Network Access Transmission Model for Evaluating xDSL Modem Performance

1
Network Access Transmission Model for
Evaluating xDSL Modem Performance

• Jack Douglass, Paradyne International
• Chair TIA TR30.3
• Sept 5, 2002, FS-VDSL
• TR30.3 302090064
• FS0261
• jackdouglass_at_hotmail.com

2
Presentation Overview

• Purpose of Presentation to FS-VDSL
• Access Network Models Projects for Evaluating
  xDSL Modem Performance
• Value of xDSL Network Model
• Network Model Overview
• Access Network Simulator
• Advantages of NMC Methodology
• Obstacles to Creating a European Network Model
  that implements NMC methodology
• Proposal for Creating European Network Model
• Discussion

3
Purpose of Presentation

• Establish a formal liaison between FS-VDSL and
  TR30.3
• Request that FS-VDSL open a project to assist in
  the develop of a European xDSL Network Model that
  uses Network Model Coverage (NMC) methodology
• TIA/EIA-876, North American Network Model, has
  been released for publication
• ETSI has opened a project in the form of a
  Permanent Document (ETSI TM6 PD (02) 07) to
  create an European Network Model
• Initial model would be for frequency range of 0
  to 1.104 MHz
• Principles can be applied to VDSL

4
Purpose of Presentation (continued)

• Work with FS-VDSL to acquire European Network
  Statistics (e.g., loop, crosstalk, ingress, etc.)
  that are needed to build the Network Model(s)
• In the past FS-VDSL has be able to reduce
  obstacles such as lack of publicly available
  information regarding loop and crosstalk
  statistics, unbundling competition issues,
  regulation issues
• Provide a template that can be used to develop
  European Network Model(s)
• Determine the best way to forward FS-VDSL/TR30.3
  work to ETSI TM6
• Contribution to ETSI TM6 concerning Network
  statistics for development of Network Model(s) in
  TM6 may be anonymous
• Contribution to ETSI TM6 that has completed
  Model(s) for inclusion into appropriate ETSI
  document (e.g., Technical Report) may be
  anonymous

5

• Access Network Model Projects
• for Evaluating xDSL Modem Performance

6
TIA/EIA-876 North American Network Model

• Uses Network Model Coverage (NMC) methodology and
  principles to evaluate and compare the
  performance of xDSL modems
• Statistically based portrait of the access
  network and impairments
• Loop Model based on a combination of loop
  surveys, including an anonymous 14 million line
  survey
• Central Office wiring models
• Central Office impairments (e.g., CEXT, Composite
  CEXT)
• Intermediate xDSL interferers
• Customer Premises drop and wiring models
• Customer Premises impairments (e.g., RFI, POTS
  signalling, splitters/filters, AC induced
  interference)
• Definitive set of loop and noise conditions for
  consistent and repeatable test results

7
TIA/EIA-876 North American Network Model

• Models DSL access network over the frequency
  range of 0 to 1.104 MHz
• Technology independent
• Can be applied to both splittered and
  non-splittered xDSL systems
• Principle of TIA/EIA-876 readily apply to higher
  frequencies such as used by VDSL
• Crosstalk, Ingress, Loop Models may require some
  modification
• Test equipment must be able to support the
  desired operating range
• Model to be used by Network Service Providers,
  PTTs, test houses, magazines, product reviewers,
  users and designers
• TIA/EIA-876 is intend to compliment the existing
  xDSL testing standards
• It is a performance test not interoperability or
  conformance test

8
TIA/EIA 876 Network Model
xDSL Network Block Diagram with Impairment
Injection Points
9
ETSI TM6 European Network Model

• Project opened in the form of a Permanent
  Document to create a European xDSL Network
  transmission Model based on the NMC principles
  and methodology

10

• Value of xDSL Network Model

11
Value to Operating Companies and Service
Providers

• Predict candidate product performance on their
  networks as Percentage of the network where
  satisfactory operation will be obtained
• Determine the potential market coverage as a
  function of different parameters/factors such as
  Quality of Service, line rate, data throughput,
  connect time, stability, technology, modulation
  technique and modem enhancements
• Select optimum technology for a proposed service
  based its Network Model Coverage Performance
• Develop Business Cases and establish Tariff
  objectives
• Minimize costs associated with loop
  qualification, loop modifications and truck rolls

