Download - 3. DOCSIS 3.1

© 2010 Cisco and/or its affiliates. All rights reserved.

DOCSIS 3.1An evolutionary approach to Gigabit Class DOCSISCABU EMEAR Marketing

Nov. 14th 2012

© 2011 Cisco and/or its affiliates. All rights reserved. Cisco Confidential 2

• Introduction

Advanced* DOCSIS

• Case Study

Advanced* DOCSIS vs EPOC

* Standardization is underway at CableLabs under DOCSIS 3.1 reference


• This year DOCSIS is 15 years old

DOCSIS 1.0 I01 Spec released March 26, 1997.

• DOCSIS 3.1 “An evolutionary approach to Gigabit-class DOCSIS”

For an era where MSO’s will continue to push fiber deeper, potentially/eventually switch to all fiber for new build, but will do everything to maximize the return on earlier investments in coax plant.

“How the next 15 years will looklike!”

* Standardization is underway at CableLabs


• More services requires more IP bandwidth.

• How much bandwidth? What is really possible?

Data as an

IP Service

Voice as an

IP Service

Video as an

IP Service


Example:

• 300 video subs per SG,� multicast for linear, � unicast for VoD, � 50% HD, 50% SD,� VBR, MPEG4,� 20 DOCSIS channels

• 150 channels collapsing to 20 – That is efficient!

Source: “HFC Capacity Planning for IP Video” by Sangeeta Ramakrishnan, SCTE Expo 2011

0

5

10

15

20

25

30

35

100 150 200 250 300 350N

um

ber

of

DS

ch

an

nels

req

uir

edNumber of IP video subs

per Service Group

CBR

VBR

• Video bandwidth will expand as new 4K and 8K formats are adopted.

• HSD will continue to grow and eventually may exceed SP video BW.


• Goals

Allow DOCSIS over HFC to compete with FTTH solutions.

Achieve 5+ Gbps in the downstream.

Achieve 1+ Gbps in the upstream

Backward compatibility story with DOCSIS 3.0, 2.0, & 1.1.

Better spectral efficiency.

• Technology

OFDM and LDPC

Re-use SCDMA MAC concepts

* Standardization is underway at CableLabs


• EPOC is EPON over Coax.

• EPOC and D3.1 will use the same PHY.

• DOCSIS (as it evolves) and EPOC are similar technologies but in different markets.


• , , and teamed together to help define and drive DOCSIS 3.1.

• The first output of this joint effort was a landmark white paper at NCTA 2012, both in terms of size and in terms of collaboration.

182 pages

83 Figures

43 Tables

10 recommendations

7 areas of further study

1. Introduction

2. Cable Spectrum Analysis

3. Solving Legacy Issues

4. Coax Network Analysis

5. HFC Optical Transport Options

6. HFC Topology

7. DOCSIS PHY

(ATDMA, SCDMA, OFDM)

8. DOCSIS MAC

9. Network Capacity Analysis

10.Network Capacity Migration

11.Recommendations

John T.

Chapman


• Evolution of the “Upstream/Downstream” Split

To get to a more Symmetrical Service Proposition (more in line with ETTH or xPON architecture)

• Extending the Frequency Spectrum

In the first place to get maximum use on coax and passives in place (and found to “easily” support approx. 1.2 Ghz so 20% more than currently spec’d for us)

• MAC & PHY Layer

To further Increase the Spectral Efficiency

• Maintaining Backwards Compatibility

Avoid forklift upgrades and provide investment protection for earlier D1.1 or D2.0 CM, QAMs, Q.


� DOCSIS 3.1 will introduce OFDM with LDPC.

� The target modulation is 1024-QAM.

� Up to 4K QAM will be specified

� Using the spectrum above current plant cut-offs(750 MHz to 1 GHz) requires an advanced

PHY to support more complex modulation such as OFDM.

� OFDM will also be used below 1 GHz, and will likely supplant legacy QAM bandwidth over time.

� The initial push is to 1150 MHz. This may be possible without replacing taps. The long term push is 1.7 GHz, but this would require tap replacement.

� The DOCSIS 3.1 downstream deployment may occur before DOCSIS 3.1 upstream deployment.

