Calix xDSL Best Practices

June 2014

#220-00427, Rev. 16

2

Contents About This Guide.................................................................... 11 Chapter 1: Introduction .......................................................... 13 About Best Practices .............................................................................................. 14 IPTV Network Model ................................................................................................ 15

Chapter 2: Underlying Technology of xDSL ........................ 17 DMT ........................................................................................................................... 18 QAM .......................................................................................................................... 19 Subcarriers and Tones.................................................................................... 20 Upstream vs. Downstream.............................................................................. 20 Packet Transfer Mode (PTM).......................................................................... 21 xDSL Protocols ........................................................................................................ 21 ADSL (G.992.1) .............................................................................................. 22 ADSL2 (G.992.3) ............................................................................................ 22 Loop Diagnostics ............................................................................................ 23 Power Modes .................................................................................................. 24 Annexes .......................................................................................................... 24 Annexes A and B ............................................................................................ 24 Annex L (Reach Extended ADSL) .................................................................. 24 Annex M (Enhanced Upstream) ...................................................................... 25 Annexes ADSL2+ (G.992.5) ........................................................................... 26 VDSL2 (G.993.2) ............................................................................................ 27 ADSL2+ Fallback ............................................................................................ 28 Power Management ........................................................................................ 28 G.INP .............................................................................................................. 36 G.Vector.......................................................................................................... 37 G.Fast ............................................................................................................. 37 Bonding (ATM/PTM) ................................................................................................ 38 ADSL2+ .......................................................................................................... 39 VDSL2 ............................................................................................................ 39

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

3

Chapter 3: Designing a Networking Service Model ............ 41 Introduction.............................................................................................................. 42 Bandwidth Requirements ....................................................................................... 42 Video Stream Bitrates ..................................................................................... 42 Video Middleware Bitrates .............................................................................. 44 Video Encryption Bitrates................................................................................ 44 Internet Bitrates .............................................................................................. 44 Supporting Tiered Service Offerings ............................................................... 45 Customer Premises Access Equipment ................................................................ 47 Residential Gateway ....................................................................................... 47 Application Networking Service Model ............................................................ 47 Different Residential Gateway (RG) Service Models ...................................... 49 RG DSL Chipset Vendor and Firmware Revision ........................................... 50 Access Node Service Radius ................................................................................. 52 Spectral Interference....................................................................................... 52 Estimated Rate vs. Reach .............................................................................. 53 Measuring Rate vs. Reach.............................................................................. 53 Zoning Model .................................................................................................. 55 Extending Reach with Bonding ....................................................................... 56

Chapter 4: Infrastructure ....................................................... 57 Best Practices at the CO and RT ............................................................................ 59 POTS/xDSL Splitter Cards and Shelves ......................................................... 59 Crosstalk ......................................................................................................... 60 Ring Trip Errors .............................................................................................. 60 Cabling: DSLAM to Splitter Shelf .................................................................... 61 Cabling: Splitter Shelf to Main Distribution Frame (MDF) ............................... 62 Cabling: MDF to Protection Panel ................................................................... 63 Connectors ..................................................................................................... 64 RJ-21 Connector for CAT5 and CAT5e Cabling ............................................. 64 Cabling: DSLAM to Pro Panel......................................................................... 65 Cabling: Protection Panel to the Cross Connect ............................................. 65 CO Surge Protectors (5-Pin Protectors).......................................................... 65 Best Practices for Outside Plant ............................................................................ 66 Noise and Crosstalk ........................................................................................ 66 Wire Gauge ..................................................................................................... 67 Splicing ........................................................................................................... 67 Drop Cables .................................................................................................... 68 Grounding and Ground Bonding ..................................................................... 68 Load Coils ....................................................................................................... 68 Bridged Taps .................................................................................................. 68 Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

4

Longitudinal Balance....................................................................................... 69 Cross Connect Blocks..................................................................................... 70 Lightning Protection ........................................................................................ 70 Line Power ...................................................................................................... 73 Reverse Line Power........................................................................................ 73 Best Practices at Customer Premises ................................................................... 74 Network Interface Devices (NIDs) ................................................................... 77 NID Protectors ................................................................................................ 77 NID Grounding and Bonding ........................................................................... 77 NID Splitters.................................................................................................... 78 Inside Wiring ................................................................................................... 79 EMI in the Home ............................................................................................. 80 Power.............................................................................................................. 81

Chapter 5: Best Practices for Deploying xDSL Service ...... 83 Feature Support by Product ................................................................................... 84 DSL Profiles ............................................................................................................. 85 Mode Selection ........................................................................................................ 86 Bitrate Configuration ............................................................................................... 86 SNR Margin .............................................................................................................. 87 Target Margin ........................................................................................................... 88 Maximum Margin ..................................................................................................... 89 Minimum Margin ...................................................................................................... 89 Upshift Noise Margin (SRA) .................................................................................... 90 Downshift Noise Margin (SRA) ............................................................................... 91 Latency Paths .......................................................................................................... 91 Impulse Noise Protection........................................................................................ 92 INP settings per Calix DSLAM ........................................................................ 93 G.INP/Retransmission ............................................................................................. 94 G.Vector.................................................................................................................... 94 Bonding .................................................................................................................... 95 Port Provisioning Recommended Settings ........................................................... 96 Port Provisioning Recommended Starting Point ............................................ 97

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

5

Chapter 6: Subscriber Activation with xDSL ..................... 101 Prequalifying a Subscriber for Service ................................................................ 102 Engineering Records Lookup ........................................................................ 102 Automated Line Testing ................................................................................ 102 MLT ......................................................................................................................... 103 OSP Checklist ............................................................................................... 104 Bonding Service ............................................................................................ 106 Subscriber Service Activation .............................................................................. 107 Customer Premise Power Inspection ............................................................ 107 Master Splitter Installation............................................................................. 107 Splitter Grounding and Bonding .................................................................... 107 Cabling Residential Gateway to the Splitter .................................................. 107 Residential Gateway Turn-up and Validation ................................................ 108 STB Installation ............................................................................................. 108 Application Validation: IP Video .................................................................... 108 Application Validation: HSI............................................................................ 108 Final Validation ............................................................................................. 108 Subscriber Service Deactivation .......................................................................... 109

Chapter 7: Monitoring xDSL Service .................................. 110 Monitoring Retrains ............................................................................................... 112 Monitoring Bitrates ................................................................................................ 112 Monitoring Operating Margin ............................................................................... 113 Monitoring Performance Counters ...................................................................... 114 Severely Errored Seconds ............................................................................ 114 Code Violations ............................................................................................. 114

Appendix A: Recommended Profile Settings .................... 117 Appendix B: Recommended Third-Party Components and Sources ................................................................................. 118 Recommended CO Cables .................................................................................... 119 Recommended MDF Terminal Blocks ................................................................. 120 Recommended CO External Splitters and Modules ........................................... 120 Recommended CO Splitter Shelves and Modules ........................................ 120 Recommended E-Series DSL and POTS Signaling Cables................................ 121 Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

6

Recommended E-Series Protection-Panel Retrofit Kits ..................................... 123 Recommended Customer Premise Master Splitters and Customer Premise NID Protectors ....................................................................................................... 125 Recommended 5 Pin Protectors .......................................................................... 126 Recommended Diagnostics Tools ....................................................................... 127 TraceSpan Communications......................................................................... 127 EXFO ............................................................................................................ 128 JDSU ............................................................................................................ 128

Appendix C: Rate Reach Data ............................................. 129 Appendix D: Pro-Panel Testing Results CAT3 vs. CAT5 .. 130 Rate Reach Testing ............................................................................................... 131 24-Hour Soak at 0ft and 4000ft ............................................................................. 133 Conclusions ........................................................................................................... 134

Appendix E: References ...................................................... 135 Online Resources .................................................................................................. 135 Standards ............................................................................................................... 135 Books ..................................................................................................................... 137 Research ................................................................................................................ 137

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

7

List of Figures Figure 1 - DSL Internal Protocol Stack in Relation to OSI and TCP/IP ..............................17 Figure 2 - ADSL,ADSL2+ and VDSL2Frequency Utilization ...............................................18 Figure 3 - VDSL2 Frequency Utilization .................................................................................18 Figure 4 - Comparative DSL Frequency Band Plans .............................................................19 Figure 5 - Sample Annex L(READSL2) Performance ............................................................25 Figure 6 - ADSL2/ADSL2+ Performance ................................................................................26 Figure 7 - Spectral Shaping in Mixed Line Length Applications ..........................................30 Figure 8 - Sample Bonding Performance .................................................................................40 Figure 9 - Zoned Serving Radius ..............................................................................................56 Figure 10 - Infrastructure Reference Diagram [DSLF 2005.195]..........................................58 Figure 11 - Left: CAT5 RJ-21 Connector with Proper Workmanship .................................64 Figure 12 - Type 525 RJ-21 Connector (110° Right/Standard exit) versus Standard RJ-21 Connector (90° Right/Standard exit) ....................................................64 Figure 13 - Typical Customer Premises Wiring for a Single STB Network .........................75 Figure 14 - Typical Customer Premises Wiring for a Dual STB Network ...........................76 Figure 15 - SNR Margin ............................................................................................................87 Figure 16 - 100-01755 - 100 pair pro-panel, 19 & 23 inch - 2RU, CAT5 .............................123 Figure 17- 100-01587 - E5-100 Octopus Cbl assembly, 50 Pair pro-panel, 19 & 23 inch - 2RU, CAT5 ...............................................................................................124 Figure 18 - CAT3 vs CAT5 Rate Reach .................................................................................132 Figure 19 - CAT3 vs CAT5 24-Hour Soak Connection Rates ..............................................133

List of Tables Table 1 - Tone and Frequency Distribution ............................................................................20 Table 2 - Layer 1& Layer 2ProtocolsUsed in xDSL Services ................................................21 Table 3 - Annex A vs. Annex M ................................................................................................25 Table 4 - ADSL2+ VDSL2 Intersection....................................................................................27 Table 5 - VDSL2 Profiles ...........................................................................................................28 Table 6 - Spectral Shaping: CO Lines ......................................................................................31 Table 7 - Spectral shaping: RT Lines .......................................................................................31 Table 8 - UPBO Settings, Annex A ...........................................................................................33 Table 9 - MPEG-2 Rates with Overhead Added .....................................................................43 Table 10 - Bitrate Calculation ...................................................................................................44 Table 11 - Protocol Overhead Estimates ..................................................................................46 Table 12 - DSLAM to Splitter Shelf Test Results: Average Crosstalk Measurements; 24 AWG; 8 Kft ....................................................................................................................61 Table 13 - DSLAM to Splitter Shelf Test Results: Worst-Case Crosstalk Measurements; 24 AWG; 8 Kft .........................................................................................62 Table 14 - Splitter shelf to MDF Test Results: Crosstalk Isolation @1.1 MHz; 24 AWG; 7 KftMDF Blocks ................................................................................................................62 Table 15 - Styles of MDF blocks ...............................................................................................63 Table 16 - 5 Pin Protector Technology .....................................................................................71 Table 17 - 5 Pin Protector Descriptions ...................................................................................72 Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

8

Table 18 - Feature Support by Product ...................................................................................84 Table 19 - DMT Protection Configuration ..............................................................................92 Table 20 - INT Settings Per DSLAM........................................................................................93 Table 21 – IPTV and HSI Downstream Recommendations ...................................................98 Table 22 – IPTV and HSI Upstream Recommendations ......................................................100 Table 23 - OSP Checklist .........................................................................................................104 Table 24 - Loop Qualification .................................................................................................105 Table 25 - C7 Recommended CO Cables ...............................................................................119 Table 26 - B-Series Recommended CO Cables......................................................................119 Table 27 - Recommended Splitter Shelves .............................................................................120 Table 28 - Recommended Splitter Cards ...............................................................................121

List of Recommendations Recommendation 1 -Managing Power Cutback......................................................................29 Recommendation 2 - Video Stream Bitrates............................................................................43 Recommendation 3 - Application Specific DSL Profiles ........................................................45 Recommendation 4 - IPTV and HSI Service Recommendation ............................................47 Recommendation 5 - Use of DHCP for IPTV Services ...........................................................48 Recommendation 6 - Use TR-69 for CPE Management .........................................................48 Recommendation 7 - QOS on the CPE ....................................................................................49 Recommendation 8 - Interoperability Testing.........................................................................51 Recommendation 9- xDSL Performance Testing ....................................................................54 Recommendation 10 -Zoning Model Recommendation .........................................................55 Recommendation 11 - Use of COMBO Line Cards ................................................................60 Recommendation 12 - Use of CAT5 Connectors and CAT5e Cable .....................................61 Recommendation 13 - Cabling between Splitter Shelf and MDF ..........................................62 Recommendation 14 - Crosstalk Mitigation ............................................................................63 Recommendation 15 - Hook Up Wire ......................................................................................63 Recommendation 16 - Best Practice recommendation for Pro Panels ..................................65 Recommendation 17 - 5 Pin Protectors ....................................................................................65 Recommendation 18 - Noise Reduction in the OSP ................................................................67 Recommendation 19 - OSP Cable Gauge.................................................................................67 Recommendation 20- Drop Cables ...........................................................................................68 Recommendation 21 - Load Coils .............................................................................................68 Recommendation 22 - Bridged Taps ........................................................................................69 Recommendation 23 - Cross Connect Blocks ..........................................................................70 Recommendation 24 – Line Power ...........................................................................................73 Recommendation 25 – Line Power placement.........................................................................73 Recommendation 26 - Customer Premises Prequalification ..................................................76 Recommendation 27- NID Protectors ......................................................................................77 Recommendation 28 - NID Grounding ....................................................................................77 Recommendation 29 - NID Splitters .........................................................................................78 Recommendation 30 - In Home Wiring ...................................................................................79 Recommendation 31 - Ethernet over COAX Cable ................................................................79 Recommendation 32 - EMI in the Home..................................................................................80 Recommendation 33 - AC Power in the Home ........................................................................81 Recommendation 34 – DSL mode and packet mode...............................................................86 Recommendation 35 - Bitrate Configuration ..........................................................................86 Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

