Noise generation

• Noise transfer

• Transient response in case of input reference switching

• Holdover performance

• Phase discontinuity in case of internal protection switching

Page 179: Synchronisation 120610202311 Phpapp02

15

ITU-T Rec. G.813

• Specification for SDH Equipment Clocks (SEC)

• It is an equipment specification

• Specifies two SEC types:

– SEC Option A: 2048 kbit/s hierarchy

– SEC Option B: 1544 kbit/s hierarchy

Page 180: Synchronisation 120610202311 Phpapp02

16

G.813 - What does it specify?

• Frequency Accuracy

• Pull-in, hold-in, and pull-out ranges

• Noise tolerance

• Noise generation

• Noise transfer

• Transient response in case of input reference switching

• Holdover performance

• Phase response to input signal interruption

• Phase discontinuity in case of internal protection switching

Page 181: Synchronisation 120610202311 Phpapp02

17

Round Up

• What does ITU-T G.811 specify?

• What does ITU-T G.812 specify?

• What does ITU-T G.813 specify?

• How many levels are defined under G.812?

Page 182: Synchronisation 120610202311 Phpapp02
Page 183: Synchronisation 120610202311 Phpapp02

1

Chapter 10

Sync E

(Synchronous Ethernet)

Page 184: Synchronisation 120610202311 Phpapp02

2

2

What’s driving the industry?

•Networks migrating to Packet but mission-critical Comms and Data

need accurate timing

•Must continue to provide the same quality of synchronisation delivered

today by TDM

•Can this be done without the need

to retain the T1/E1 links used today to transfer synchronisation?

Base Station

Ethernet

TDM

Base

Station

Future:Ethernet-only

Today:

Hybrid Network

Core Network

2

Page 185: Synchronisation 120610202311 Phpapp02

3

Sync-E and IEEE1588V2 compared

• less than 100 nanosecond (target 50nS) time-of-day precision over Ethernet LAN.

• Less than 1µs time-of-day precision over switched Ethernet WAN.

• stable frequency (1.6x10-8 or 16ppb) recovery (from time-of-day reference).

IEEE1588V2 delivers Frequency, Phase & ToD

1588V2 CAN MEET & EXCEED MOBILE NETWORK REQUIREMENTS

Synchronous Ethernet delivers Frequency Only

• Regular heartbeat on link signal provides frequency synchronization.

Application Frequency Time

GSM 50 ppb N/A

UMTS/ W-CDMA 50 ppb N/A

UMTS/ W-CDMAfemtocells

250 ppb N/A

GSM, UMTS, LTE Network Interface

16 ppb, suggested to meet 50ppb RF specification

N/A

CDMA2000 50 ppbShould +/-3µs, shall +/-10µs

TD-SCDMA 50 ppb +/- 1.5µs

LTE (FDD) 50 ppb N/A

LTE (TDD) 50 ppb+/- 1.5µs small cell, +/-5µs large cell

LTE MBSFN 50 ppb+/- 1-32µs, implementation dependent

LTE-A CoMP (Network MIMO)

50 ppb+/- 500 ns (0.5 µs), pre-standard

WiMAX (TDD)2 ppm absolute, ~50 ppb between base stations

+/-1 - 8 µs, implementation dependent

Page 186: Synchronisation 120610202311 Phpapp02

4

Building a N'wk with SyncE/1588v2

G.8265.1(PTPprofileFrequency)

G.8275.1(PTPprofileTime/phase)

Basics

Clock

Methods

Profiles

Frequency: G.826x Time/Phase:G.827x

G.8265.m(PTP Profile frequency m)

G.8261.1(NetwkPDV_frequency)

G.8271.1(NetwkPDV_time/phaseNetwork

requirements

SyncE NetwkJitter-Wander: Included in G.8261

G.8261 G.8271

G.8272PRTC

G.8273

G.8275(Packet-architecture-time)

G.8262 (SyncE)

G.8263

G.8264(SyncE-architecture)

G.8265(Packet-architecture-Frequency))

