Professional Documents
Culture Documents
HP-AN1258 - Evaluating Tributary Jitter From The SDH Network
HP-AN1258 - Evaluating Tributary Jitter From The SDH Network
3
Analysis of the network In order to achieve this with the Also, as a further result of the
32 island model, the maximum experimentation, it has become
In order to specify the jitter limits jitter from a PTE is limited to clear that test methodology
on a PTE, analysis has been 1.3UI. Table 1 shows how this guidelines need to be produced in
performed to predict the expected budget has been allocated be- order to achieve accurate and
jitter accumulation that might tween mapping jitter, single repeatable results.
occur as a tributary signal passes pointer movements and degraded
through multiple SDH islands. The synchronization conditions. Liasion between ANSI and ITU has
objective of this study is to prompted ITU to also review the
produce a model that represents a pointer test sequences. Study
practical worst-case example of a Characterization of jitter Group 13 is, at the time of printing,
network that may be used to performance considering how the sequence in
transfer a PDH signal. {A similar G.783 should be expanded/modi-
study carried out by Bellcore for G.783 presently includes pointer fied.
the SONET world produced a 32 test sequences [3]. These se-
SONET island model, each of quences are aimed to emulate
which contains 10 pointer expected network degradations.
processing nodes [1], (Figure 1).}
During 1992, Telecom Canada
To ensure that the jitter accumula- carried out testing to verify the
tion does not cause service theoretically predicted responses
degradation at the output of the to various types of pointer activity
last SDH island, the total on PTEs [4]. As well as verifying
network jitter must not exceed the theoretically predicted re-
that specified for the tributary sponses to variations in pulse
rate [2]. Therefore, each PTE must stuffing ratios (used to map the
not only meet this specification tributary signal into the VC), and
but will have to exhibit a far to single pointer movements,
better performance if the jitter at Telecom Canada also showed that
the output of the last SDH island the defined tests did not fully
is to meet this requirement. represent the pointer sequences
that a real network might
Once the size and structure of the produce.
network model has been agreed,
the allocation of the amount of With the results from the practical
jitter which can be generated experimentation and the specifica-
by the various jitter-producing tion of jitter performance in terms
effects will be performed to of three network conditions, (an
ensure that the total network jitter example of which is shown in
does not exceed the specified Table 1), ANSI have reviewed the
limit on a tributary signal. As an pointer movement sequences [5].
example of the order of magnitude The aim was to produce tests that
that is likely to be settled upon, more closely emulated real
ANSI tackled a similar problem network conditions and also allow
for the DS3 interface which measurement of the specified
requires that the peak-to-peak jitter thresholds.
jitter shall not exceed 5UI, (in the
10Hz to 400 kHz range).
4
Figure 1: Hybrid network model produced for the transfer of DS3 through a SONET Network. ITU Study Group 13 is investi-
gating the generation of a similar model for each of the CEPT rates.
Island 1
Pointer Processing
A
DS3 M D A
OC-N OC-N
Jitter
Free Ptr
Adj Jitter
Network
Output Jitter
Island 2 Island 32
SONET Jitter
M D SONET
Island M D DS3
Island A
Table 1: ANSI jitter specification for a DS3 signal demultiplexed from a single SDH/SONET island.
5
Figure 2(a) : Single isolated pointer test.
Single
Pointer
Movement
500 µs
µ 500 µs
µ
The new test sequences The second criterion defines the Figure 2(b) shows the test defined
response to a Single Isolated to cover this condition. Like the
The test sequences developed by Pointer movement. This is tested single pointer test, the bursts are
ANSI have already been accepted using a sequence which has a set 30 seconds apart to allow time
for inclusion in G.783 for the single pointer movement every for the effects of each burst to die
North American portions of SDH. 30 seconds, Figure 2(a). The out before the next burst is
It is likely that similar network pointer movements are spaced 30 applied.
scenarios will be considered by seconds apart in order to allow
the Study Group in order to time for the effects of each move- The second test emulates the
produce test sequences for the ment to completely die out before network condition when the
CEPT-mapped signal. another pointer movement is originating PTE loses lock to the
applied. This ensures that the system clock. This condition will
The first of the new test resultant jitter measured is that cause continuous pointer move-
sequences was developed to produced by a single pointer ments to be generated. On top of
emulate the situation when there movement. this background, extra pointer
is no pointer activity. This test activity may occur due to phase
requires the frequency of the Two test sequences have been noise from the other nodes of a
tributary signal to be varied to defined to measure the perform- network. Therefore, on top of the
find the maximum jitter produced ance of a PTE in a Degraded background of continuous pointer
due to the pulse stuffing ratio Synchronization Condition. movements, an added or canceled
used to map the tributary signal The first test emulates the net- perturbation to the background
into the VC, ie, Mapping Jitter. work condition where phase noise sequence is performed every
This test checks the specification in a chain of network elements 30 seconds.
of the maximum jitter on the (eg, ADMs, cross-connects, etc)
output due to the process of accumulates to produce a burst of
extracting the tributary signal pointers with minimum spacing.
from the VC.