9

Recommendation 36 - Downstream and Upstream Target Margin for IPTV .....................88 Recommendation 37 - Downstream and Upstream Target Margin for HSI ........................88 Recommendation 38 - Maximum SNR Margin for IPTV ......................................................89 Recommendation 39 - Maximum SNR Margin for HSI .........................................................89 Recommendation 40 - Downstream and Upstream Minimum SNR Margin for IPTV .......90 Recommendation 41 - Upstream and UpstreamMinimum SNR Margin for HSI ..............90 Recommendation 42 - Upshift Noise Margin for IPTV ..........................................................90 Recommendation 43 - Upshift Noise Margin for HSI .............................................................90 Recommendation 44 - Downshift Noise Margin for IPTV .....................................................91 Recommendation 45 - Downshift Noise Margin for HSI ........................................................91 Recommendation 46 - IPTV Latency Configuration ..............................................................91 Recommendation 47 - HSI Latency Configuration.................................................................91 Recommendation 48 - Downstream Impulse Noise Protection for IPTV .............................93 Recommendation 49 - Upstream Impulse Noise Protection for IPTV ..................................93 Recommendation 50 -Impulse Noise Protection for HSI Applications .................................93 Recommendation 51 – G.INP or PhyR for IPTV ....................................................................94 Recommendation 52 – G.INP or PhyR for HSI.......................................................................94 Recommendation 60 – G.Vector group. ...................................................................................94 Recommendation 61 – G.Vector deployment. .........................................................................95 Recommendation 62 – G.Vector and Legacy CPEs ................................................................95 Recommendation 63 – G.Vector and G.INP ............................................................................95 Recommendation 53- MLT .....................................................................................................103 Recommendation 54 - Retrain Per Day Threshold: 1 or less ...............................................112 Recommendation 55 -Insufficient Bitrate Threshold: 32 Kbps ...........................................113 Recommendation 56 - Margin Change Threshold: 1.5 dB ...................................................113 Recommendation 57 - SES-L Threshold: 0 Seconds .............................................................114 Recommendation 58-Coding Violation Threshold: 2 CVs per IP Video Stream, Downstream Only .............................................................................................................115 Recommendation 59 - Best practice for a Network Outage .................................................134

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

10

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

11

About This Guide Statement of Purpose This document offers service providers a foundation necessary to plan, deploy and support High Speed Internet and IP video (IPTV) services over copper using standards-based DSL access technologies (xDSL). This document describes the physical layer factors that may directly influence the quality of data and video services delivery, and recommends the best practices necessary to achieve optimal results. Calix recognizes that there are costs associated with achieving optimal results, and that optimizing some factors will yield better improvements and have a bigger pay off than others. The best practices recommendations in this document are those that Calix believes make economic sense, and have yielded proven results in real deployments. The primary audience for this document includes personnel responsible for design and service configuration at the Network Operations Center (NOC), carrier Central Office (CO), or Remote Terminal. Calix assumes that users have experience with computer systems and software, and have knowledge of telecommunications and engineering standards. Familiarity with the Calix xDSL services platforms (B6, C7, E-Series system hardware, as well as the Calix CMS network management interfaces is recommended). Some features and functionality described herein may not be available on all Calix products; however product-specific differences are captured where possible. For detailed information on individual products, refer to the product-specific documentation sets.

New in Rev 16: This guide has been updated with sections on G.INP and G.Vector. The bonding section has been updated to include 17a bonding, and the Calix platforms E7-48(c) and E5-48(c) have been added to the tables. The reference section has also been revamped and the rate reach appendix has been updated to point to the document with rate reach data.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

12

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

Chapter 1 Introduction As service providers enhance their High Speed Internet (HSI) services and pursue the exciting opportunities of Internet Protocol (IP) based information and entertainment services, and specifically IP video services, they are challenged with many new technologies. Service providers that are well-versed in Digital Subscriber Line (DSL)technologies, ADSL and VDSL (xDSL), have successfully used them to provide Internet access services for years. Performance demands of new applications and associated bandwidth requirements, particularly IP-based video services (IPTV), far exceed the capabilities of traditional DSL deployments. When using xDSL for loss-sensitive applications like IPTV, a clear understanding of its features and limitations is crucial. This document presents the technology, requirements and best practices to successfully deploy HSI and IPTV over xDSL data lines. It is intended for a broad audience of people from planners to technologists, upper management to central office and outside plant technicians. We begin with a discussion of DSL and its underlying technology, move onto key analytical tasks necessary to successfully plan for commercial deployment of IPTV, and then follow up with recommended DSL profile settings. Consideration is given to the challenges providers face in determining what service profiles they will offer and to what distance xDSL can support those profiles. Calix has a long and successful track record of deploying IP video service over xDSL, with hundreds of our Service Provider customers serving hundreds of thousands of end subscribers with IPTV over Calix access equipment. The best practices information and conclusions presented in this document are a culmination of years of gaining valuable insight, knowledge, and experience from live deployments as well as extensive and continual testing and research, which takes place in our development labs. IPTV is still maturing, and the ways we interact across a HSI connection continue to evolve; this document aims to evolve accordingly, to arm you with the best information to solve your service delivery challenges.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

14

About Best Practices Deploying HSI and IPTV services together over xDSL presents a complex and challenging opportunity, with the new features of G.INP and Vectoring further increasing the complexity and allowing for even faster connections. Many variables within the copper access network directly influence the quality of the data/video signal and each must be addressed, managed, and optimized in order to ensure successful IPTV deployment (and delighted video subscribers). In general, the variable elements fall into four categories:    

The planning and layout of the network design and service offerings. The quality and characteristics of the outside plant (such as loop length, wire gauge, splices, impairments, protectors, etc). The quality and characteristics of the inside wiring (such as wire gauge and type, connectors, bridged taps, splices, and impairments). The data rate, strength and robustness of the xDSL signal (such as provisioning bandwidth, xDSL modes and VDSL2 profiles, signal-to-noise ratio, impulse noise protection, interleave depth, etc).

Successful IPTV roll out requires that service providers address all the variable factors within these categories. Optimizing one area and ignoring others typically does not satisfy the requirements for "perfect quality" video. And for IPTV to be successful, the picture quality must be perceived as equal to or better than the video quality offered by competitive cable and satellite operators. The recommendations in this document represent the ‘best practices’ available to address the key factors listed above. You can employ any of these individual best practice recommendations to achieve immediate incremental performance improvements. However, leveraging many or all of the best practice recommendations in combination will absolutely provide substantial performance gains in quality and/or range.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

15

IPTV Network Model This document uses the following simplified IPTV network model: DSLAM

Head End

Splitter Shelf

Cross Connect Box

NID

xDSL Router/ RGW

STB

TV

Cross Connect Box

NID

xDSL Router/ RGW

STB

TV

w/xDSL Cards

MDF/ Pro Panel

DSLAM

1

MDF/ Pro Panel

w/Combo Cards

2

3

4

This model consists of four distinct sections: 1. The Video Headend and associated equipment (video headend encoders, routers, application servers) 2. The Central Office (CO) and/or Remote Terminal (RT) equipment (Equipment, splitter card/shelf, cabling, main distribution frame, protection panel) 3. The Outside Plant (OSP) (cross connect box, cable binders, cabling) 4. The Customer Premises Equipment (CPE) (network interface device, wiring, xDSL router, set top box, television) Each section, and the individual network elements contained within, is addressed in separate chapters in this document.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

16

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

Chapter 2 Underlying Technology of xDSL This chapter is an introduction to xDSL technology. See the “Best Practices for Deploying xDSL Service” chapter for recommended settings. Between a twisted pair of wire and the data or IPTV for transmission across the wire, a number of protocol layers exist. DSL encompasses the physical layer of the OSI network model and within that, has a number of its own layers. The following diagram illustrates the relationship between the OSI network model, the TCP/IP network model (which is the fabric of the Internet), and the DSL layers. The following sections delve into the DSL layers in depth, focusing on the information critical to provisioning and troubleshooting services.

Figure 1 - DSL Internal Protocol Stack in Relation to OSI and TCP/IP

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

18

DMT Each variant of DSL uses Discrete Multi Tone (DMT) technology to partition transmission into small frequency bands called subcarriers across the spectrum between 0 KHz and 30 MHz. ADSL and ADSL2 use the frequencies between 0 and 1.1 MHz. The ADSL2+ standard doubled the available ADSL2 spectrum to 2.2MHz, while VDSL2 added frequencies up to 30MHz. Each subcarrier is 4.3125 KHz wide, providing a total of 256 subcarriers for ADSL2, 512 subcarriers for ADSL2+, and 4,096 subcarriers for VDSL2.DSL Annexes discussed in this document do not use the first 7 subcarriers to avoid interfering with POTS transmission signals, thus the actual frequencies used begin at 28KHz. There are options for “All Digital” DSL—a mechanism that employs the use of the first 7 subcarriers for additional performance on lines that do not carry POTS.

POTS/ ISDN

UPSTREAM

DOWNSTREAM

DOWNSTREAM

DOWNSTREAM & UPSTREAM DS1

US1

DS2

US2

DS3

US3

VDSL2 ADSL2+ ADSL1 ADSL2 MHz

0.14

1.1

2.2

Figure 2 - ADSL,ADSL2+ and VDSL2Frequency Utilization

Figure 3 - VDSL2 Frequency Utilization Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

30

19

QAM During pre-xDSL standard days, few equipment providers developed their own band plan for QAM-based (Quadrature Amplitude Modulation) DSL: upstream (US) 324960KHz, downstream (DS) 960KHz to 9.3MHz. This technology was also referred as “Classic” VDSL. The frequency plan was chosen in order to prevent the upstream band from being the limiting factor for the xDSL reach. Allocating the lowest frequencies to US makes the DS the limiting factor (limitation due to cable attenuation of signal); by providing slightly lower DS rates, the reach could be extended. The US band start at 324KHz was set to avoid interference with any sort of ISDN or HDSL2 services that might co-inhabit the binder group. When ADSL and the 992 standards arrived, ADSL was defined as (approximately) 25135KHz US band, and 135KHz-1.1MHz DS band. This DS band overlaps the DSL-Classic US band, and stomps on that signal, not allowing modem traffic in binder groups containing both “Classic” VDSL and ADSL. In general, this tended to not be a big practical problem, as operating companies engineered their deployments through placement of cards/services within shelves/cabinets, such that shared binder groups was avoided as much as possible. The use of standards based xDSL within binder groups shared with “Classic” VDSL is incompatible and problematic at best. Once 998 VDSL came out, this problem disappeared; although 998 VDSL and “Classic” VDSL are also incompatible, for exactly the same reason explained already.

Figure 4 - Comparative DSL Frequency Band Plans The 998/997(*) standard band plan was specifically designed for ADSL, ADSL2+ and VDSL2 to be able to co-exist within the same binder group and cable. Only the old, prestandard “Classic” VDSL is problematic. Mixed deployments of this technology with standards based xDSL are very messy and not recommended. (*) Note: VDSL or ITU-T Recommendation G.993.1 defines two band plans A and B, also previously known as plan 998 and plan 997 respectively.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

20

Subcarriers and Tones Signal loss is not constant across the DSL spectrum; high frequencies experience more attenuation than low frequencies at the same distance. Each subcarrier functions as an independent communication channel, determining its capabilities are during the handshake process, and adapting as necessary over time to maintain a Bit Error Rate (BER) of 10-7. A subcarrier transmits between 1 and 15 bits or “symbols” per clock cycle. Subcarriers may be referred to as “bins” and symbols as “bits”. A subcarrier using 15 symbols may be described as “15 bits per bin”. The number of symbols per subcarrier is determined by how well the DSLAM and residential gateway can hear each other at that particular frequency; this measurement is called Signal to Noise Ratio (SNR). SNR is analogous to a conversation in a noisy room or with someone far away. The louder the room or the further they are, the less you can communicate and be understood. ADSL uses a fixed data rate of 4000 baud per subcarrier. With quadrature amplitude modulation (QAM) 256, there are 256 symbols per constellation. Eight bits is required to represent 256 symbols, therefore a subcarrier can communicate at 8x4000baud or 32,000 bits per second (32 Kbps). In the context of DSL, a “tone” represents a subcarrier at a specific frequency. Tone 0 corresponds to the subcarrier in the range of 0 and 4.3125 KHz. Tone 255 represents the subcarrier between 1100 KHz and 1104 KHz.

Upstream vs. Downstream The DSLAM uses Frequency Division Multiplexing (FDM) to separate the downstream transmission coming from the DSLAM from the upstream traffic of the residential gateway. Upstream traffic uses the first 32 subcarriers of Annex A, or 64 subcarriers in the case of Annex M. The remaining subcarriers are used for downstream traffic. The first 6 subcarriers are not used by Annex A (tones 0 through 5, 0-25.875 KHz) as this spectrum is used for narrowband voice services (POTS). The table below shows the allocation of frequency and the corresponding achievable data rates, assuming 15 bits per bin. Direction Annex A

Protocol

Tones

Frequency

Theoretical Bitrate

Upstream

ADSL 2/2+

6-31

25.875 KHz – 138 KHz

1.56 Mbps

Downstream

ADSL2

32-255

138 KHz – 1.104 MHz

13.44 Mbps

Downstream

ADSL2+

32-255

138 KHz – 2.208 MHz

28.8 Mbps

Upstream

ADSL 2/2+

6-63

25.875 KHz – 276 KHz

3.12 Mbps

Downstream

ADSL2

64-255

276 KHz – 1.104 MHz

11.88 Mbps

Downstream

ADSL2+

64-255

276 KHz – 2.208 MHz

27.24 Mbps

Annex M

Table 1 - Tone and Frequency Distribution

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

21

VDSL2 uses the same signaling spectrum characteristics as ADSL2 and ADSL2+ for the first 255 tones.VDSL2 expands the available tones to 1,971, 2,770, and 4,095 for 8.5 MHz, 12 MHz, and 17 MHz solutions. Each of the VDSL2 signaling profiles up to 17 MHz use the same 4.312 KHz frequency spacing as ADSL2 and ADSL2+. The 6-band 30 MHz band uses 2,098 tones with 8.625 KHz frequency spacing (not currently supported). VDSL2 modulates the upstream and downstream signaling frequencies.