G.8275.n(PTPprofileTime/phase n)

G.8271.2may be needed in future

Definitions /

terminology

G.8260 (Definition)

73.1-GM

73.2 BC

73.3 TC

G.8260(metrics)

ConsentDec2011

OngoingAgreed

ITU-T Standards

Page 187: Synchronisation 120610202311 Phpapp02

5

Synchronous Ethernet (SyncE)

– Line rate of the Ethernet Interface used to transfer timing

– No impact/demand on packet layers

– Defines the use of a high stability oscillator to generate

line frequency

– Ethernet ‘Classic’: ±100ppm

– Synchronous Ethernet: ±4.6ppm

– ITU-T Standards in place

• G.8262: Timing Characteristics for Synchronous Ethernet Equipment

• G.8261: Timing & Synchronisation in Packet Networks

• G.8264: Distribution of Timing Through Packet Networks (ESMC)

Page 188: Synchronisation 120610202311 Phpapp02

6

Challenges

• Cost: All interfaces need to be Sync-E compatible

• Cannot be used with existing Ethernet equipment

when transferring synchronisation

~

PRC�

�~

PRC

Page 189: Synchronisation 120610202311 Phpapp02

7

Conformance Testing - Jitter

1G/10G SyncE

EEC

1G/10G SyncE

Jitter Generation

Jitter Tolerance

Sync-E (1GbE/10GbE) ITU-T G.8262

Page 190: Synchronisation 120610202311 Phpapp02

8

Conformance Testing - Wander

Sync-E (100M/1GbE/10GbE) ITU-T G.8262

100M/1G/10G SyncE

Synchronisation

Source

EEC

Reference

100M/1G/10G SyncE

Wander generation

Wander tolerance

Wander transfer

Frequency Accuracy

Pull-in, Pull-out, Hold-in

SyncE

Under Test

SyncE

(Wander-free)

EEC

Phase transient response

Reference

Page 191: Synchronisation 120610202311 Phpapp02

9

Jitter/Wander measurement

narrow-band filter(jitterless)

wide-band filter(jittered)

internal

internal

Page 192: Synchronisation 120610202311 Phpapp02

10

Sync architecture for Sync-E

PRC

G.812

Type I

(SSU)

SEC

SEC

SEC

SEC

SEC

EEC

EEC

SEC

G.812

Type I

(SSU)

G.8262 is

Compatible

with

G.813

Total number of

G.813 clocks in a

sychronisation trail

should not exceed 60

Number of G.812 type I

clocks < 10

Number of G.813 option 1

clocks < 20

Page 193: Synchronisation 120610202311 Phpapp02

11

White Paper & Application Note

Page 194: Synchronisation 120610202311 Phpapp02

12

G.8260

Metrics

Metrics – released as ratified by ITU-T

Includes pktfilteredMTIE, MATIE, MAFE, etc.

1588v2

Network

pktfilteredMTIE

Page 195: Synchronisation 120610202311 Phpapp02

13

G.8261 – Appendix VI

Slave Clock (Frequency) Test (Old)

Apply G.8261 or Network Profiles, Measure E1/T1 MTIE/TDEV (now)

Stress multiple slaves with G.8261 or Network Profile (H1 2012)

Stress multiple slaves with multiple profiles (H2 2012)

Page 196: Synchronisation 120610202311 Phpapp02

14

G.8263

Slave Clock (Frequency) Test (New)

Apply G.8263 Profile, Measure E1/T1 MTIE/TDEV (now)

Apply G.8263 Profile to multiple slaves, Measure E1/T1 MTIE/TDEV (H1 2012)

Apply G.8263 Profile to multiple slaves, Measure multiple E1/T1 MTIE/TDEV (H2 2012)