12
Value to Manufactures and Design Engineers

• Helps find design weaknesses
• Facilitates isolating and resolving field
  problems
• Assists in evaluating different technologies
• Predicts real access network performance

13
Comparison Testing

• Model can be used by test houses, magazines and
  product reviewers to compare the performance of
  different brands of xDSL modems or systems
• Test results are intended to reflect the customer
  experience

14

• Network Model Overview

15
Example General European Access Network Model

• Cable lengths and types are intended as a basis
  for discussion.
• Intermediate noise injection (Remote DSLAM) point
  may not be necessary.

16
Cumulative Distribution for Crosstalk Models

• Cumulative Distribution Values
• Basis for the crosstalk mix used in Crosstalk
  Impairment Combination Tables
• Residential/Multiunit Model
• asymmetrical weighting
• Business Model
• symmetrical weighting
• Projected for the year 200x
• Current xDSL deployment statistics
• Projected xDSL deployment
• Assumes 25 (?) -pair binders with yy vacant
  pairs
• Churn/disconnect cross-connected at street
  cabinet to reserve loop assignment for the next
  tenant
• Defective pairs
• Reserved for future growth

17
Residential/Multiunit Cumulative Distribution
(CD) Number of Disturbers of Each Type
18
Business Cumulative Distribution (CD) Number of
Disturbers of Each Type
19
Crosstalk Impairment Combinations

• Crosstalk Impairment Combinations (IC) are
  specified for each Loop
• Residential/Multiunit model
• Business model
• A, B, C and D Crosstalk severity levels
• A Most severe
• D Least severe
• LOOs A 5, B 15, C 30 and D 50 (Total
  100)
• FEXT may be handled differently in mathematical
  analysis and hardware simulation
• Hardware simulator
• NEXT is inserted at both ends so that tests can
  be run in both directions simultaneously
• Insertion of NEXT at one end of the loop produces
  an approximation of FEXT at the other end
• Mathematical analysis
• FEXT should be included at both ends
• Assumes Worst-case crosstalk coupling
• Disturber Model may vary between Exchange and CPE
  end
• CPE Crosstalk is xx co-located and yy
  distributed
• Distributed crosstalk may be do to operating
  range of some system is less than the loop can
  accommodate
• Crosstalk may be distributed as a result of
  distributing services to other customer along the
  way.

20
Crosstalk Impairment Combinations (IC) Loop XX
(LOO/Length) Residential/Multiunit
21
Crosstalk Impairment Combinations (IC) Loop XX
(LOO/Length) Business
22
Specified Steady-State Impairments

• Specified Steady-State Impairment Combinations
  Severity levels 0 - 3
• Primarily ingress noise
• Severity 0 is a baseline null case
• Severities 1 through 3 have increasing levels of
  ingress noise
• Do not have an associated LOO

23
AM Radio Interference

• Severity level 1, 2, and 3

24
Specified Transient Impairments

• Not part of the NMC calculation
• Important part of the Access Network Transmission
  Model
• Must be accounted for in testing

25
Example General Loop Diagram
26
Example Test Loop Make-up and LOOs
Loop Loss values _at_ 100 kHz and _at_ 300 kHz are
approximate and assume same cable type is used
for entire length
27
Premises Wiring Models

• Based on G.996.1, section 6.2.2
• Single Family and Small Office Premises Models
• Daisy Chain Wiring
• Star Wiring
• Star Wiring with Central ADSL Splitter and Direct
  Line
• Multi-Unit/Business Wiring
• Multi-Tenant Residence / Business -- Daisy Chain
  Wiring
• Multi-Tenant Residence / Business -- Star Wiring
• Small Office Wiring
• Large Office Wiring

28
Example Customer Premises Models Based on
G.996.1, section 6.2.2
Daisy Chain Wiring Model
29
Network Model Coverage Tables