6
In the original sequences, the Similar effects can be predicted Table 2:
clock synchronization loss was for all types of SDH VCs. It is Experimental results showing effect of
87/3 sequence.
represented by continuous evenly possible by the use of sophisti-
spaced pointer movements. cated pointer processing nodes to
Jitter on
However, this is not always the remove this effect and produce Pointer demultiplexed DS3
case in practice. Experimentation regular, evenly spaced pointer sequence (UIp-p)
has shown that gaps are generated movements. However, this is not
G.783 test:
in the pointer sequence due to the specified as a requirement for the regular pointers 0.15
effects of the positioning of the network equipment, and as the with one missing
SDH overhead bytes. For exam- most straight forward techniques pointer
ple, when an VC-4 is cross-con- produce this effect, PTEs will 87/3 sequence 0.60
nected, a repetitive sequence of have to be designed to cope with
87 evenly spaced pointer move- this sequence. Notes
ments followed by a gap equiva- 1. In both cases, the pointer spacing
lent to 3 missing pointer move- Figure 2(c) shows the new se- was set at 33 ms, equivalent to a
ments is generated. Table 2 quence that has been defined to frequency offset of 4.6 ppm.
shows the effect on the jitter emulate the clock synchronization 2. The results were obtained by testing
the DS3 drop port from the
output from the 87/3 sequence loss condition for DS3 testing. HP 37704A SONET test set.
compared to that produced by the
original G.783 test.
Pattern 3
Pattern 2
T
t = 500 µs
AAAA AAAA
AAAA AAAA
AAAA AAAA
AAAA AAAA AAAA AAAA
AAAA AAAA
AAAA AAAA
AAAA
AAAAAAAAAAAAAAAA
AAAA
AAAA AAAA
AAAAAAAA
Group A Group A Group A Group A Group A
7
Figure 3: Typical block diagram of a E3 PTE.
Front-End Internal E3
Receive Pointer STM-n Demultiplexer Drop
STM-n Processing Block E3
Block
Recovered
E3 Clock
Internal
STM-1
Oscillator
TX RX
RX STM-n E3
STM-n
CLK
MTS
MTS
Clock CLK
RX
STM-n RX
TX E3
STM-n
Clock Extracted
and used to lock
Test Signal
8
Test methodology The second effect of this structure References
comes from the elastic store
As a side-effect of the practical which will be present in the front- 1. T1X1.2, “SONET Hypothetical
experimentation being carried end pointer processing block. Reference Circuit (HRC)",
out, it became clear that there was This store will absorb some of the T1X1.2/93-015, March 1993.
a need to clarify the test pointer activity in the received
methodology in order to obtain STM-n signal before producing 2. ITU Recommendation G.823,
accurate and repeatable measure- any pointer movements in the " The Control of Jitter and
ments. Factors like the number of internal STM-n signal. (Cases have Wander within Digital Net-
repetitions of each pointer been observed where as many as works which are based on the
sequence test are being 15 movements are absorbed.) 2.048 Mbit/s Hierarchy".
considered to produce implemen- If a pointer sequence test is Vol III - Fascicle III.5, Blue
tation guidelines for each test performed, it is necessary to Book.
sequence. Two of the most prime this elastic store to ensure
important requirements arise from the pointer sequence applied to 3. Revised Recommendation
the structure of the PTEs avail- the PTE is that which appears at G.783, CCITT COM XV-R
able today. the input to the demultiplexer 110-E, Nov'92.
block. (This can be achieved by
By testing PTEs, it has been found applying a test signal containing 4 K. Mahon, “Significance of
that many contain front-end continuous pointer movements of Telecom Canada’s SONET
pointer processing blocks that the same polarity as those used Jitter Accumulation Measure-
extract the VC-n from the received during the testing until jitter ments”, Telecom Canada,
signal and pass it into an inter- spikes are detected on the tribu- T1X1.3/92-129, November 1992.