Page 197: Synchronisation 120610202311 Phpapp02

15

Introducing ESMC

• ESMC: Ethernet Synchronization Messaging Channel

• ESMC has been built first and foremost as the transport

channel for SSM (QL) over Synchronous Ethernet links

• Key outcome: Simple and efficient

• ESMC does not aim to become a complex protocol

• However, in the future it may support some extensions

• It is not a control plane and does not need a control plane

Page 198: Synchronisation 120610202311 Phpapp02

16

G.8264 – ESMC testing

Stimulus to EEC Response from EEC

1. ESMC = PRC1.ESMC = PRC2.Wander locked to

Reference

2. ESMC = DNU

1.ESMC changes to EEC1/22.Wander shows offset as

EEC now locked to local clock

3. ESMC = PRC1.ESMC changes to PRC2.Wander locked to reference

•Wander Graph shows Line Clock Rate

switching into and out of Holdover

•ESMC Graph shows ESMC

messages changing state to reflect status

Change QL of ESMC

100M/1G SyncE (Wander free)

SynchronisationSource

EEC

Port 1

Paragon GUI&

TimeMonitor

10MHz, 2.048MHz, E1, T1

Clock with Freq

offset

Port 2

ESMC with EEC defined QL

Page 199: Synchronisation 120610202311 Phpapp02

1

Chapter 11

IEEE 1588v2 - PTP

(Precision Timing Protocol)

Page 200: Synchronisation 120610202311 Phpapp02

2

Packet Sync Technologies

• Separate packet flow used to transfer timing

– Timestamps embedded in packets to transfer timing

– Two-way protocol employed to measure delay between Master and Slave devices

• Able to transfer frequency (syntonisation) and phase/time-

of-day (synchronisation)

• Standards define devices/techniques to reduce uncertainty (Peer-to-peer & End-to-end Transparent Clocks) and to

create hierarchical clocking topology (Ordinary Clocks,

Boundary Clocks)

Page 201: Synchronisation 120610202311 Phpapp02

3

Standards

• IEEE: 1588v2; Precision Timing Protocol, PTP

• IETF: RFC1305; Network Time Protocol, NTPv3, RFC5905: NTPv4 (TICTOC group) Deployed

by Ericsson

• ITU-T: G.8264; Distribution of Timing through

Packet Networks

Page 202: Synchronisation 120610202311 Phpapp02

4

IEEE 1588v2 (PTP)

Benefits:

– 1588v2 Standard ratified March 2008

– Independent of services

– Suitable for layered/complex clock distribution

topologies

– Compatible with currently deployed packet networks

Drawbacks:

– Cost: Extra bandwidth required

– Protocol assumes symmetrical delays in up- and down-stream paths

– Sensitivity to PDV

Page 203: Synchronisation 120610202311 Phpapp02

5

IEEE1588 Network Sync - 1

Phase 1 - Establishes the Master-Slave hierarchy via the session protocol and a local state machine

Page 204: Synchronisation 120610202311 Phpapp02

6

Phase 1 - Session Management

Session Start-up

Signalling (Request Announce)

Signalling (Acknowledge)

Signalling (Request Sync)

Signalling (Acknowledge)

Signalling (Request Del_resp)

Signalling (Acknowledge)

Announce and Signaling

messages configure and

maintain the clocking structure - they include:

• Clocking Topology

• Grand Master identity and priority

• Timestamps

• Current UTC offset

Master

Clock

Slave

Clock

Page 205: Synchronisation 120610202311 Phpapp02

7

IEEE1588 Network Sync - 2

Phase 2 - Synchronises the clocks

Tmpd = (T2 – T1) – (T4 – T3)

2

1) Mean Progation Delay

Offset = T2 – T1 – tmpd

2) Clock offset correction

*simple model

3) Slave Clock Synchronisation

*Calculation differs by vendor

Master

Clock

Slave

Clock

Page 206: Synchronisation 120610202311 Phpapp02

8

Phase 2 – Slave Clock Sync

Propagation Delay Message Exchange

t-sm

t-ms

Follow_Up

Sync

Delay_Req

Delay_Resp

t1

t4

t2

t3

Clock Output must comply with the

relevant ITU-T clock specification (MTIE & TDEV specification)

G.81x series of specifications

(G.823/4 for TDM delivery)