Packet Transfer Mode (PTM) As the VDSL standards were being developed, both the IEEE 802.3ah task force and the ITU-T groups worked on defining and adopting a generic Packet Transfer Mode (PTM). The VDSL2 standard included PTM as an improvement over ATM cell based communication. Previously, ADSL encapsulated Ethernet via UTOPIA into ATM cells (AAL5) to be transported into the DSL physical interface layer. Within VDSL2, adoption of PTM allows both the DSLAM and the CPE to improve throughput by eliminating the Segmentation and Reassembly (SAR) block. After its adoption, PTM mode also became available for ADSL2+ applications. Several CPE manufacturers include both ATM and PTM modes in their ADSL2+ products.

xDSL Protocols ADSL encompasses a family of protocols initially defined in T1.413 and then adopted by the ITU and defined in G.992.1–G.992.5. ADSL uses a handshake protocol shared by other DSL technologies referred to as “G.handshake” and defined in G.994.1. Operations, Administration and Maintenance (OAM) specifications including all statistics available in ADSL are defined in G.997.1. IETF RFC-2662 captures these parameters in the standardized ADSL LINE MIB. Protocol

Description

Specification

ADSL

ADSL “G.dmt”

G.992.1

ADSL Lite

Splitterless ADSL or “G.lite”

G.992.2

ADSL2

ADSL2 “G.bis”

G.992.3

ADSL2 Lite

Splitterless ADSL2

G.992.4

ADSL2+

ADSL2Plus “G.bis.plus”

G.992.5

VDSL(1)

VDSL

G.993.1

VDSL2

VDSL2 (DMT)

G.993.2

Handshake

Handshake protocol “G.hs” used by the xDSL protocol family.

G.994.1

Test

Test procedures, configurations and environments for xDSL.

G.996.1

Test

Single-ended line testing for digital subscriber lines (SELT)

G.996.2

Management

Physical layer management for DSL

G.997.1

Bonding

ATM Bonding for DSL

G.998.1

Bonding

PTM Bonding for DSL

G.998.2

G.INP

Improved impulse noise protection for DSL transceivers (G.inp)

G.998.4

G.Vector

Self-FEXT cancellation (vectoring) for use with VDSL2 transceivers (DSM Level 3)

G.993.5

Table 2 - Layer 1& Layer 2ProtocolsUsed in xDSL Services Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

22

ADSL2 and ADSL2+ recommendations include multiple annexes describing modes of operation for different applications. Annex A is the most common means of deploying ADSL2 and ADSL2+ in North America. Annex L provides extended reach capabilities to better serve subscribers at the farthest distances possible for ADSL, and Annex M enables higher upstream bitrates, primarily for business users.

ADSL (G.992.1) ADSL was initially introduced as ANSI T1.413 and later adopted by the ITU as ITU-T G.992.1, ratified in 1999. ADSL is the foundation for the newer ADSL2 and ADSL2+ technologies. It utilizes DMT between 0 and 1.1 MHz.

ADSL2 (G.992.3) ADSL2 improves and builds upon the ADSL specification. While not a complete list, the following identifies the most important developments in ADSL2 for triple play applications: • • • •

Performance Online Reconfiguration Diagnostics Power Modes

Performance ADSL2 improves the coding techniques used for digital transmission, provides better error correction and optimizes the use of spectrum by allowing data transmission on subcarriers that used to be reserved (pilot tone), and allowing 1-bit constellations. Online Reconfiguration Online reconfiguration allows the DSLAM and residential gateway to change characteristics of the data plane without interrupting data flow. Improvements have been made to the bitswapping algorithms—a technique for changing the allocation of bits when one subcarrier improves and another degrades. SRA Seamless Rate Adaption (SRA) targets optimal data rate performance without operator intervention, providing a safeguard from changes in the cross-talk levels in the binder caused by various noise sources. SRA is used to maintain good Signal to Noise Ratio (SNR) levels, while automatically reducing the line rate when these noise conditions persist for the selected downshift time.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

23

Use the following key parameters to control SRA: •

Rate Adaptation Downshift Noise Margin If the current upstream noise margin is below the Downshift Noise Margin for more than the time specified by theDownshift Time, the ATU-C triggers an SRA downshift.

•

Rate Adaptation Downshift Time This parameter specifies the minimum time interval during which the upstream noise margin stays below the Downshift Noise Margin before the ATU-C triggers an SRA downshift.

•

Rate Adaptation Upshift Noise Margin If the current upstream noise margin is above the Upshift Noise Margin for more than the time specified by the Upshift Time, the ATU-C triggers an SRA upshift.

•

Rate Adaptation Upshift Time This parameter specifies the minimum time interval during which the upstream noise margin stays above the Upshift Noise Margin before the ATU-C triggers an SRA Upshift.

SRA must be enabled for the above settings to operate. Note: Do not to allow the port to adapt to a rate lower than the minimum rate for the service applied to the port. If this occurs, the modem retrains.

Loop Diagnostics Loop diagnostics standardize a mechanism for the DSLAM and residential gateway to evaluate line conditions and report on: • • •

Attenuation Quiet Line Noise per subcarrier Signal to Noise Ratio per subcarrier

SELT Single Ended Line Testing (SELT) is an automated method of testing a DSL loop from one end of the line (typically from the central office). SELT can determine the wire gauge, length of the loop, and presence of noise and attenuation of the line—in essence the quality of the copper and its ability to deliver service. SELT also provides the Shannon capacity (theoretical maximum information transfer rate of the channel). SELT is an excellent tool to prequalify lines. Note: When running a SELT test, do not plug a modem into the LOOP under test.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

24

DELT Dual End Loop Test (DELT)differs from SELT, which tests an un-terminated copper line; DELT uses both the line card and a remotely deployed modem to perform bi-directional loop tests. DELT is able to collect much more data for loop analysis than SELT alone. DELT is an excellent troubleshooting tool. Note: DELT requires both a compliant DSLAM at the central office or remote terminal and a modem to operate properly.

Power Modes ADSL2 enables power savings by using OLR capabilities to configure a line into low power mode when no activity is detected. “L2 Power Mode” maintains a minimal communication path, minimizing power output until data traffic is detected. Once traffic is detected the line returns to “L0 Power Mode”, re-establishing the original data rate in less than 1ms. Note: IP video streams constantly to set top boxes, even when TVs are off. The power mode features may not apply to IP video applications, but certainly reduce power consumption in data only applications.

Annexes ADSL2 introduced three important annexes which put DSL into special modes of operation including: ADSL2 line shared with North American POTS (Annex A), reach-extended ADSL2 (Annex L), and enhanced upstream bandwidth ADSL2 (Annex M).

Annexes A and B Annex A is the predominant form of ADSL used in North America. Annex A reserves the lower frequencies for use by analog telephony. Annex B allocates additional headroom to allow the line to be shared with BRI (2B+D) ISDN. You may find both Annex A and Annex B deployed in Europe.

Annex L (Reach Extended ADSL) Annex L is selected when auto-moding is enabled and the training algorithms determine that the residential gateway is sufficiently far from the DSLAM so that a special power spectral density (PSD) mask can improve performance. Auto-moding is a method that establishes DSLAM-to-CPE communication based on using the highest attainable line rate. As the DSLAM establishes the line the auto-mode capability uses different protocols until the line is trained.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

25

Figure 5 - Sample Annex L(READSL2) Performance

Annex M (Enhanced Upstream) Annex M enhances upstream bandwidth by re-allocating lower frequency DMT tones from downstream to upstream use. Like Annex A, Annex M service operates in the band above analog voice. By reallocating bandwidth from downstream to upstream, use of Annex M can more than double the potential upstream bandwidth available to a subscriber. Frequency Band

0–25 KHz

25 – 138 KHz

138- 254 KHz

254KHz -1.1 MHz

1.1 - 2.2 MHz 256-511

DMT Tone Numbers

0-5

6-31

32-64

65- 255

G992.3 Annex A

POTS

Upstream

Downstream

Downstream

G992.3 Annex M

POTS

Upstream

Upstream

Downstream

G992.5 Annex A

POTS

Upstream

Downstream

Downstream

Downstream

G992.5 Annex M

POTS

Upstream

Upstream

Downstream

Downstream

Table 3 - Annex A vs. Annex M Annex M supports upstream bandwidth as high as 3 Mbps on short loops (under 3 Kft). Because the low frequency tones are attenuated less on longer loops than on higher frequencies, Annex M continues to provide good upstream performance as the loops lengthen. Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

26

Although the bandwidth available for downstream traffic is somewhat reduced when using Annex M, there is still sufficient downstream capacity to support robust service toward the subscriber. Annex M ADSL services are quite effective with ADSL2+ (G992.5) on shorter loops, as ADSL2+ supports 256 additional tones in the band from 1.1 to 2.2 MHz providing significant extra downstream capacity on loops less than 12 Kft in length. However, even when ADSL2/2+ Annex M is used on longer loops the subscriber receives a downstream bitrate at least as high as available with ADSL (G992.1) supporting the service. Note: To avoid spectral compatibility issues, do not mix Annex M with Annex A.

Annexes ADSL2+ (G.992.5) ADSL2+ is ADSL2 with twice the downstream spectrum allocation, doubling the maximum downstream bitrate. Where ADSL and ADSL2 use the spectrum between 0 and 1.1 MHz, ADSL2+ uses the spectrum between 0 and 2.2 MHz. ADSL2+ has a maximum theoretical downstream bitrate of 32 Mbps.

Figure 6 - ADSL2/ADSL2+ Performance

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

27

VDSL2 (G.993.2) VDSL2 is a significant milestone in the development of ADSL2+ standards. Adopted by the ITU in 2005 as G.993.2, VDSL2 adds new upstream and downstream signaling profiles above those of ADSL2+, and is capable of downstream bitrate performance up to 100Mbps with upstream rates up to 56Mbps. The selected VDSL2 profile adds the subsequent number of tones associated with the signaling hertz created with signaling profiles described as 8a, 8d, 12a, and 17a. These profiles refer to the signaling bandwidth (MHz) and are commonly referred to as 3-band, 4-band, and 5-band plans. Each band plan offers significant downstream and upstream capabilities over ADSL2 capabilities. Generally 3-band, 4-band, and 5-band plans are used for subscriber services and are asymmetric. 30 MHz 6-band VDSL2 is a symmetric based solution.

Distance in feet

0

1000

2000

3000

4000

5000

6000

ADSL2+

29,407

28,844

27,678

27,789

25,799

22,952

19,846

VDSL2 (17a)

100,015

100,015

67,370

49,656

33,999

24,028

16,327

Table 4 - ADSL2+ VDSL2 Intersection As shown in the table above there is an intersection point between the ADSL2+ and VDSL2 performance curves. On shorter loops VDSL2 outperforms ADSL2+, yet on longer loops the VDSL2 performance curve dips below the performance capabilities of ADSL2+. The DSLAM and DSL profile settings are configured by default to select the highest performance. With the default setting, the DSLAM would use VDSL2 signaling through 5000 feet and would then change to ADSL2+ signaling. US0 Values US0 has a selection of BW settings required for profiles 8a, 8b, 8c, 8d and 12a. US0 is not required for profiles 12b and 17a, but can also be used. All VDSL2 profiles can be configured with or without US0, as appropriate. The range of settings for US0 is from EU32 to EU128. EU indicates “End Upstream” and the number following it specifies the bin or tone count. For example, EU32 indicates End Upstream at bin 32. This is the same as ADSL, ADSL2 and ADSL2+ for Annex A Upstream. EU64 is similar to the bin count used for Annex M in ADS2 and ADSL2+. G.993.2 allows for an additional selection to increase the amount of BW allocated to US0 as required by the service. Note: The typical default setting for US0 is EU32. VDSL2 Profiles The table below is an excerpt from ITU G.993.2 listing the available VDSL2 profiles. Frequency plan

Parameter

Parameter value for profile 8a

8b

8c

8d

12a

12b

17a

30a

All

Maximum aggregate downstream transmit power (dBm)

+17.5

+20.5

+11.5

+14.5

+14.5

+14.5

+14.5

+14.5

All

Minimum aggregate downstream transmit

For further

For further

For further

For further

For further

For further

For further

For further

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

28 Frequency

Parameter

Parameter value for profile

power (dBm)

study

study

study

study

study

study

study

study

All

Maximum aggregate upstream transmit power (dBm)

+14.5

+14.5

+14.5

+14.5

+14.5

+14.5

+14.5

+14.5

All

Minimum aggregate upstream transmit power (dBm)

For further study

For further study

For further study

For further study

For further study

For further study

For further study

For further study

All

Sub-carrier spacing(s) (kHz)

4.3125

4.3125

4.3125

4.3125

4.3125

4.3125

4.3125

8.625

All

Support of upstream band zero (US0)

Required

Required

Required

Required

Required

Not Required

Not Required

Not Required

All

Minimum net aggregate data rate capability (Mbit/s)

50 Mbit/s

50 Mbit/s

50 Mbit/s

50 Mbit/s

68 Mbit/s

68 Mbit/s

100 Mbit/s

200 Mbit/s

Table 5 - VDSL2 Profiles

ADSL2+ Fallback ADSL2+ fallback is a feature available on newer DSL chipsets. If the preferred VDSL2 profile cannot be achieved the fallback feature enables the DSLAM to ‘fall back’ to ADSL2+ signaling and rate/reach performance. This feature allows for the highest attainable bitrate to always be selected for subscriber lines.