• Tables for Network Model Coverages (NMC) of 100,
  95, 90 and 65 are typically provided
• Used for both Residential/Multiunit and Business
  Models
• Test Channel Score
• intersection of the IC and test loop
• Score is Product of Loop LOO and IC LOO
• lt 100 NMC Tables
• Remove Loop/IC combinations with lower percentage
  Scores
• Run on Test Channels that have Scores
• Reduces the test time with slightly reduced
  resolution

30
Network Model Coverage Tables
31
Example Network Model Coverage Tables

• Examples NMC100 and NMC90 Tables are provided
  to illustrate how to construct and use NMC Tables
• Arbitrary values have been assigned to the loop
  LOO, so that the example test channel scores can
  be calculated
• A Test Channel Score is calculated by taking the
  product of the loop LOO and the IC LOO
• All Test Channels are included in an 100 NMC
  Table
• Lower percentage scores have been removed from
  90 NMC Table (actual total score is 90.05)
• Actual NMC Table can be constructed once the Loop
  LOOs have been assigned based on loop network
  statistics

32
Example Network Model Coverage 100
33
Example Network Model Coverage 90
34
Test Procedure and Network Model Coverage (NMC)
Curves

• Run each test channel (that has an associated
  score), in the NMC Table along with Specified
  Steady-State Impairment Severity 0 (null case)
  and one of the Premises Wiring Models. Note The
  number of tests can be reduced by using a lower
  percentage NMC Table.
• Measure desired parameter(s) (e.g., connect rate,
  throughput, connect time, etc.).
• Repeat each test channel with Specified
  Steady-State Impairment Severities 1 through 3.
  Tests may also be repeated with different
  Premises Wiring Models and/or Specified Transient
  Impairments.
• Sort measured parameter(s) along with associated
  NMC Scores in a descending order using a
  spreadsheet or similar mechanism.
• Plot the measured parameter(s) on the Y axis and
  the associated NMC Score on the X axis.
• The resulting curve shows the performance (in
  terms of the measured parameter) as a percentage
  of the Network Model.

35
Family of 65 NMC Curves for Steady-State
Impairments Severity 0 to 3
36

• Access Network Simulator

37
Network Model Simulator Implementation

• Network Model Simulators
• Mathematical Simulator
• Hardware Simulator
• Ideal Network Model Simulator
• Separate Loop sections
• Separate Noise sources
• Practical and Cost-Effective Simulator
• Single loop simulator
• Exchange wiring
• Distributed Cable
• Branch Cable
• Drop wire
• Composite Exchange Interferers and the Composite
  CPE Interferers
• FSAN mixed crosstalk combination method
• Account for associated loop sections
• Account for noise injection points.
• Typically use Arbitrary Waveform Generator (AWG).
• Premises wiring simulator
• Device(s) Under Test (DUT).

38
Ideal Network Model Simulator
39
Practical and Cost-Effective Network Model
Simulator
40
Typical Test SetupxDSL Simulator and Modems
xDSL CPE Modems (ATU-R)
Telephone Network Simulator (Line Current/Dial
Tone) ADSL only
xDSL DSLAM s (ATU-C)
Loop Simulator
Premises Wiring Simulator
AWG
AWG
41
Screen of Arbitrary Waveform Generator (AWG)
showing Crosstalk Impairment on CO Side
Uses Loop and Crosstalk transfer functions to
accurately simulate impairment combinations
42
Screen of Arbitrary Waveform Generator(AWG)
showing Crosstalk and RFI Impairment on CPE Side
Uses Loop and Crosstalk transfer functions to
accurately simulate impairment combinations
43

• Advantages of NMC Methodology

44
Advantages of NMC Methodology

• Predicts the performance of the system/modem over
  the real access network
• Statistically accurate indication of overall
  performance based on operation over good, medium
  and worst case loop and noise conditions
• Evaluates more than just stress conditions
• Test results are displayed as a family of NMC
  Curves
• Performance differences between products or
  technologies can be easily seen
• NMC testing can be viewed as running many
  individual SNR points over a wide range of loop
  and noise conditions
• NMC methodology which was implemented in both TIA
  Standards and ITU Recommendations was a key
  factor in improving the quality and performance
  of voiceband modems, over the years
• NMC principles can do the same for xDSL Technology