nally generated STM-n before the tary output each time a pointer
signal is demultiplexed to extract movement occurs in the test 5. ANSI, "Synchronous Optical
the tributary signal, Figure 3. signal.) Network (SONET): Jitter at
Network Interfaces", T1X1.3/
For effective testing of a PTE 93-006R3, Aug. '93.
with this architecture, it is impera- Summary
tive that the frequency of the
internal STM-n signal is locked to By tightly specifying the jitter
the frequency of the test signal generated by PTEs before there
being applied to the PTE. is significant deployment of SDH
If this is not accomplished, equipment, network operators can
frequency and/or phase differ- avoid major problems once the
ences between the two STM-n multiple SDH island scenario
signals will produce pointer becomes a reality. The definition
activity on the internal STM-n of sequences which simulate real
signal. These pointer movements network conditions, accompanied
will produce spurious spikes of by test methodology guidelines,
jitter on the tributary signal. will allow network operators to
Figure 4 shows two possible test gain confidence that equipment
configurations which will synchro- they are installing in their net-
nize the test signal to the PTE and works will interwork with the
hence avoid this problem. existing asynchronous networks
both now and in the future.
9
Appendix A
PTE NE NE PTE
10
Test set emulation model The defined pointer test Summary
sequences consist of repetitive
The reduced network model patterns of pointer movements, By defining a test set transmitter
which must be emulated in order all of which are the same polarity. in terms of a network emulation
to accurately reproduce signals This implies that an offset exists model, it is possible to highlight,
containing pointer movements, between the line rate (f3) and VC-4 and hence ensure reproduction of
consists of a PTE plus a network rate (controlled by f2). As the the important characteristics of
element which provides pointer definition for the sequences signals containing pointer move-
processing functions. Figure A2 requires that the line rate remains ments. This model clarifies the
shows the emulation model. In constant, the internal STM-n (f2) need for the pulse stuffing process
this model, the pulse stuffing rate rate must change to generate these to be independent from the gen-
is controlled by the relative pointer movements. Changing f2 eration of pointer movements.
frequency of the E4 rate (f1) to the will also cause a step change in The implementation of this emula-
internal STM-n line rate (f2). The the pulse stuffing ratio. A step tion model in a SDH test set will
rate of pointer additions is con- change of this nature would not ensure accurate production of the
trolled by the relative frequency of occur in the network as line stimuli required to measure the
the internal STM-n line rate (f2) to frequencies are restricted by tributary jitter performance of a
the output STM-n line rate (f3). network equipment specifications terminating PTE.
With this model, (as in the real from suddenly changing rate. The
network), when pointer move- cooling down period defined at the
ments are introduced, there will start of the sequences provides
not be a step change in the pulse time for the effects of the step
stuffing rate around the pointer change to die out before the jitter
movement to counteract the measurements are performed.
apparent step change in VC-4 rate.
PTE NE
(Pointer
STM-n
E4 Processing STM-n
Functions)
f1 f2 f3
11
Appendix B
Measuring tributary jitter This table clearly shows the need This means that if it is more than
out of an SDH network to test SDH demultiplexing 200 ms between the peak values,
equipment with test signals that then the actual peak-to-peak jitter
SDH tributary jitter contain pointer movements in will not be measured.
characteristics order to accurately characterize its
performance. 2. The bandwidth of the circuit
The innovation of using pointers to used to recover the carrier
track the position of the Virtual Traditional jitter frequency is such that the
Container (VC) within SDH signals measurement circuits recovered clock phase will track
has produced many benefits that transients that last as long as that
will minimize the cost and The jitter that has existed in PDH shown in Figure B.2. When this
complexity of network equipment. networks prior to the introduction signal is compared to the input
However, due to the large inherent of SDH has been sinusoidal in signal in order to measure the
phase step associated with a nature. Jitter measurement jitter, it will not be possible to see
pointer movement (eg, 24UI per circuits, therefore, were produced the true magnitude of the jitter due
VC-4 pointer movement), that were capable of accurately to the tracking effect in the
compared to that produced by measuring the peak-to-peak jitter reference clock being used.
pulse stuffing techniques used in of repetitive waveforms. Figure
asynchronous multiplexing, the B.1 shows a typical measurement
SDH network has the potential of technique that requires the jitter
creating large jitter transients in waveform to be repetitive.