Power Management VDSL2 and ADSL2+ have a number of intelligent features related to power. These are discussed in the following sections. Power Cutback Also called power backoff, power cutback can be thought of as a friendly neighbor algorithm. The goal of power cutback is to transmit at a power level sufficient to achieve the best possible bitrate up to the provisioned maximum bitrate with an SNR Margin equal to or greater than the provisioned target SNR Margin. If that is achievable and sufficient excess margin (SNR Margin above target) exists, power may be reduced while still adequately maintaining signal. When all DSLs in a binder are able to provide service but transmit no louder than necessary, the overall crosstalk in the binder is minimized. Power cutback introduces the potential for performance improvements but introduces challenges as well. Two adjacent pairs transmitting at different power levels will interfere with each. Assuming two adjacent pairs are trying to support the same DSL Profile but one is on a short loop and another is a long loop, the DSL service on the long loop will operate at a higher power level than the short loop. In trying to be a good neighbor, the short loop will perform power cutback and then suffer from an elevated noise floor due to the adjacent long loop transmitting at a higher power.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

29

Service providers rarely engineer short lines and long lines in the same binder, distributions tend to go to a serving area interface (SAI) where an entire binder is terminated at the same distance. However, multiple distributions leave the central offices and remote terminals and these get multiplexed together at the main distribution frame (MDF) when they are patched to the DSLAM. The connections between the DSLAM and the MDF become the playground for long lines to bully short lines. The excess margin or “max additional margin” as it is defined in the ITU recommendations is a parameter provisioned at the DSLAM. This can be used as a coarse control on power cutback. As we will discuss, spectral shaping can be used as well. Recommendation 1 -Managing Power Cutback In order to leverage the benefits of power cutback without suffering the consequences of high powered and low powered DSL lines adjacent to one another, power cutback must be managed. Calix recommends the following considerations for managing power cutback: •

Loop Segregation: Where possible, plan to terminate short loops and long loops on different J connectors at a DSLAM or on different DSLAMs. This segregates the run of short loops from long loops with separate cabling between the MDF or Pro Panel and DSLAM.

•

Max-additional margin setting: When short loops must run in the presence of adjacent pairs using DSL at higher power (longer loops) and spectral shaping does not sufficiently insulate the interference introduced to a low power line (or is not available) , Calix recommends increasing maxadditional margin to manage power cutback. The net effect of increasing the max margin is to increase the power level to compensate for other high power line in the binder. This gives additional SNR margin to these lines.

The generic max margin recommendations for IPTV and HSI are shown below. •

IPTV: Calix recommends enabling power cutback by setting max- margin to 16 dB.

•

HSI Access: Calix recommends a 10 dB max- margin setting.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

30

Spectral Shaping ADSL2+ has spectral shaping features allowing for selective use of the allocated spectrum. This is important in applications where a known disturber must be avoided. Spectral Shaping provides the ability to manipulate the power spectral density (PSD) of transmission across the frequencies utilized by ADSL2+. This capability is exposed to the operator as a set of transmission filters that range from maximized performance with maximum spectral and power utilization to reduced performance with minimal utilization of the frequencies between 0 and 1.1 MHz. This feature is useful when ADSL transmitters from different distances are present in the same binder, for example when services from both the central office (CO) and a remote terminal (RT) are present in the same binder. There are 15 filter settings available for controlling spectral shaping. The example below demonstrates the effect of each filter setting for a specific application. Figure 7 describes an application where a binder group of 24 copper pairs is providing DSL service on each pair. The result tables show the effect of applying each filter setting in the example where 12 of the pairs are served from a CO that is 7000 ft from the residence and the other 12 pairs are served from an RT that is 3000 ft from the residence. Selecting filter setting 0, which disables filtering, yields an adverse effect on the training rates of the CO lines due to the presence of other lines trained at ADSL2+ speeds in the bundle. Also note that the RT lines for filter setting 0 achieve higher training rates. By selecting higher numbered filter settings, the service provider can effectively remove the cross talk interference caused by the RT lines in the bundle. Filter setting 15 has the maximum effect forcing the RT pairs to only operate in the higher spectrum and reserving the lower spectrum for the CO lines. CENTRAL OFFICE

REMOTE TERMINAL MHz

1.1 2.2

ADSL2+ Common Binder

ADSL2

MHz

1.1 2.2

Figure 7 - Spectral Shaping in Mixed Line Length Applications

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

31 Central Office Line Averages with 12 RT Lines This shows the average downstream bitrate, SNR margin, power output and attenuation for 12 Central Office Lines with 7000ft loops in the presence of 12 Remote Terminal lines. The RT lines are configured with the filter setting shown in the first column.

Filter

Bitrate (Kbps) DS

SNR margin US

Power x10

Attenuation x10

DS

US

DS

US

DS

US

Filter 0

7408

1387

8.5

7.4

185

120

365

206

Filter 1

9400

1390

6.4

7.5

181

120

359

202

Filter 2

9545

1394

7.1

7.5

177

120

357

200

Filter 3

9569

1390

7.5

7.4

180

121

354

198

Filter 4

10119

1390

6.5

7.4

175

120

353

197

Filter 5

9692

1386

8.4

7.3

180

121

352

197

Filter 6

10062

1388

7.9

7.3

178

120

352

196

Filter 7

9922

1388

8.5

7.3

177

120

350

196

Filter 8

10599

1387

7.9

7.3

177

121

351

196

Filter 9

11013

1384

7.4

7.4

177

121

350

196

Filter 10

11221

1389

7.4

7.3

178

121

350

196

Filter 11

11454

1390

7.6

7.3

178

120

353

197

Filter 12

11808

1389

7.5

7.3

180

121

352

197

Filter 13

10166

1156

5.9

6.6

177

121

295

180

Filter 14

12606

1386

6.9

7.4

176

120

351

196

Filter 15

12945

1389

6.9

7.5

177

120

354

197

Table 6 - Spectral Shaping: CO Lines Remote Terminal Line Averages with 12 RT Lines 12 Remote Terminal lines with 3kft loops and spectral shaping filter settings shown in the first column. These lines share a binder with 12 CO lines that have 7kft loops.

Filter

Bitrate (Kbps) DS

US

SNR margin DS

US

Power x10 DS

US

Attenuationx10 DS

US

Filter 0

22836

1377

8.6

7.2

167

122

170

85

Filter 1

21861

1379

8.5

7.2

165

122

226

83

Filter 2

22113

1378

8.0

7.3

161

122

237

82

Filter 3

22075

1380

8.0

7.2

159

122

239

82

Filter 4

22141

1379

7.7

7.3

157

122

244

81

Filter 5

22298

1378

7.7

7.2

156

122

251

81

Filter 6

21910

1383

8.1

7.2

157

122

257

81

Filter 7

21763

1377

8.1

7.2

153

122

260

81

Filter 8

21766

1376

8.1

7.3

146

122

267

81

Filter 9

21731

1381

7.9

7.3

146

122

273

81

Filter10

21500

1380

8.0

7.2

142

122

276

80

Filter 11

21400

1381

8.4

7.2

139

122

287

81

Filter 12

20787

1381

8.9

7.3

135

122

285

81

Filter 13

20100

1379

9.0

7.3

129

122

289

81

Filter 14

20231

1376

9.1

7.3

124

122

297

81

Filter 15

19483

1379

9.3

7.3

120

121

303

81

Table 7 - Spectral shaping: RT Lines Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

32

Power Back off (PBO) VDSL2 introduces US and DS PSD masks. Loop lengths may be different from line to line within a cable binder group, causing the Power Spectral Density (PSD) to differ between lines. This varying degree of PSD on every line can cause cross talk on adjacent lines within the binder group. Upstream PSD originating from the CPE induces US cross-talk or far-end cross talk (FEXT). Downstream PSD from the CO side of the VDSL2 link induces near-end cross talk (NEXT) on adjacent lines. Because the power is higher at the source or the transmitter, NEXT is almost always worse than FEXT. This can be at the transmitter at the CPE (for DS bands)or the transmitter at CO side of the link (for US bands). The power back off (PBO) feature helps to reduce cross-talk noise interference on adjacent lines within a cable binder group by limiting the PSD on a given range of frequencies. Upstream Power Back Off (UPBO)

The UPBO reduces the FEXT on adjacent lines. There are two distinct bands for UPBO - US1 and US2 (excluding US0). The PSD masks on the mandatory US bands are managed through configurable parameters. Note that depending on the negotiated VDSL2 profile and bandplan, not all the US bands are available. US1 and US2 assume the 17a profile, 998 North American Frequency (band plan). For more information on UPBO, please refer to ITU-T G.993.2. Guidelines for making UBPO settings Background: ITU-T G.993.2 defines UPBO in section 7.2.1.3, and provides the formulas for calculating the US PSD Mask. Unfortunately, G.993.2 does not include any background in defining parameters “a” and “b” in the PSD calculation formula. The ATIS 0600027 document provides guidance for the setting the “a” and “b” parameters and explains them. Essentially parameter “a” defines the nominal reference PSD and parameter “b” models the loop loss (known as reference length or RL). These parameters are defined for US1, US2, US3, and US4 bands (See ITU-T-G.993.2-2007-04-Amendment-1.pdf regarding US4. UBPO US0 is not currently defined in G.993.2 or G.997.1 for UPBO and is effectively turned off.) Parameter values for “a” and “b” vary by VDSL2 profile and US band, but follow these conventions (per G.997.1): • Parameter “a” ranges from 40 dBm/Hz to 80.95 dBm/Hz in steps of 0.01 dBm/Hz. • Parameter “b” ranges from 0 to 40.95 dBm/Hz in steps of 0.01 dBm/Hz. • The set of parameter values “a” = 40 dBm/Hz, “b” = 0 dBm/Hz is a special configuration to disable UPBO in the respective upstream band.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

33

TR-114 defines a group of UPBO settings for Annex A testing as follows: Parameter

UPBOKLF

Setting

Description

1

CPE SHALL be forced to use the kl0 of the CO-MIB (UPBOKL) to compute the UPBO

UPBO setting for upstream performance requirements: UPBOA US0

40

UPBOB US0

0

A and B values US band 0 (These values imply no UPBO)

UPBOA US1

53

A value US band 1

UPBOB US1

21.2

B value US band 1

UPBOA US2

54

A value US band 2

UPBOB US2

18.7

B value US band 2

Table 8 - UPBO Settings, Annex A The appropriate values for parameter “a” are the same as TR-114 and agrees with ATIS 0600027. • • • •

Value “a” Value “a” Value “a” Value “a”

for US0=40 for US1=53 (limit PSD mask -3.5dB) for US2=54 (limit PSD mask -3.5dB) for US3=40

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

34

Appropriate default for parameter “b” TR-114 seems to define “b” for US1 as a Reference Length of approximately 2200-2300 ft and for US2 as about the same; these settings are for profile 8d testing. WT-114 sets these values for profiles 8a, 12a and 17a to 16.2 for US1 and 10.2 for US2. This implies a 2000 ft loop for US1 and 1200-1300 ft for US2 (26 AWG). The value of “b” is dependent on the amount of US power cutback required and the actual loop length in the network. The higher the “b” value in dB, the greater the cutback and the lower the US rates. • • • •

For US0 the answer is easy; parameter “b” is set to 0 (along with setting “a” to 40, turns off UPBO) For US1 set to 16 (approximates 2000 ft loop) For US2 set to 16 (approximates 2000 ft loop) For US3 set to 0

Where kl0 comes into play Kl0 (electrical length) can be “forced” by the CO-MIB or derived by various combinations of CO and CPE negotiations. Per WT-114, Issue 2; CPE SHALL NOT be forced to use the kl0 of the CO-MIB (UPBOKL) to compute the UPBO. This means we use the kl0 values derived by the MIB settings as applicable. Neither G.997.1 nor G.993.2 provide a precise definition of how the CO and CPE are to automatically determine kl0. Consequently, Calix’ primary xDSL vendor, Broadcom, has defined the following methods. The method to use is selected by the implementation, or on some of Calix’ products, by the end-user. Broadcom’s document recommends kl0=-1, as defined below. • • • •

kl0=-1: final electrical len = max(kl0_CO,kl0_CPE) (this is the method used by the E7 VDSL2 implementation when Kl0 is set to “no-force”) kl0=-2: final electrical len = min(kl0_CO,kl0_CPE) kl0=-3: final electrical len = kl0_CO kl0=-4: final electrical len = kl0_CPE

It is important to note that no UPBO is applied for loops whose electrical length kl0 is greater than the RL that is determined by the setting of “b”. Downstream Power Back Off (DBPO)

Subscriber lines in a binder group can have some lines directly connected to the Exchange and others connected to closer DSLAM in RTs. Power on lines that run from the RT tends to be higher than on lines that run from the CO, or exchange, leading to problems with NEXT on the lines in the binder group.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

35

DPBO applies PSD Spectral Shaping to the frequencies of the RT lines that overlap with the CO lines. This allows you to reduce the strength of the RT signal while maintaining a sufficiently high data rate on the CO signal. This method is called the “exchange side model” and is also known as the “E-side model” as defined in G.997.1. The PSD from the RT is configured to be equal or below the PSD from the CO on the frequency range used for downstream traffic by both the RT and the CO. This is done by predicting the shape of the attenuated signal (exchange PSD (EPSD) when it reaches the RT by configuring the E-Side cable Model and E-side electrical length (ESEL) parameters. There are three “E-Side Cable Model” parameters - A, B and C. These parameters define the frequency behavior of the E-Side Cable. Each cable has different signal characteristics. The “ESEL” parameter is an estimate of the loop attenuation. Besides the “E-Side Cable Model” and ESEL parameters the following other considerations must be configured: •

The minimum and maximum frequency values (FMIN/FMAX) define the frequency range that should have attenuation applied. This is the frequency spectrum that should be protected from NEXT. The FMIN and FMAX values are based on the services type (ADSL2+, VDSL2 with and without US0) provided at the CO site.