Master

Clock

Slave

Clock

Page 207: Synchronisation 120610202311 Phpapp02

9

Time Transfer Example

Master Clock Slave Clock

sync (t1)

delay_req(uest)

delay_resp(onse) (t4)

t2

t3

Data At

Slave Clock

t1 = 100 seconds

t2 = 152 seconds

(150+2)

t3 = 157 seconds

(152+5)

t4 = 109 seconds

(100+2+2+5)

Assume at an instant in time:

Master clock value = 100 seconds

Slave clock value = 150 seconds

(the slave clock error = 50 seconds)

One way path delay = 2 seconds

Sync message is sent at t = 100 seconds

For illustration, Delay_Req is sent 5 seconds after

the Sync message is received:

Round Trip Delay

RTD = (t2 - t1) + (t4 - t3)

RTD = (152 - 100) + (109 - 157)

RTD = 4 seconds

Slave clock error eliminated

Slave Clock Error = (t2 - t1) - (RTD ÷ 2)

= (152 - 100) - (4 ÷ 2)

= 50 seconds

Round trip error eliminated

If the slave clock is adjusted by -50 seconds, the

Master & Slave will be synchronized.

t1

t4

•2s

Page 208: Synchronisation 120610202311 Phpapp02

10

Influences on 1588v2 Accuracy

� Packet Delay Variation

(PDV) appears as a

change in frequency or

phase of the recovered

clock–Increases with number of

network elements and traffic load

–Multiple causes, including queuing delays, routing changes,

congestion, use of switches

versus routers, etc.

� Network asymmetry

� Prolonged Packet

Loss (Outage)–Causes clock recovery process

to enter holdover

� Slave Performance�Vendor A – PDV tolerance – X

�Vendor B – PDV tolerance – Y

� Packet Delay, Packet

Loss, and Packet

Errors are not an

issue for packet timing protocols

Page 209: Synchronisation 120610202311 Phpapp02

11

Today’s test challenges (G.8261)

• Building a ref. network/designing algorithms is time-consuming and expensive

• Not repeatable (varying behaviour) and inconsistent (many

different implementations)

• Does it truly represent YOUR network?

Page 210: Synchronisation 120610202311 Phpapp02

12

1588v2 Slave (Ordinary Clock)

• In a network with legacy routers, PDV and Asymmetry accumulation (in each direction) can be significant

• Slave clock recovery is a challenge

Master

Clock

Slave

Clock

Router Router

PDV Accumulation

Router Router

Page 211: Synchronisation 120610202311 Phpapp02

13

Testing 1588v2 Ordinary Clocks

• Testing 1588v2 Ordinary Clocks involves applying PDV and testing the recovered Slave clock

– Connect equipment as shown

– Apply PDV profile

– Test recovered clock to ITU-T limits

Page 212: Synchronisation 120610202311 Phpapp02

14

G.8261 – Appendix VI

Slave Clock (Frequency) Test (Old)

• Apply G.8261 or Network Profiles, Measure E1/T1 MTIE/TDEV (now)

• Stress multiple slaves with G.8261 or Network Profile (H1 2012)

• Stress multiple slaves with multiple profiles (H2 2012)

Page 213: Synchronisation 120610202311 Phpapp02

15

G.8263

Slave Clock (Frequency) Test (New)

• Apply G.8263 Profile, Measure E1/T1 MTIE/TDEV (now)

• Apply G.8263 Profile to multiple slaves, Measure E1/T1 MTIE/TDEV (H1 2012)

• Apply G.8263 Profile to multiple slaves, Measure multiple E1/T1 MTIE/TDEV (H2 2012)

Page 214: Synchronisation 120610202311 Phpapp02
Page 215: Synchronisation 120610202311 Phpapp02

1

Chapter 12

Boundary and

Transparent Clocks

Page 216: Synchronisation 120610202311 Phpapp02

2

Boundary Clocks

Boundary Clocks reduce PDV

accumulation by:

• Terminating the PTP flow and recovering the reference timing

• Generating a new PTP flow

using the local time reference, (locked to the recovered time)