the demultiplexed tributary
outputs. Not only are the phase The repetition rate of the jitter
steps much larger, the transients in the pure PDH
characteristics of the jitter network was relatively high (eg, Table 1:
produced are completely different 20 Hz for 2.048 Mbit/s). The carrier Experimental results highlighting the
from these previously experienced frequency recovery circuit, relative effect of pointer movements on
the tributary jitter.
in the PDH network. therefore, would track any phase
changes at rates less than this
The effect of pointer bandwidth.
movements
Jitter
SDH tributary jitter Network condition measured
Experimentation has confirmed characteristics (UI p-p)
that the jitter created at a
demultiplexer output is A single pointer movement No pointer movements 0.09
significantly greater when the SDH equates to an instantaneous phase
signal arriving at the Path step. This causes the output phase Single pointer movements 0.12
Terminating Network Element of the desychronizer to respond in The "87/3" sequence 0.60
(PTE) contains pointer a similar manner to that shown in produced when an node
movements. One natural Figure B.2. Two effects stop the is offset by 4.6 ppm
phenomenon known as the "87/3" traditional measurement circuit
sequence (see Note, Page 14), from giving a true measure of the
which occurs in the VC-4 pointer peak jitter:
when a node within the SDH
network loses timing 1. Positive and negative peaks
synchronization, can have the need to occur within 200 ms of
effect of generating large jitter one another, ie, the peak-to-peak
transients. The test results shown value is the difference between the
in Table 1 highlight the relative positive reading and the negative
magnitude of jitter that can result reading during a measurement
from each effect. period.
12
Figure B.1: Jitter measurement in a PDH network.
Jitter Threshold
100 ms 100 ms
x
Search for Positive Peak Search for Negative Peak
Figure B.2: Input phase and desychronizer response in the presence of a single pointer movement.
Desync Input
Phase
Desync Output
Phase Response
24-
-
18-
-
Phase
(UI) 12-
-
6-
-
0-
-3-
1 2 3 4 5 6 7 8 9 10
Time (seconds)
13
HP 37717A jitter 1.The voltage-controlled oscillator 3. The maximum peak-to-peak
measurement circuit (VCO) used to track the carrier jitter displayed by the instrument
frequency signal being received will be the difference between the
The jitter measurement module for has a bandwidth of 4 Hz. This will positive and negative peaks within
the HP 37717A PDH/SDH BER and provide a better reference for the the gating period, ie, the use of
jitter test set is the first in a new measurement of long transients fixed measurement periods within
generation of jitter measurement produced by pointer activity. the gating period will no longer be
implementations. To allow employed.
accurate measurement of jitter 2. Separate positive and negative
created by the SDH network into peak detectors will ensure that all By the use of these techniques, the
the PDH network, the following spikes of jitter are captured HP 37717A provides a jitter tester
techniques have been employed: irrespective of when they occur that will give accurate measure of
within the measurement period. jitter in a PDH signal, irrespective
of the network topology that the
signal has passed through.
Note
14
15
For more information on Hewlett-Packard
Test & Measurement products, applications or
services please call your local Hewlett-
Packard sales offices. A current listing is
available via Web through AccessHP at http://
www.hp.com. If you do not have access to the
internet please contact one of the HP centers
listed below and they will direct you to your
nearest HP representative.
United States:
Hewlett-Packard Company
Test and Measurement Organization
5301 Stevens Creek Blvd.
Bldg. 51L-SC
Santa Clara, CA 95052-8059
1 800 452 4844
Canada:
Hewlett-Packard Canada Ltd.
5150 Spectrum Way
Mississauga, Ontario
L4W 5G1
(905) 206 4725
Europe:
Hewlett-Packard
European Marketing Centre
P. O. Box 999
1180 AZ Amstelveen
The Netherlands
Japan:
Hewlett-Packard Japan Ltd.
Measurement Assistance Center
9-1, Takakura-Cho, Hachioji-Shi
Tokyo 192, Japan
Tel: (81-426) 48-0722
Fax: (81-426) 48-1073
Latin America:
Hewlett-Packard
Latin American Region Headquarters
5200 Blue Lagoon Drive
9th Floor
Miami, Florida 33126
USA
(305) 267 4245/4220
Australia/New Zealand:
Hewlett-Packard Australia Ltd.
31-41 Joseph Street
Blackburn, Victoria 3130
Australia
1 800 629 485
Asia Pacific:
Hewlett-Packard Asia Pacific Ltd.
17-21/F Shell Tower, Times Square
1 Matheson Street, Causeway Bay
Hong Kong
Fax: (852) 2506 9285
Printed in UK (5/94)
Data subject to change
5962-9551E