•

The minimum usable signal level (MUS) is the amount of attenuation applied between FMIN and FMAX. The higher the attenuation applied, the lower the available bitrate will be at the RT location.

More information on how to configure DPBO can be found in ITU G.997.1 and TR-115 documentation. RFI Notching Electromagnetic interference (or EMI, also called radio frequency interference or RFI) is a disturbance that affects an electrical circuit like a phone line due to either electromagnetic conduction or electromagnetic radiation emitted from an external source. The disturbance may interrupt, obstruct, or otherwise degrade or limit the effective performance of the circuit. HAM radios are major sources of RF. Other sources are artificial or natural objects that carry rapidly changing electrical currents, such as an electrical circuit, AM radio station, the sun or the Northern Lights. Any interference that occurs within the 30 kHz to 17 MHz frequency band (ADSL/VDSL2 frequency range) could cause some severe service degradation to a DSL line. Note: Be careful not to exclude VDSL pilot sync tones when creating RFI gaps, as this may lead to the modem not being able to train. These essential VDSL tones differ from line to line. If you are using RFI notching to screen for HAM radio bands you will typically not need to worry about this issue, as the bands that HAM radios normally use are outside these VDSL pilot sync tones.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

36

G.INP Impulsive Noise on xDSL lines can contribute to a poor customer experience, especially with respect to video services. Previously, mitigation techniques have revolved around additional SNR (by setting a higher SNR target margin) and in the past, by leveraging the error correction that is achieved through Interleaving. Video delivery based on UDP packet traffic is insensitive to delay, however TCP/IP based video delivery is very susceptible to delay. Modern Video delivery has been moving to TCP/IP based delivery and therefore the delay caused by Interleaving is no longer tolerated by the system. Prior to G.INP being standardized, Broadcom Corporation created a retransmission technique known as “PhyR®’ and Ikanos Corp had a technique known as “Erasure®”. Both of these techniques contributed to the standards based retransmission now known as G.INP. G.INP is defined in ITU document G.998.4 Key parameters for G.INP are as follows: •

MINETR_RTX: Minimum effective throughput. Roughly equivalent to minimum train rate when using interleaving

•

MAXNDR_RTX: Maximum net data rate. Same as maximum train rate when using interleaving.

•

DELAYMAX_RTX : Maximum allowed delay for retransmission

•

INPMIN_SHINE_RTX: Minimum impulse noise protection against SHINE.

•

INPMIN_REIN_RTX: Minimum impulse noise protection against REIN

•

IAT_REIN_RTX: REIN inter-arrival time. This is how you provision that REIN is caused by 60Hz AC or 50Hz AC.

•

SHINERATIO_RTX: The loss of rate in a 1 second interval expressed as a fraction of NDR due to a SHINE impulse noise environment expected by the operator to occur at a probability acceptable for the services.

The current defaults for the Calix E7 VDSL2-48(C)/E3-48C and E5-48(C) platforms are as listed below and pending further characterization are the recommended settings. Minimum effective throughput

: 32K

Maximum net data rate

: 128.000M

Max delay for retransmission

: 20 ms

Min INP against SHINE

: 4 symbols

SHINE Ratio

: 0.010 NDR

Min INP against REIN

: 0 symbols

REIN inter-arrival time

: 120 Hz

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

37

G.Vector G.Vector is described in ITU-T document “G.993.5 Self-Fext cancellation (vectoring) for use with VDSL2 transceivers”. The purpose of G.Vector is to help VDSL lines overcome the biggest factor effecting link speed: Cross Talk. G.Vectoring allows the VCE(Vectoring Control Entity) to analyze on all the ports in the G.Vector group (For G.Vector to work all ports in a binder need to be in the G.Vector group). The analyses of the signals then allows the VCE to mathematically induce compensating/pre-distorting into each DLS link to minimize the effect of cross-talk . There are two types of Vectoring; System level vectoring and Unit level vectoring, which are defined by the location of the VCE. In Unit level vectoring the VCE is located on the ESeries VDSL2 product while in System level vectoring the VCE is located on a dedicated processor card. Calix currently supports Unit level vectoring and is working on System level vectoring solution. Further reading on G.Vector and in depth look how it works can be found in BBF MD257i2 and BBF TR-320. The G.Vector interoperability test plan is TR-249i1.

G.Fast G.Fast allows speeds of up to 1 Gbit/sec over a copper pair. The standard ITU G.9700 and G.9701 is undergoing final approval and work is ongoing defining the products that will support G.FAST. To flow the work, keep tabs on BBF WT-301 and BBF WT-308.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

38

Bonding (ATM/PTM) DSL bonding allows multiple loops to be aggregated into what appears to the applications as a single facility. The standard which Calix follows with bonding is ITU-T G.998.1 - ATMbased multi-pair bonding and ITU G.998.2 G.bond for PTM Ethernet pair bonding. ATM based Multi-pair bonding: • • • •

Supports dynamic removal and restoration of pairs without human intervention Supports disparate data rates, up to a ratio of 4-to-1 (fastest to slowest), amongst its pairs Incurs a maximum overall one-way bonding delay of 2 ms Is transparent to the applications and the overall access network

Additionally, G.998.1 has a number of advantages for supporting the rapid deployment of IP-based services: • • •

It leverages the existing ATM based infrastructure that supports existing ADSL deployments. It requires minimal modification to existing ADSL transceivers shortening development times and costs ATM-based DSL bonding does not require a common clock across the bonded links. This both simplifies development on the DSLAM and allows bonding to be implemented as a centralized function allowing links in the bonded group to be assigned to any port on the Access Node.

ATM-based bonding utilizes a modified ATM Header in which certain bits in the cell header are used to support a sequence number that allows reassembly of the inverse multiplexed cells, regardless of differences in the bitrate or differential delay of the DSL loops making up the bonded facility 1. The following figure illustrates the layout of a normal ATM Cell Header and those used in ATM-based DSL bonding. Unused bits in the ATM Virtual Circuit Identifier (VCI) or the VCI and Generic Flow Control (GFC) are overwritten to contain the bonding Sequence ID (SID). Either an 8 bit or 12 bit sequence ID can be supported. The longer SID supports greater differential delay between individual loops than the shorter one). Since the ATM UNI for a DSL interface does not use the GFC field, or the high order 8 bits of the VCI 2, overwriting this data in the header will not affect the ATM information in the header. Ethernet based Multi-pair bonding (PTM Bonding): The PTM bonding recommendation specifies portions of clause 61 of IEEE Standard 802.3ah-2004 amendment to Carrier Sense Multiple Access with Collision Detection (CSMA/CD) access method and physical layer specification as a normative reference and identifies the requirements for multi-pair bonding in IEEE 802.3ah-2004 that are different in the United States. Further, this recommendation specifies the requirements for extending the 1

This is contrasted with the Inverse Multiplex over ATM (IMA) protocol that cannot support differential delay or bandwidth differences between the links in a bonding group. 2 Typical ADSL ATM UNIs support far fewer than 255 Virtual Circuits at the ADSL “U” Interface. Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

39

bonding methods in IEEE 802.3ah-2004 to xDSL technologies other than VDSL and SHDSL. The following are the objectives this recommendation: • To provide support for operation of xDSL technologies on multiple pairs of voice grade twisted pair cable. • To provide 100 Mbit/s burst data rate at the Ethernet media independent interface using Rate Matching. • To provide full duplex operation. • To provide a communication channel with a mean BER at the α/β service interface of less than 10–7. Ethernet-based PTM pair bonding is another method of delivering additional bandwidth over longer distances to subscribers. PTM bonding effectively flattens the steepness of the performance degradation curve allowing for higher bandwidth subscriber services to be pushed further away from the DSLAM. In addition to the VDSL2 spectrum performance, another advantage of PTM bonding is that pairs need not be adjacent to one another on the DSLAM. Historically, ATM bonding typically enabled across adjacent pairs within a bonding group (a group of ports that may be bonded). PTM bonding allows for an available port to be bonded to a previously deployed subscriber line.

ADSL2+ The ADSL2+ standard accommodates booth ATM and PTM based data transport. PTM based ADSL2+ single port and bonding modems are just now entering the market For a list of Calix DSLAM’s that will support PTM over ADSL2+ services for single pair or bonded pair, please see the feature matrix Appendix.

VDSL2 The VDSL2 standard directly supports PTM data delivery and modern DSL chipsets will also support ATM transport in ADSL2+ fallback mode. Note: VDSL2 bonding is restricted to a max 120 Mbits/sec with 17a bonding on most Calix DSLAMs. Note: Current VDSL2 CPEs may be restricted to profiles 8a,b,c,d on each bonding member, meaning nothing is gained by deploying VDSL2 bonded groups as short line lengths. Testing has shown that in a 26 Gauge plant until the loop is beyond 2,000ft nothing is gained over a 17a VDSL2 connection.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

40

Figure 8 - Sample Bonding Performance Note: 17a bonding performance will be added

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

Chapter 3

Designing a Networking Service Model

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

42

Introduction In order to develop an effective business plan, service providers must determine what services they will offer their subscribers, what technologies will be used to offer these services and how far these services can reliably reach into the network over xDSL. The decisions are influenced by a number of elements, each of which must be considered in relation to the others in order to establish a business plan. Regulatory elements such as franchise agreements, technical elements such as encoding technologies and the elements of current facilities including customer serving areas (CSA), loop gauge and spectral interference from other services all weigh in on each other when considering the services that can be offered and the corresponding resources that will be required to successfully deploy and support these services.

Bandwidth Requirements Video Stream Bitrates Many different manufacturers offer video headend encoders. Calix has tested and certified several encoders as part of its Calix Compatible™ Program. The video encoder digitally encodes and compresses all analog programming per MPEG-2 and MPEG-4 standards. Additionally, the encoder encapsulates the MPEG video frames into User Datagram Protocol (UDP) over IP packets to support Internet Group Management Protocol (IGMP) video signaling for standards-based IPTV implementation. Service providers should optimize the encoding rate to the lowest rate possible without sacrificing video quality. Optimizing the encoding rate benefits the whole network by minimizing bandwidth requirements for both the transport and access portions of the network. MPEG-2 encoding rates for typical deployments vary between 2.7 Mbps and 4.0 Mbps per video channel for standard definition, depending on the desired quality of encoding and the sophistication of the headend vendor. Each service provider must determine its preferred encoding rate. MPEG-4 is a digital multimedia transmission standard that has a more efficient compression capability than MPEG-2 (more than 200:1 compression).A key feature of MPEG-4 is its ability to manage separate media components within image frames. These media objects can be independently controlled and compressed more efficiently. MPEG-4 AVC (High Definition) encoding rates will vary between 6.0 and 8.0 Mbps. MPEG-4 can model media components into 2 dimensional or 3 dimensional scenes. It has the ability to sense and adjust the delivery of media dependent on the media channel type such as reliable broadcast or unreliable Internet. MPEG-2 and MPEG-4 video is generally transported as either Capped Variable Bitrate (VBR) or Constant Bitrate (CBR).

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

43

Recommendation 2 - Video Stream Bitrates There are no universal best practices for setting the optimal MPEG-2 or MPEG-4 encoding rate. Each video deployment has its own unique requirements. For example, a deployment using 50 broadcast channels has different requirements than a deployment using 200 broadcast channels. Also, different channels can be encoded at different rates. Channels with high action content, such as sports programming, may require higher encoding rates than other channels. Further, video quality may vary from one encoder vendor to the next. A 2.7 Mbps MPEG-2 stream from Vendor A's encoder may produce a perfectly acceptable picture, while a 2.7 Mbps MPEG-2 stream from Vendor B's encoder may produce an unacceptable picture. From a best practices perspective, it is important to account for the channel(s) with the highest encoding rate requirement, and recognize that each video stream requires additional overhead in excess of the video payload. Regardless of the MPEG encoding rate selected, service providers should plan to add an additional 3.5% of bandwidth to account for the Ethernet/IP/UDP overhead. Also, an additional 14.5% of bandwidth must be added to account for the ATM, AAL5, 1483/2684 overhead, for a total of 18% additional overhead. Some examples of the associated overhead added to video streams (at various baseline MPEG-2 encoding rates) are shown below.