• There is no direct transfer of

PDV from input to output

A Boundary Clock is in effect a

back-to-back Slave+Master

Q2

Qn

Q1

Clock

Sla

ve

Maste

r

Page 217: Synchronisation 120610202311 Phpapp02

3

1588v2 with Boundary Clocks

MasterClock

SlaveClock

BC BC BC BC

PDV PDV PDV PDV PDV

• BCs recover and re-generate the 1588v2 clocking

• With a network of BCs, PDV contribution (per hop) is only

from BC and link

• PDV experienced by Slave is minimised

Page 218: Synchronisation 120610202311 Phpapp02

4

Boundary Clocks - Specifications

• The ITU-T are working on a performance specification, to appear in G.8273.2

• Considered performance areas:

1) Frequency/Time Accuracy

2) Noise Generation

3) Noise Tolerance

4) Noise Transfer

5) Phase Response

6) Holdover

Page 219: Synchronisation 120610202311 Phpapp02

5

Boundary Clocks - G.8273.2

Conformance Testing - Wander

Noise generation

Noise tolerance

Noise transfer

Frequency Accuracy

Phase transient response

Master ClockMaster Clock Slave Clock

BC

Holdover

Master ClockMaster Clock Slave Clock

~

BC

Master Clock

~

Page 220: Synchronisation 120610202311 Phpapp02

6

Transparent Clocks

Q1

Q2

Qn

Packet Delay in TC Device inserted into correctionField

at output of Transparent Clock device

Transparent Clocks reduce PDV by:

• Calculating the time a PTP packet resides in the TC device (in nsec) and inserting the value into the CorrectionField

• Using the CorrectionField, the Slave or terminating BC can effectively remove the

PDV introduced by the TC

Page 221: Synchronisation 120610202311 Phpapp02

7

1588v2 with Transparent Clocks

End Slave removes PDV Accumulation using

CorrectionField

PDV is written by each TC into CorrectionField and this

accumulates, so CorrectionField = PDV

Accumulation at the End Slave

MasterClock

SlaveClock

TC TC TC

PDV Accumulation

TC

Page 222: Synchronisation 120610202311 Phpapp02

8

Transparent Clocks - Specifications

• Performance specifications will be given in ITU-T G.8273.3

• IEEE C.37.238 specifies less than 50ns error for a TC.

• The performance of a TC is essentially the accuracy of the CorrectionField

• This accuracy must be tested in the presence of traffic for forward path (Sync) and return path (Del-req).

• (Note 1-step and 2-step methods differ in the location of the CorrectionField)

• Tests should be done with varying congestion traffic: Packet

size, traffic priority and utilisation

Page 223: Synchronisation 120610202311 Phpapp02

9

Transparent Clock Test Plan

CorrectionField Accuracy

•Master Clock

~

Traffic Generator

TC•`1`

Master Clock

•`1`Slave Clock

Connect as shown, and

broadcast Traffic Generator traffic on all ports of the TC

1. Capture PDV before and after

TC, with CorrectionField

2. Perform differential calculation

of PDV - Confirm

CorrectionField error is <50ns

3. Repeat under multiple traffic

conditions:1. Vary traffic packet size

2. Vary traffic priority

3. Vary traffic utilisation

Test Procedure to measure

accuracy of CorrectionField

Page 224: Synchronisation 120610202311 Phpapp02

10

Packet /Traffic Generator Errors

~

TC

e.g. GPS

Traffic Generator

Claim: Emulating Master and Slave can verify CorrectionField accuracy

• TC has 50ns error limit (e.g. IEEE C.37.238)

• TC test equipment must have better accuracy – Meeting the ns challenge

• Even synchronised to GPS, hardware architecture of traffic generators has 10’s/100’s of ns error. So these are not fit for purpose

• Paragon is accurate to 5ns and fit for purpose.