45
Advantages of NMC Methodology

• Operating Companies and Service Providers
• Predict candidate product performance on their
  networks as Percentage of the network where
  satisfactory operation will be obtained
• Determine the potential market coverage as a
  function of different parameters/factors such as
  Quality of Service, line rate, data throughput,
  connect time, stability, technology, modulation
  technique and modem enhancements
• Select optimum technology for a proposed service
  based its Network Model Coverage Performance
• Develop Business Cases and establish Tariff
  objectives
• Minimize costs associated with loop
  qualification, loop modifications and truck rolls

46
Advantages of NMC Methodology

• Manufacturers and Design Engineers
• Predict real access network performance
• Find design weaknesses
• Isolate/resolve field problems
• Evaluate different technologies

47
Advantages of NMC Methodology

• Model to be used by test houses, magazines and
  product reviewers to compare the performance of
  different brands of xDSL modems or systems
• Test results are intended to reflect the customer
  experience

48

• Proposal
• for Creating European xDSL Network Model

49
Proposal

• Establish a formal liaison between TR30.3 and
  FS-VDSL to develop a European xDSL Network Model
  that uses Network Model Coverage (NMC)
  methodology
• FS-VDSL Committee opens a Network Model Project
• Study NMC Methodology
• Acquire European Network Statistics (e.g., loop,
  crosstalk, ingress, etc.) that are needed to
  build the model(s)
• Assist in developing a European xDSL Network
  Transmission Model(s) based on NMC principles and
  sample templates
• Initial model would not include VDSL
• Jointly determine the best method to forward
  FS-VDSL / TR30.3 work to ETSI TM6
• Contribution to ETSI TM6 concerning Network
  statistics, so a Network Model(s) can be
  developed in ETSI may be submitted anonymously
• Contribution to ETSI TM6 that have Model(s) for
  inclusion into appropriate ETSI document (e.g.,
  Technical Report) may be submitted anonymously

50
Key Committees and Role

• FS-VDSL
• Vehicle for acquiring network statistics and
  creating network model
• Ad hoc meetings to analyze network statistics and
  draft document
• TIA TR30.3
• Experience in creating network model
• Liaison/Work with FS-VDSL and ETSI TM6 to create
  network model
• Work on drafting the model during TR30.3 meetings
• Possible vehicle to anonymously submit final
  network model
• ETSI TM6
• European Access Network Model

51
Obstacles

• Country to country variations of
  loop/crosstalk/noise statistics and
  characteristics
• Lack of publicly available information regarding
  loop/crosstalk statistics
• Unbundling Competition
• Regulations

52
Proposed Procedure

• Create straw-man Network Model(s) using sample
  template and experience
• Gather statistical information on European Access
  Network
• Loop (configuration, binder size, type of cable,
  gauge, etc.)
• Crosstalk data (numbers and types of interferers
  currently installed and marketing deployment
  information)
• Steady-State Impairments (e.g., Ingress
  impairments, AM Radio, etc.)
• Transient Impairments
• Revise straw-man Network Model(s) based on
  statistical information
• Validate model using real xDSL equipment of
  different technologies
• Compare validation results with known real world
  performance
• Submit Model(s) and/or network statistics to ETSI
  TM6 for possible inclusion into an appropriate
  ETSI document (e.g., Technical Report)
• may be submitted anonymously

53
Key FS-VDSL Contributors

• Operating Companies
• Chip manufacturers
• 8 Companies supported opening the project at the
  ETSI TM6 meeting

54
Time Table

• Time table is mainly dependent on how quickly
  statistical information can be obtained
• Model can be created fairly quickly using sample
  template
• Time to create model can be reduced by having ad
  hoc meetings and working on the document at the
  FS-VDSL and TR30.3 committee meetings.

55
Discussion

• Determine how work can be applied to a VDSL
  version of the model
• Comments, Suggestions and Recommendations