MPEG-2 Stream

Example 1

Example 2

Example 3

Basic MPEG-2 Payload Rate from Encoder

3.000 Mbps

3.200 Mbps

3.750 Mbps

Ethernet/IP/UDP Packet Rate (adds 3.5%)

3.105 Mbps

3.312 Mbps

3.882 Mbps

ATM, AAL5, 1483/2684 (adds 14.5%)

3.555 Mbps

3.792 Mbps

4.445 Mbps

Table 9 - MPEG-2 Rates with Overhead Added Note that when you source pre-encoded content from another provider, you must understand the encoding rate used on the sourced content so that connection bandwidth can be planned accordingly. For example, Video On Demand (VOD) content providers who also provide content to the Multiple Service Operators (MSO) may use MPEG-2 encoding rates as high as 3.75 Mbps, which is the maximum allowed by the standard*. In such cases, ensure that your xDSL profiles for IPTV are configured with enough bandwidth to accommodate such rates (in this example 4.445 Mbps per channel). Note: See CableLabs specification MD-SP-VOD-CONTENT-I01-020327 or the CableLabs Web site (www.cablelabs.org) for more information.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

44

Video Middleware Bitrates STBs present a channel guide, often called the electronic program guide to the viewer, which typically includes program information for all channels in a given package. Packages are compositions of channels with different pricing targeting different demographics. The same home may have a children’s package associated with one STB and a sports package associated with another. In order to present the correct program guide and keep the corresponding data current, the STB needs to interact with a middleware server. In addition to program guide updates, other interactions occur between the STB and middleware. Enumerating every interaction is unnecessary, but a clear accounting from the middleware vendor of the total capacity required for upstream and downstream communications is very important. Example Middleware Bitrate Estimation • Real time communication between middleware and STB – 16kbps to 512kbps downstream • Real time communication between middleware and STB – 16kbps to 64 kbps upstream • STB boot streams – 3mbps downstream

Video Encryption Bitrates Video encryption typically adds a very small amount of overhead and thus must be considered in the video stream bitrate calculation. Video encryption may also require data exchange between the STBs and a certificate authority. Be sure to determine the overhead added to each individual stream as well as any signaling for key transfer that must occur in order for the STBs to decrypt the incoming video.

Internet Bitrates Internet access is typically advertised as a best-case rate. Actual Internet throughput is difficult to measure; protocol overhead varies with packet size and higher layer protocols like TCP use congestion control mechanisms that can significantly affect throughput in the presence of congestion and delay. Many speed tests are web based, click a button and a result is returned indicating how fast the link is. Bottlenecks between the web server and the subscriber may have nothing to do with the service provider-s network. If a web server sat right behind a DSLAM and a subscriber did a speed test expecting to see IP throughput equal to what was advertised in their service contract, headroom must be left for IP, Ethernet and ATM framing. Using the overhead estimations provided in Table 11 - ; we would need to allocate 12% overhead for Internet traffic. Example Internet Bitrate Calculation – 12% Overhead Downstream

Upstream

Advertised Bitrate:

4.0 Mbps

512 Kbps

Required Bitrate:

4.48 Mbps

574 Kbps

Table 10 - Bitrate Calculation Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

45

Supporting Tiered Service Offerings Service providers often sell different packages to their customers based on the number of TVs in a home and the speed of the Internet connection. The different packages have different bitrate requirements, giving the service provider the option to offer certain packages deeper into their network than others. As an example, a customer with 3 TVs and a 4 Mbps Internet service may be a high-end package, while some subscribers may only be interested in 1 TV and 1 Mbps Internet. Recommendation 3 - Application Specific DSL Profiles Calix recommends provisioning DSL profiles to reflect the bitrate requirements of specific service offerings. This requires service providers to represent each service offering with a unique DSL profile and coordinate the provisioning of DSL ports on a DSLAM with the corresponding service offered to the customer. This allows lines closer to the DSLAM to enjoy added margin and reduced power, reducing over-all system crosstalk and increasing service reliability.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

46

Application Bitrate Requirements The internal architecture of a CPE modem often includes a small processing engine for ATM segmentation and reassembly (SAR) and application protocols such as PPP or NAT. The use of different applications at different bitrates will exercise this processor and could present a bottleneck that affects loss sensitive applications like IP video. To understand the performance requirements of a CPE modem, a careful study of the worst-case application bitrates is required. The sections in this chapter discuss common applications and considerations that must be taken in determining the required bitrates for a service offering. The details of protocol overhead computation are discussed later in this document. To gain a general sense of the application bitrate requirements the following overhead estimate 3 are used:

Protocol

ProtocolOverhead

Estimated % Overhead

TCP

20 bytes/packet

1.5%

UDP

8 bytes/packet

0.5%

IP

20bytes/packet

1.5%

Ethernet

26 bytes/frame

2.0%

ATM

5 bytes/cell

10.5%

Table 11 - Protocol Overhead Estimates

3

Overhead estimations built using the following assumptions: • 1300 byte TCP or UDP payload • No TCP timestamps • No IEEE 802.1q tag on the Ethernet frame • ATM AAL-5 with perfect cell alignment Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

47

Customer Premises Access Equipment Residential Gateway Selecting a residential gateway with the correct feature set is an obvious task, but determining the correct feature set requires a careful look at the application requirements. The application’s networking service model and the application’s bitrate requirements must be understood. Once those two parameters are understood the residential gateway can be selected based on required features and performance.

Application Networking Service Model The application’s networking service model can be thought of as the subscriber home network scheme and the corresponding requirements placed on the residential gateway, the DSLAM, and the core network infrastructure. There are many options to consider in selecting a service model. The highest level decision is if video and data are equally available from any device in the home or exist on fundamentally different networks. Traditional IPTV deployments utilize separate networks for set top boxes (STB) and personal computers (PC). New technologies, especially those offered with Microsoft middleware and STBs encourage the integration of functionality in the home where a PC or STB would have access to the same resources, be it games, HSI, VOD, multicast video or other IP based service innovations sure to penetrate the home entertainment market in the coming years. Application Separation If STBs and PCs are to be kept separate, STBs are typically given private IP addresses and communicate on a network wholly separate from PCs. The residential gateway is responsible for maintaining this traffic separation back to the DSLAM. Once the choice is made to separate services the question of how must be answered. In the VDSL2 case the residential gateway may support a virtual local area network (VLAN) atop a single service. In the ADSL2+ case the residential gateway may support the separation at the Asynchronous Transfer Mode (ATM) layer by using a virtual circuit (VC) per service or at the Ethernet layer by using a VLAN per service atop a single VC. In either case, the residential gateway must ensure quality of service (QoS) to prioritize video signaling traffic above best effort Internet traffic. Failure to do so can result in a subscriber saturating the upstream link with Internet activity, preventing STBs from managing video streams and potentially causing intermittent video blackouts or other unacceptable user experiences throughout the home. Recommendation 4 - IPTV and HSI Service Recommendation IPTV and HSI should be on separate services.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

48

PPP vs. DHCP Traditional broadband access was often implemented with Internet Protocol (IP) over Point to Point Protocol (PPP). IP video deployments utilize IP over Ethernet for all video related traffic. PPP adds protocol overhead, does not support multicast and PPP clients embedded in CPE rarely have the horsepower to process the kinds of bitrates required for IPTV. STBs typically obtain an IP Address via DHCP and though DHCP is an application, not a transport protocol, when discussing connectivity options people often discuss these two approaches as “PPP vs. DHCP”. When utilizing application separation between video and Internet networks, a hybrid approach may be taken as each application can be viewed as its own network presence. The CPE can map Ethernet ports used by PCs to a PPPoE client that communicates on one service while video is bridged on a second service. Recommendation 5 - Use of DHCP for IPTV Services Use DHCP to manage the STBs boxes for IPTV services. In-band Residential Gateway Management It is crucial to have network access to the CPE for Operations, Administration and Maintenance (OAM) functions. Calix recommends using a separate VC for in-band CPE management. The CPE must provide an interface that supports a DHCP client that is not accessible from the LAN ports or the other VCs in use. This provides a secure channel between the CPE Modem and the provider’s network operations center for in band management of network elements. Used in conjunction with DHCP Relay, this allows for CPE Modem deployment with no customer specific configuration. Recommendation 6 - Use TR-69 for CPE Management Use the network access CPE deployment model as it allows firmware upgrades and reconfigurations of CPEs without the need for truck roles. Testing in the labs has shown, especially for VDSL CPEs, that having the latest firmware on CPEs is crucial to the operation of the CPE and DSLAM.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

49

Upstream Quality of Service (QoS) The residential gateway must be configured to support QoS upstream. The DSLAM will prioritize downstream traffic appropriately, insuring that video is never impacted by bursty data traffic. However, the DSLAM cannot prevent the subscriber from unleashing high volumes of traffic from a PC, effectively initiating a denial of service attack against itself. Without QoS this subscriber traffic has the potential to starve critical network control traffic like IGMP. Recommendation 7 - QOS on the CPE On the CPE always enable upstream QoS with the IPTV service configured as a higher priority than all other services.

Different Residential Gateway (RG) Service Models RG Model • • • • •

Microsoft Mediaroom use only—all STBs and PCs are assigned private addresses behind a single IP interface on the RG. Other middleware vendors do not yet support this ‘true’ RG model. Data and video share the same VLAN between the RG and the up-stream router. The Calix DSLAM passes the single VLAN through from the RG to the up-stream router. The up-stream router separates video from data and routes to the appropriate destination. The up-stream router employs QoS to limit data bandwidth to the amount of remaining bandwidth not used by video.

Pseudo RG Model • • • • •

The RG operates at layer two, switching the data between STBs and PC to the up-stream router. Data and Video share the same VLAN between the RG and the up-stream router. The Calix DSLAM passes the single VLAN through from the RG to the up-stream router. The up-stream router separates video from data and routes to the appropriate destination. The up-stream router employs QoS to limit data bandwidth to the remaining bandwidth not used by video.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

50

QoS Enabled RG Model • • • • • •

Microsoft Mediaroom use only as all STBs and PCs are assigned private addresses behind a single IP interface on the RG. Requires an RG configured to route and capable of its own QoS functions as well as supporting Dual Routed WAN Interfaces. Video and data are separated by the RG and sent to the up-stream router on two different VLANs. The RG uses built in QoS to bandwidth limit data to a rate available after video. The Calix DSLAM separates the two VLANs from the RG and sends them through to the up-stream router. The up-stream router forwards the video and data VLANs to their separate destinations.

Bridged Video and Routed Data Model • • • •

This model can be used for Minerva, Myrio or other middleware providers that do not require large amounts of unicast traffic. Video is bridged and data is routed at the RG. Different ATM VCs or VLANs separate video and data traffic at the RG. The Calix DSLAM, or an up-stream router, forwards the separated traffic to the appropriate destination.

RG DSL Chipset Vendor and Firmware Revision The specifications of ADSL2, ADSL2+, VDSL2 and many interoperability test initiatives have come a long way in improving interoperability of DSL between different vendor equipment. The Broadband Forum has published interoperability test plans, initially with TR048 and then updated to TR-067 and sponsors interoperability “plug fests” through the DSL Consortium at the University of New Hampshire’s Interoperability Test Lab (UNH-IOL). The follow on to TR-67 is TR-100 for ADSL2+ and both TR-114 and TR-115 for VDSL2. All the major ADSL2+ and VDSL2 chipset vendors continue to work to perfect interoperability with one another but each vendor tends to have best performance with their own chipsets on either side of the line. This is primarily due to proprietary techniques vendors implement to improve performance but also due to the fact that they have a better ability to test between their own components than those of a competitor. ADSL2+ and VDSL2 chipsets are typically implemented with digital signal processors (DSPs) controlled by firmware. The firmware implements the majority of the DSL protocol and thus has the potential to significantly impact interoperability. A change to the DSL firmware at either the residential gateway or the DSLAM has the potential to affect the performance and reliability of the network.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

51

Calix does extensive testing against a selection of popular residential gateways with the best of breed chipsets including but not limited to Broadcom, ADI, Ikanos, Texas Instruments and Lantiq. Calix also works with the DSL Consortium at UNH-IOL to enable neutral 3rd party interoperability. UNH-IOL has best of breed calibrated test equipment and experts whose sole purpose is to test DSL interoperability. UNH-IOL co-chairs the Broadband Forum Test and Interoperability Working Group and remains focused on continually improving industry-wide performance, feature and functional tests found in the Broadband Forum Technical Reports. This work is done in concert with their development of automated test suites, providing them a unique ability to keep their test capabilities current with the latest industry innovations and interoperability requirements. Calix is a member in good standing of the UNH-IOL DSL Consortium and strongly endorses the quality, thoroughness and impartiality of their interoperability testing.A key change in Calix’s interoperability recommendations is based on the UNH-IOL’s unique capability to quickly test any version of a member’s DSL Modem against Calix DSLAMs. Calix encourages all DSL Modem partners to provide UNH-IOL test results as proof of interoperability at a level of detail that is crucial to service provider deployment planning. Recommendation 8 - Interoperability Testing Calix strongly recommends that thorough interoperability testing be performed prior to changing residential gateway DSL firmware in a production network. Contact your Calix representative to determine if a given residential gateway DSL firmware revision has been tested and is supported with the DSLAM DSL firmware revision being used.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

52

Access Node Service Radius Once the planning work has been done to determine the total bitrate requirements of a service we must determine how far into the network those bitrates are achievable. This determination is crucial to a service provider’s business plan, the further the technology can be pushed the higher the number of potential customers. However, performance declines as the technology is pushed further. ADSL2+ and VDSL2 use Discrete Multi Tone (DMT) technology to partition transmission into small frequency bands called subcarriers across the spectrum between 30 KHz and 30 MHz. Signals lose strength over distance, especially at the higher frequencies. Signals are further impacted by many factors including the quality of the facilities, the number and types of other signals present in the plant and external disturbers such as AM radio or street lamp fluorescent ballasts that operate in the same frequency band as xDSL. One of the most difficult challenges in deploying IP video is accurately determining the rate/reach capabilities of DSL in a specific environment. Consider the following variables: • • • •

DSL configuration Outside plant condition Inside wiring practices External noise sources

All of these will have an effect on the attainable bitrates at a given distance. To complicate matters, service providers rarely have perfect engineering plant information so the actual distance of a subscriber from a CO or RT may be different than the plant records indicate. Add all these variables together and you can see that determining whether a service can reach a specific subscriber can be quite a challenge. The ultimate business goals compete with each other and can be summarized as follows: • •

Maximize the number of customers served Minimize the number of service issues

Spectral Interference xDSL performs best when its neighbor pairs are dormant or only running POTS. The term “neighbor pairs” refer to neighboring pairs in a binder or pairs in an adjacent binder. A signal traveling through a twisted pair will emit some amount of electro-magnetic interference (EMI), as twisted pair cabling does not perfectly insulate the signals traveling across it. Other transmission systems like ISDN, T1 and even neighboring xDSL will interfere with an xDSL line when run on pairs in the same binder. The worst disturbers will affect adjacent binders as well. When determining how far an xDSL service can reach, consideration must be given to the other services running with it. From cable trays in the CO through the distribution plant, other services have the potential to impact xDSL. In addition, xDSL will affect itself as “fill”, or the number of adjacent pairs running xDSL is increased.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

53

Standard CO wiring practices bring twisted pair from an MDF to a DSLAM, filling up ports on the DSLAM and thus creating high concentrations of xDSL in the cabling that runs from the DSLAM to the MDF. As xDSL becomes more popular in a provider’s serving area the distribution plant will see increases to the total fill rate as well. In order to plan for success, any estimates on xDSL performance at a particular distance must take into account all disturbers anticipated to be present. At minimum this should include planning for fully utilized cabling between the MDF and DSLAM, i.e. 100% utilization of DSL on the pairs in all cables and components between the MDF and DSLAM. Requirements for spectrum management are standardized in ANSI T1.417-2001. This standard is distilled in a white paper published by Telcordia, available on-line. For details refer to [SPEC-WP] in the bibliography. New xDSL research and emerging standards are now focusing on dynamic spectrum management (DSM). If you are curious about the future of xDSL and the exciting new developments being made in the area of DSM we encourage you to visit the website of Stanford Professor John M. Cioffi [CIOFFI-DSM].