Page 225: Synchronisation 120610202311 Phpapp02

11

The 1µs Challenge

An increasing number of applications require accurate transfer of frequency and time through Ethernet networks:

•TDD base stations

– need phase synchronisation to 1µs accuracy

•High Frequency Trading (HFT)

– needs 1µs timestamp resolution

•Power Substations

– specify a maximum time variation through a system of 1µs

•For Mobile Backhaul

– frequency accuracy must be better than 16ppb

Page 226: Synchronisation 120610202311 Phpapp02

12

Test Boundary/Transparent Clocks

Confirming 1µs accuracy

• The 1µs inaccuracy limit is for a network as a whole – but each network element plays a part

• Standards bodies are creating performance specs for BCs and TCs

• Errors in a BC or TC can only be a fraction of 1µs– Standards (e.g. IEEE C.37.238) propose 50ns limits

• Test equipment must be accurate to ns– Traffic Generators/Packet Capture Platforms have 10’s if not

100’s of ns error

Page 227: Synchronisation 120610202311 Phpapp02

13

Network Conditions in the Lab

Base Station

TDM-o-Ethernet

TDM

Core Network

Capture PDV profiles from

live or trial networks . . .

. . . additionally:

Record and replay from G.8261 Reference Network

(Removes repeatability issues and minimizes resource effort.)

Use PDVs from a library of profiles (including G.8261, MEF-18)

Generate pseudo-real world profiles

(edit real-world profiles to test margins of operation and clock

recovery)

Generate theoretical models

(Gamma or Gaussian PDV distributions)

. . . then replay the real-world PDV profiles

back in your lab.

•895.477518885•894.977768110•894.478017335

•Offset=0.002 ppm

•-0.000296349

•0.000645879

•Time Interval•Erro r (TIE)•vs Nominal•(seconds)

•MKR-1:x=894.874691063, y=0.000360070

•MKR-2:x=894.874691063, y=0.000360070•Delta: x=0.000000000, y=0.000000000

•x=895.477518835

•y=0.000444870

•T (a)

Page 228: Synchronisation 120610202311 Phpapp02
Page 229: Synchronisation 120610202311 Phpapp02

1

Chapter 13

Clock Measurements

Page 230: Synchronisation 120610202311 Phpapp02

2

Clock Performance measurements

Page 231: Synchronisation 120610202311 Phpapp02

3

Tie Interval Error (TIE)

TIE (sec)

Page 232: Synchronisation 120610202311 Phpapp02

4

Maximum Time Int Error (MTIE)

Page 233: Synchronisation 120610202311 Phpapp02

5

Time Deviation (TDEV)

1 10 100 1000 10000 100000

Page 234: Synchronisation 120610202311 Phpapp02

6

Modified Allen Deviation (MDEV)

Page 235: Synchronisation 120610202311 Phpapp02

7

G.8271.1 N'twk PDV – Time/Phase

GPS1588v2 Slave

1588v2 Master

pktfilteredMTIE

1588v2

Network

1pps ToD Accuracy

LAB NETWORK

Lab – G.8271.1 profiles stress Slave, 1pps and ToD accuracy measured

Network – Measure Network PDV pktfilteredMTIE to G.8271.1/G.8260(available on ratification of G.8271.1)

Page 236: Synchronisation 120610202311 Phpapp02

8

G.8273.2 – Boundary Clock Test

1588v2 + PDV + Traffic

GPS

BC

Ref

1588v2 + PDV + Traffic

1pps ToD

Noise generation

Noise tolerance

Noise transfer

Frequency+Time Accuracy

Pull-in, Pull-out, Hold-in

BC

Ref

1pps ToD

1588v2

Phase transient response

Page 237: Synchronisation 120610202311 Phpapp02

9

G.8273.3 – Transparent Clock Test

1588v2 + PDV + Traffic

GPS

TC

Ref

1588v2 + PDV + Traffic

Measure Latency vs Correction Field

Page 238: Synchronisation 120610202311 Phpapp02
Page 239: Synchronisation 120610202311 Phpapp02