Estimated Rate vs. Reach Included in the appendices are data sets Calix has accumulated testing rate versus reach using calibrated test equipment and different noise models. These were tested using Calix’s recommended video DSL profile and show train rates at distances in one thousand foot increments. This data provides service providers an indication of xDSL performance in very specific test environments, which can be used as a starting point in determining how far a service can reach. To better characterize the performance of a particular provider’s plant testing must be done with the DSLAM, residential gateway and DSL profile with which the service is to be offered.

Measuring Rate vs. Reach The estimated rate/reach data is a great starting point but cannot be used by itself to plan a serving radius. In order to understand how your facilities match up to the estimates, tests must be run in the field. The maximum distance a specific xDSL profile can reach using a specific xDSL firmware load on the DSLAM and residential gateway must be taken into account. Understanding the delta between the estimated rates and measured rates at a given distance gives you a sense of how well your facilities compare to the simulated facilities used to model the estimate. Calix maintains a relationship with UNH (University of New Hampshire) IOL (Interoperability Laboratory). UNH performs “Standards” based xDSL testing based on the following Broadband Forum specifications: •

TR-67 for ADSL

•

TR-100i2 for ADSL2+

•

TR-114i2 for VDSL2

•

TR-115i2 for VDSL2 (functionality)

•

TR-249 for G.vector

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

54

•

TR-273 for bonding

The goal of Calix is to maintain a standardized testing strategy to insure that we meet the test requirements outlined by the Broadband Forum. This does not imply that we meet all “expected” test results for all Calix platforms, it simply means we maintain a standardized set of test methods for applicable test cases. Specialized testing that is outsideof the scope of the standards is also a regular practice in Calix IOL labs and may include such testing as: •

Empirical ring trip testing with or without video.

•

OSP testing to verify network elements such as: MDF’s, cross connects, CO POTS splitter shelves, protection panels, and so forth.

•

Calix generated bonding testing.

•

Participating in industry Plugfests at the IOL/UNH lab.

•

Custom test cases based on customer requests. Recommendation 9- xDSL Performance Testing Calix recommends performing measurements to establish actual rate/reach performance in your environment. Measurements should be taken using the DSLAM, residential gateway and DSL Profiles corresponding to the service being offered. Results at a given distance will vary, it is important to take multiple samples and average the results in order to gain an accurate idea of what the technology is capable of on your plant.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

55

Zoning Model The periphery of the DSLAM serving radius will tend to have marginal performance. Some lines just outside the serving radius may work fine and lines just inside it may not be able to support the service. No two pairs are the same. To accommodate for these discrepancies a zoning model can be used. Consider a red zone to start at the maximum distance required bitrates are achievable. It is not likely that the service can be established in the red zone. Subtract 10% of that distance to define the perimeter of the green zone. The service provider should expect that the service can be established for all subscribers in the green zone. In between the green and red zones is a yellow zone where the required level of service may or may not be achievable. Pre-qualification is recommended for subscribers in the yellow zone. Recommendation 10 -Zoning Model Recommendation Calix recommends defining green, yellow and red serving areas for each DSL profile used in a deployment. Red Zone: Any distance greater than or equal to the distance measured as the maximum rate/reach capability of a DSL Profile. Green Zone: A distance 10% less than the Red Zone. Any subscriber within the green zone should be provided service with confidence, unless their facilities contain abnormal impairments. Yellow Zone: The area between the green and red zones wherein DSL performance may be acceptable for one subscriber and unacceptable for another. Calix recommends service pre-qualification for any customer in this area.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

56

Calix DSLAM

Figure 9 - Zoned Serving Radius

Extending Reach with Bonding If a subscriber is in the red zone for a service, one option is to provision multiple DSLs to the home. This can be achieved with DSL bonding, which requires aresidential gateway (RG) that has two DSL transceivers, mated to a DSLAM capable of supporting bonding. Another approach is to simply use multiple gateways in the home, each with its own DSL and corresponding DSLAM port. Franchise agreements sometimes require service providers to offer identical services throughout a serving area. In situations where a subscriber is in a red zone but franchise regulations require they be allowed to order the service, multiple DSL lines may be a solution. When deploying multiple independent gateways to a home, each should be treated as an individual entity. If two gateways are provided to a home in order to meet an advertised service offering of 3 TVs and 4 Mbps data, two profiles could be used: • •

RG 1 - 2 TVs RG 2 – 1 TV + 4 Mbps data

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

Chapter 4 Infrastructure Every component between the DSLAM and the residential gateway has the potential to interfere with the DSL communications path and thus render a service unsatisfactory or completely inoperable. Common xDSL disturbers like bridge taps and loading coils are incompatible with IPTV applications over xDSL. The infrastructure closest to the BLC aggregates the highest number of xDSL signals close together and thus must be the most immune to crosstalk. Cabling between the MDF and DSLAM must be top quality, properly manufactured and distributed through a CO or cabinet with spectral management considerations taken into account. CPE master splitters must be used. Inline filters at each phone are unacceptable for IPTV applications. This chapter seeks to examine the typical infrastructure between a DSLAM and residential gateway, identifying the key areas a service provider has the most control over and the corresponding infrastructure or planning that should go in place. Note: Calix Qualified Components. Calix has qualified cabling, splitters and other components from multiple vendors. A list of preferred components is included in the “Recommended Third-Party components and Sources” section. The figure below depicts a reference diagram for infrastructure and facilities typically found between a CO or RT and the customer premise. A number of variations exist in the models and components service providers use to distribute lines from a central location to each subscriber, we do our best to capture the generic elements and place an emphasis on those points that are most likely to impact an IPTV service offering.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

58

Figure 10 - Infrastructure Reference Diagram [DSLF 2005.195] 1. DSLAM 2. Cabling between DSLAM and Splitter 3. Splitter 4. Cabling between Splitter and Terminal Block 5. Terminal Block 6. Connector/Protector 7. Feeder Cable 8. Distribution Cable 9. Drop Cable 10. Cross Connect/Serving Area Interface (SAI) 11. Pedestal 12. NID & Protector 13. CPE Master Splitter 14. Inside Wiring

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

59

Best Practices at the CO and RT

DSLAM

Head End

Splitter Shelf

Cross Connect Box

NID

xDSL Router/ RGW

STB

TV

Cross Connect Box

NID

xDSL Router/ RGW

STB

TV

w/xDSL Cards

MDF/ Pro Panel

DSLAM w/Combo Cards

1

MDF/ Pro Panel

2

3

4

KEY: 1. The Video Headend and associated equipment (video headend encoders, routers, application servers) 2. The Central Office (CO) and/or Remote Terminal (RT) equipment (Equipment, splitter card/shelf, cabling, main distribution frame, protection panel) 3. The Outside Plant (OSP) (cross connect box, cable binders, cabling) 4. The Customer Premises Equipment (CPE) (network interface device, wiring, ADSL router, set top box, television) The reference diagram shown in the figure above is relevant for both central offices and remote terminals. Central offices may support many different types of services and often have relatively long runs between the MDF and the BLC. Cabling between the MDF and the BLC is likely to be laid in cable trays alongside power or cabling supporting other services. Conversely, remote terminals are necessarily compact and support a limited number of services. In both scenarios, Calix recommends the use of CAT5e cabling between the MDF and the BLC (CO), or the protector panel and the BLC (RT). The traditional practice of CAT3 cabling on the plant side of the MDF is acceptable for both CO and RT applications.

POTS/xDSL Splitter Cards and Shelves xDSL applications require POTS/DSL splitter components at each end of the subscriber phone line. On the CO side, the splitter combines the baseband POTS service from the voice switch with the xDSL service. On the subscriber side, the splitter separates the baseband POTS service from the xDSL service. POTS/xDSL splitter components are available in a variety of form factors, but the two most common are integrated directly with the DSL line card (combination) or physically separate from the DSL line card (splitter shelf). Each has its advantages and disadvantages. When evaluating types and brands of POTS/xDSL splitters for use in video networks, the two key considerations should be how effectively splitters suppress crosstalk and ring trip errors.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

60

Crosstalk Crosstalk is a condition that contributes to xDSL performance degradation. Crosstalk is defined as interference caused by two signals becoming partially superimposed on each other due to electromagnetic (inductive) coupling between the conductors carrying the signals. The twisting of wire pairs keeps the crosstalk coupling between the balanced circuits on each pair at a reasonably low level at voice frequencies. However, crosstalk generally grows with increasing frequency, and since DSL frequencies extend into the megahertz region, crosstalk becomes one of the major limitations to highspeed DSL transmission. Therefore, for optimal performance, crosstalk management is of utmost importance.

Ring Trip Errors When a subscriber takes a phone off-hook while ringing voltage is applied, a surge of current runs through the circuit. The transformers on typical POTS/xDSL splitters are not robust enough to handle this surge. They saturate, causing a burst of errors on the xDSL line. Ring trip errors occur in every xDSL network, regardless of the xDSL equipment manufacturer. Note that the severity of the condition is influenced by such factors as the type of telephone used at the customer premises, loop length, and wire gauge. When ring trip errors occur on an xDSL line providing only HSI access service, the errors typically go unnoticed. The errors may cause lost packets, but higher-layer protocols request retransmission of the missing data, and the transmission continues as normal. However, when the xDSL line is also used for IPTV services, the burst of errors causes dropped packets that cannot be recovered. As a result, the subscriber may see a brief period of macroblocking or tiling on the television screen. In severe cases, the xDSL modem may even enter a re-train cycle, causing total loss of video for an extended period of time. Video subscribers will not tolerate the error conditions caused by ring trip, so the POTS/xDSL splitter used in video networks must withstand ring trip current surges and eliminate these line errors. For maximum effectiveness, such splitters must be used at both ends of the loop. Recommendation 11 - Use of COMBO Line Cards When possible, Calix encourages the use of COMBO (Integrated Voice and Data) line cards (with integrated splitters) instead of separate splitter DSLAM and POTS line cards, because the addition of another network element (the splitter shelf) into the system may potentially increase signal loss and crosstalk. It also increases cost. If the use of COMBO plug-in cards is not possible and a separate splitter shelf is required, see the “Recommended Third-Party components and Sources” section at the end of this document. Calix does not recommend the “overlaying” of COMBO equipment. There are too many variables to this topology to certify proper operation of services. Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

61

Note: See NID Protectors for more information about NID splitters for the customer premises. Remember that the CO-side splitter products operate as a matched set with the approved Customer Premises-side NID splitter products. Substituting an alternative product on either side may compromise performance. Calix has performed detailed tests comparing downstream ring trip coding violations between the recommend vendor splitter equipment and generic splitter equipment at both ends of the loop. Coding violations cause video tiling and macroblocking (and modem retrains in severe cases). In all cases, use of the recommended splitter equipment dramatically reduced or eliminated coding violations.

Cabling: DSLAM to Splitter Shelf Calix testing has shown that the type and quality of cabling used between the DSL shelf and the POTS/xDSL splitter shelf can significantly affect the SNR margin and performance. Recommendation 12 - Use of CAT5 Connectors and CAT5e Cable Use CAT5e rated cabling with CAT5 rated connectors for all xDSL connections between the Equipment and the splitter shelf. The cabling MUST be contiguous between the DSLAM and the POTS splitter shelf. Test results have shown that the use of CAT5e cabling (versus CAT3 cabling) between the DSL shelf and the POTS/xDSL splitter shelf reduces crosstalk from 2 dB to 4 dB in all cases. SNR margin improvements of this magnitude equate to higher throughput and/or longer loop reach to serve more customers, and should justify the modest increased costs of using CAT5 materials. Results shown in the tables below represent a typical example. Individual results may vary.