1

Chapter 14

CALNEX Testing

Page 240: Synchronisation 120610202311 Phpapp02

2

Testing E1/2.048MHz

PRC

Node-B

MTIE/TDEV against

G.823 PDH Interface

masks

E1 Clock out

Page 241: Synchronisation 120610202311 Phpapp02

3

Testing 10MHz

Calnex Freq

Converter

PRC

Node-B

MTIE/TDEV against

G.823 PDH Interface

masks

10MHz Clock out

Page 242: Synchronisation 120610202311 Phpapp02

4

Testing Sync-E

PRC

MTIE/TDEV against

G.8261 Opt1 mask

Node B

Page 243: Synchronisation 120610202311 Phpapp02

5

Testing 1588v2

GPS

Node-B

1588v2 Reverse PDV

1588v2 Forward PDV

1588v2

Master

(RNC)

1588v2 Protocol Analysis

1588v2 pktMTIE (future)

Delay Distribution

Page 244: Synchronisation 120610202311 Phpapp02

6

How to measure Forward PDV

895.477518885894.977768110894.478017335Offset=0.002 ppm

-0.000296349

0.000645879

Time IntervalErro r (TIE)vs Nominal(seconds)

MKR-1:x=894.874691063, y=0.000360070

MKR-2:x=894.874691063, y=0.000360070Delta: x=0.000000000, y=0.000000000

x=895.477518835

y=0.000444870

T (a)

t-ms

Follow_Up

Sync

t-sm

Delay_Req

Delay_Resp

Propagation Delay

Message Exchange

t1

t4

t2

t3

Sync PDV

Master Clock

Slave Clock

Does Master > Slave PDV impact clock recovery?Does Master > Slave PDV impact clock recovery?

• Paragon uses timestamp from Sync message to calculate Sync PDV

• Sync PDV = variation of (arrival time at Paragon – timestamp)

Page 245: Synchronisation 120610202311 Phpapp02

7

How to measure Reverse PDV

895.477518885894.977768110894.478017335Offset=0.002 ppm

-0.000296349

0.000645879

Time IntervalErro r (TIE)vs Nominal(seconds)

MKR-1:x=894.874691063, y=0.000360070

MKR-2:x=894.874691063, y=0.000360070Delta: x=0.000000000, y=0.000000000

x=895.477518835

y=0.000444870

T (a)

t-ms

Follow_Up

Sync

t-sm

Delay_Req

Delay_Resp

Propagation Delay

Message Exchange

t1

t4

t2

t3

Master Clock

Slave Clock

Delay_Req PDV

• Measure variation between the launch time of the Delay_Req message (arrival time of Del_Req in Paragon) and the embedded timestamp, t4, in the corresponding Delay_Resp message

Does Slave > Master PDV impact clock recovery?

Page 246: Synchronisation 120610202311 Phpapp02

8

Simultaneous Measurements

GPS

Node-B

1588v2 Reverse PDV

1588v2 Forward PDV

Hybrid

1588v2

Master and

Sync-E EEC

1588v2 Protocol Analysis

ALL AT THE SAME TIME!

Sync-E MTIE / TDEV

ESMC Monitor / Overwrite

E1/T1 MTIE / TDEV

ESMC Monitor / Overwrite

Sync-E MTIE / TDEV

1pps ToD Accuracy

1pps ToD Accuracy

E1/T1 MTIE / TDEV

1588v2 Reverse PDV

1588v2 Forward PDV

1588v2 Protocol Analysis

Delay Distribution

Delay Distribution

Page 247: Synchronisation 120610202311 Phpapp02

9

Calnex Products

Portable GPS Rubidium reference

• Provides stable frequency (10MHz) and 1pps reference in the field

• GPS Antenna and cables included

Paragon

• Lab & Field up to 1G

• 1588v2, SyncE, CES, NTP,

OAM

Paragon-X

• Lab & Field up to 10G

• 1588v2, SyncE, CES, NTP,

OAM

• E1/T1 and 1pps

Paragon-m

• Field up to 1G

• 1588v2, SyncE, CES, NTP,

OAM

• E1/T1 and 1pps


Recommended