Equipment

CAT3

CAT5

Improvement

Comtest Splitter

-41.1 dB

-44.6 dB

3.5 dB

Generic Splitter A

-37.4 dB

-40.0 dB

2.6 dB

Generic Splitter B

-40.4 dB

-44.2 dB

3.8 dB

Table 12 - DSLAM to Splitter Shelf Test Results: Average Crosstalk Measurements; 24 AWG; 8 Kft

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

62 Equipment

CAT3

CAT5

Improvement

Comtest Splitter

-38.2 dB

-42.0 dB

3.8 dB

Generic Splitter A

-35.5 dB

-37.7 dB

2.2 dB

Generic Splitter B

-37.7 dB

-41.7 dB

4.0 dB

Table 13 - DSLAM to Splitter Shelf Test Results: Worst-Case Crosstalk Measurements; 24 AWG; 8 Kft The increase in SNR margin gained by using CAT5e cables translates into average downstream rate increases of 500 Kbps to 1.5 Mbps. The following test results compare CAT3 vs. CAT5e cabling by evaluating average downstream rates and worst-case rates using Comtest splitters and generic splitters. In all cases, the use of CAT5e cables between the DSLAM and splitter shelf provides a considerable increase in data rate.

Cabling: Splitter Shelf to Main Distribution Frame (MDF) Calix testing has shown that the type and quality of cabling used between the splitter shelf and the Main Distribution Frame (MDF) can significantly affect the SNR margin and performance. Recommendation 13 - Cabling between Splitter Shelf and MDF Use CAT5e rated cabling with CAT5 rated connectors for all POTS/xDSL connections between the splitter shelf and the MDF. The cabling MUST be contiguous between the DSLAM and the MDF. Test results have shown that the use of CAT5e cabling (versus CAT3 cabling) between the splitter shelf and the MDF improves crosstalk isolation by an average of 2 dB in all test cases. SNR margin improvements of this magnitude equate to higher throughput or longer loop reach to serve more customers, and should justify the modest increased costs of using CAT5 materials. Results shown in the table below represent three typical test cases. Individual results may vary.

Test Case

CAT3

CAT5

Improvement

1

46.1 dB

48.1 dB

2.0 dB

2

45.8 dB

48.1 dB

2.3 dB

3 46.3 dB 48.1 dB 1.8 dB Table 14 - Splitter shelf to MDF Test Results: Crosstalk Isolation @1.1 MHz; 24 AWG; 7 KftMDF Blocks

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

63

Several different styles of Main Distribution Frame (MDF) blocks are available on the market today. Calix tested three common types of blocks to determine whether different types had any impact on crosstalk isolation and performance improvement. The three styles tested are described as follows: WW-WW

Wire-wrap terminals on each side of the block. Within the block, the terminals are interconnected with CAT3 wire pairs.

CON-WW (CAT3)

RJ-21 connector on one side of the block (facing the DSLAM or splitter shelf) and wire-wrap terminals on the other side of the block (facing the protector panel.) Within the block, the terminals are interconnected with CAT3 wire pairs.

CON-WW (CAT5e)

RJ-21 connector on one side of the block (facing the DSLAM or splitter shelf) and wire-wrap terminals on the other side of the block (facing the protector panel.) Within the block, the terminals are interconnected with CAT5e wire pairs.

Table 15 - Styles of MDF blocks Recommendation 14 - Crosstalk Mitigation For crosstalk testing and xDSL rate/reach testing, CAT5 is the recommended MDF block.

Cabling: MDF to Protection Panel Most service providers use twisted pair jumper wire (hook-up wire) to cross connect each individual subscriber line from the MDF to the protection panel. Calix tested both CAT3 and CAT5e jumper wire (100 ft @ 24 AWG) to determine whether these two different types of wire had any impact on crosstalk isolation and performance improvement. Although there were minimal differences between CAT5e and CAT3 hook up wire for ADSL services at 1.1MHz, Calix recommends in all cases that all xDSL installations use CAT5 hook up wire. Recommendation 15 - Hook Up Wire Use CAT5e rated hook up wire for all connections to the MDF. The cabling MUST be contiguous between the MDF and the protection panel.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

64

Connectors

Figure 11 - Left: CAT5 RJ-21 Connector with Proper Workmanship

Right: UNACCEPTABLE RJ-21 Connector Special Note: Any connector that brings twisted pair out to pins to be mated with another connector must be manufactured with the utmost care to maintain the twist down to the pin. The butterfly tool commonly used to mate 25 pair cables with CHAMP AMP RJ-21 connectors must be used carefully as poor workmanship often leaves many pairs with unacceptable long untwisted regions.

RJ-21 Connector for CAT5 and CAT5e Cabling It is not sufficient for CAT5 cable assemblies to be put together with CAT5 or CAT5e grade UTP cables. This is specifically the case for VDSL2 service deployment. Cables assemblies must also include, at minimum in the equipment side, Type 525 RJ-21 connectors. 525 connectors provide a 6 dB margin which cuts crosstalk and assures Category 5 performance throughout the systems.

Figure 12 - Type 525 RJ-21 Connector (110° Right/Standard exit) versus Standard RJ-21 Connector (90° Right/Standard exit)

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

65

Cabling: DSLAM to Pro Panel See Appendix D for Pro Panel test results driving this recommendation. Recommendation 16 - Best Practice recommendation for Pro Panels Based on the test results, Calix recommends CAT5 Pro-Panels for all VDSL2 implementations and for all high BW ADSL2+ applications (>10mbps). Although certain xDSL profile settings i.e. INP, PhyR, G.INP and additional SNR Margin can be leveraged to mitigate xDSL line errors, the CAT5 Pro-Panels show improved rate vs. reach and much lower line error counts in CAT5 pro-panel networks. The most significant data point that forms the basis for this recommendation is line stability. This includes the RJ-21style CAT5 connector and CAT5e wiring.

Cabling: Protection Panel to the Cross Connect CO to Cross Connect The protection panel to cross connect wiring is commonly OSP. Since the cross connect is generally a long distance from the CO, this wiring tends to have a low impact on services. RT to Cross Connect Although there are multiple cases, the cross connect generally sits very close to the RT equipment and typically uses OSP between the RT cabinet and the cross connect cabinet. Note: The RT to cross connect wiring is currently under investigation for VDSL2 Services. Keep the wiring well dressed and properly terminated.

CO Surge Protectors (5-Pin Protectors) The 5-pin surge protectors used at the CO can create an imbalance in the pair (longitudinal balance), which adversely affects xDSL performance. (See Lighting Protection Section for information on 5-Pin Protectors and how to select them). Recommendation 17 - 5 Pin Protectors Use 5-pin surge protectors with a capacitive balance less than 1pF from tip-to-ground and ring-to-ground. Most protector manufacturers offer protectors with this option, generally indicated with a "BC" in the part number. When selecting 5-pin protectors, verify a capacitive balance of less than 1pF. Specifically Calix recommends the “BC” version of the Bourns 2410 for all xDSL applications. Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

66

Best Practices for Outside Plant DSLAM

Head End

Splitter Shelf

Cross Connect Box

NID

xDSL Router/ RGW

STB

TV

Cross Connect Box

NID

xDSL Router/ RGW

STB

TV

w/xDSL Cards

MDF/ Pro Panel

DSLAM w/Combo Cards

1

MDF/ Pro Panel

2

3

4

KEY: 1. The Video Headend and associated equipment (video head end encoders, routers, application servers) 2. The Central Office (CO) and/or Remote Terminal (RT) equipment (Equipment, splitter card/shelf, cabling, main distribution frame, protection panel) 3. The Outside Plant (OSP) (cross connect box, cable binders, cabling) 4. The Customer Premises Equipment (CPE) (network interface device, wiring, XDSLXDSL router, set top box, television)

Noise and Crosstalk Noise is any unwanted signal that adversely affects xDSL that, in turn, affects the subscriber's IPTV experience. If the noise level is high enough, the video stream can be disrupted for an extended period of time. At noise levels that are marginal, the xDSL rate can be reduced, resulting in pixelization and tiling. Noise is a general term. Noise can be in the form of Crosstalk, Electro-Magnetic Interference (EMI), and Radio Frequency Interference (RFI), to name a few. Crosstalk within binder groups and between binder groups adversely impacts xDSL performance. Repeatered T1 services that use AMI or B8ZS line coding typically cause the worst interference. HDSL, HDSL2, and ISDN services also cause unwanted crosstalk. EMI can be caused by proximity to power transmission lines and factories. Natural sources of EMI include lightning, sunspots and solar flares. EMI can affect the proper operation of equipment in addition to introducing unwanted noise onto the subscriber's line. With respect to RFI, your favorite AM radio station could be interfering with your xDSL signal as both technologies share a part of the same frequency band.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

67

Recommendation 18 - Noise Reduction in the OSP Reduction of noise and/or noise sources in the OSP is the only way to ensure a high quality video stream. Noise mitigation is perhaps the least understood method of improving circuit quality. OSP test equipment can assist service providers find the cause of poor video performance by measuring copper pairs in the OSP. To minimize the effects of crosstalk, keep all "like" services in the same binder group. For example, keep all xDSL services in the same binder group, keep all HDSL services in their own separate binder group, and so forth. Under no circumstances should you mix T1 (AMI/B8ZS) services in the same binder as xDSL services. EMI and RFI can be reduced by ensuring equipment and cabling in the outside plant is properly grounded and bonded. Replace corroded grounding and bonding straps in all RTU’s, Cabinets, Pedestals, and so forth.

Wire Gauge The use of heavier-gauge wire allows less line attenuation at any given distance, resulting in better xDSL performance. But because most xDSL deployments require the use of existing copper loop plant (already in place), changing the wire gauge used in the network may not be cost-effective. Recommendation 19 - OSP Cable Gauge For networks requiring new cable, if you can specify the wire gauge, use the heaviest available wire gauge that is cost effective. Commonly used wire gauges include 19, 22, 24, and 26 AWG. In cases where only a portion of the copper cable can be replaced, be aware that wire gauge transitions can sometimes cause reflections and negatively impact performance. Use of proper splicing methods and equipment will ensure potential reflections are minimized.

Splicing High quality splicing must be maintained. Any high resistance connections on a pair can severely impact the attainable DSL train rate and SNR margin.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

68

Drop Cables Parallel line wire drops have been widely used in the past. These should be replaced with shielded twisted pair. Drops should be shielded, grounded and bonded. Aerial drop cable may be susceptible to interference from such sources as AM radio, sunspots, and power lines. Buried drop cable may be subject to EMI radiation contamination, bridged taps, ground faults, and foreign voltages. Recommendation 20- Drop Cables Change out straight pair drop cables for twisted pair drop cable. The twist is there to minimize noise pickup.

Grounding and Ground Bonding Improper grounding can result in noise being generated between pairs instead of going to the ground sheath. Low resistance to ground at each splicing point to the cable is a must. Any high resistance or broken grounds will severely impact DSL speeds.

Load Coils DSL technology does not function with load coils. Originally designed to improve voice performance over long distances, load coils act as filters, effectively blocking the high frequency signals used by xDSL. Recommendation 21 - Load Coils If load coils exist on a subscriber pair, you must remove them prior to turn-up of xDSL service. Standard OSP test equipment can determine the number of load coils on a circuit. Use the TDR function to determine the distance to the first load coil.

Bridged Taps Unused, unterminated subscriber lines are called bridged taps, representing one of the most performance-affecting faults on an xDSL circuit. Bridged taps generally do not affect traditional POTS circuits, but do cause problems with high-frequency xDSL signals. The length of a bridged tap is referred to as a lateral, which is any length of cable not in the direct path between the CO and customer. A lateral creates a second path for the xDSL signal. The signal travels down the lateral and reflects back from the open end of the bridged tap, creating noise on the actual subscriber cable pair.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

69

Two key factors determine the effect of a bridged tap on xDSL performance: •

The length of the lateral: Shorter bridged taps often cause more harm than longer ones, because the reflected signal encounters less attenuation over a short distance and thus is more powerful.

•

The distance from the bridged tap to the modem on either end: Close proximity of the XDSL router at the central office or subscriber premises to the noise source causes more damage than does a distant source that has been attenuated.

Recommendation 22 - Bridged Taps Understand the impact of a bridged tap on signal loss, and ensure your field service force is well trained and equipped with tap finding tools such as a Time Domain Reflectometer (TDR) or Frequency Domain Reflectometer (FDR). Standard OSP test equipment typically offers both these measurements. Service personnel must check SNR margins at the NID, diagnose low or missing SNR margins, and determine whether the fault is due to a tap or some other inhibitor. When a bridged tap causes a fault, remove the bridged tap.

Longitudinal Balance Longitudinal balance is the electrical symmetry of the two wires within a pair. It indicates how well matched are the impedances of tip and ring conductors with respect to ground. The better a pair is balanced resistively and capacitively, the better it will reject unwanted noise. Poor line balance can result from a variety of line problems, including:        

Variations in pair twist ratios Split pairs Faulty CO equipment or protectors Faulty subscriber equipment or protectors Partial opens (series resistance) Dirty opens (combination resistive and capacitive faults) Grounds Faulted bridged taps (opens, partial opens, dirty opens, grounds)

Maximizing the longitudinal balance on the subscriber pair is one of the best ways to ensure reliable and robust video service delivery.

Proprietary Information: Not for use or disclosure except by written agreement with Calix. © Calix. All Rights Reserved.

70

Cross Connect Blocks Several different styles of cross connect blocks are available on the market today. Calix tested three common types of blocks to determine whether these different types had any impact on crosstalk isolation and performance improvement. The tested cross connect block styles follow:   

IDC/Screw-down with CAT5e wire and MS2 connectors URLS Tool-less IDC with CAT3 wire and MS2 connectors RLS Tool-less IDC with CAT3 wire and MS2 connectors

Recommendation 23 - Cross Connect Blocks For crosstalk and ADSL2+rate/reach testing, the effect of the three different cross connect blocks was not statistically significant, however BW requirements have changed over time and CAT5 blocks are being re-assessed for higher BW applications. Note: Testing to date has focused on CO infrastructure where the cross connect is located from 2kft to 4kft from the DSLAM. Cross connect testing for RT locations where the DSLAM is =60dB

Loop Current

>=20mA

Ground Resistance