Improved FEC (Forward

Error Correction) enables

higher modulation at the

same CNR performanceQ

RF Spectrum Efficiency

Improvement:

~ 25% with same CNR(8 – 10 bits / symbol, 1024QAM

Equivalent)

~ 50% with improved CNR(8 – 12 bits / symbol, 4096QAM

Equivalent)


� DOCSIS 3.1 upstream will use OFDMA with an LDPC FEC

� Target modulation is 256-QAM.

� Up to 4K will be spec’ed.

� Existing spectrum will be shared between ATDMA/SCDMA and OFDM. New spectrum will be OFDM only.

� Frequency split options:

� The immediate goal is to maximize sub-split. (42/65 MHz)

� The short-term recommendation is mid-split. (85 MHz)

� The long-term recommendation is high-split. (~230 MHz)

� Mid-split triples upstream throughput and is available today with D3.0.

� DOCSIS 3.1 high-split provides 1 Gbps.

Improved FEC (Forward

Error Correction) enables

higher modulation at the

same CNR performanceQ

RF Spectrum Efficiency

Improvement:

~ 33% with same CNR(256QAM Equivalent)

~ 66% with improved CNR(1024QAM Equivalent)


DOCSIS 3.0 DOCSIS 3.1

Now Phase 1 Phase 2 Phase 3

DS Range (MHz) 54 - 1002 108 - 1002 300 - 1152 500 - 1700

DS QAM Level 256 256 ≥ 1024* ≥ 1024*

# DS Channels 8 24 “142” “200”

DS Capacity (bps) 300M 1G 7G 10G

US Range (MHz) 5 - 42 5 - 85 5 - 230 5 - 400

US QAM Level 64 64 ≥ 256* ≥ 1024*

# US Channels 4 12 “33” “60”

US Capacity (bps) 100M 300M 1G 2.5G

Note: TBD values are underlined, Channels in quotes = Equivalent # of SC-QAMs * On OFDM PHY


• As extra measures to increase flexibility as well as system throughput DOCSIS AMP is likely to bring following news related to MAC & PHY:

• US PHY: Additional OFDMA* with new LDPC FEC**

• DS PHY: Additional OFDM with new LDPC FEC**

• Order of modulation: With a plan to investigate 4k QAM for OFDM, see table on previous slide that new standard targets supporting at least 1kQAM in both US and DS

• MAC: Likely for current SCDMA MAC functionality to be the basis for the new OFDMA MAC Layer

• Remark: Goal as set out for throughput is an increase of min. 50% in bits/Hz

* Orthogonal Frequency Division Multiple Access

** Low Density Parity Check Forward Error Correction providing the equivalent of 5 to 6 dB performance improvement,

in other words a system that runs today 64-QAM could run 256-QAM with new LDPC FEC


• DOCSIS 3.0 uses single carrier QAM (SC-QAM) in the downstream and upstream.

Two sine waves, I and Q, each with separate amplitude and phase are added together to create symbol within a constellation.

• Each instance is referred to as a symbol.

16-QAM is 4 bits per symbol





Example: 16-QAM


• OFDM is identical to coded OFDM (COFDM) and discrete multi-tone modulation (DMT)

• A frequency-division multiplexing scheme used as a digital multi-carrier modulation method using a large collection of very narrow QAM subcarriers.


• FEC = Forward Error Correction

FEC adds redundant bits so that errored bits can be re-created.

FEC requires an interleaver in order to be truly effective.

• LDPC = Low Density Parity Check

Invented by Robert Gallager in 1960.

Could not be implemented in HW until recently.

• LDPC is much more robust than Reed-Solomon.


� 2 dB for FEC improvementsQ Reed Solomon versus LDPC

� 1 dB for OFDM filter edge improvements versus Single Channel QAM

� 1 dB for lowest common denominator for 12 KHz channel versus 6 MHz channel width

� Total = 4 dB equivalent improvement

� Remark: Assuming that most plants have 2 dB headroom for existing modulation depthQ there might be 2 modulation depth increases possible Q

Resulting in 25-33% increase bits/Hz depending current DS/US constellation


Modulation Bits/Sym

MER

SCQAM

MER OFDM w/

LDPC

Approximate Data

rate for 6 MHz

Channel Mbps

RF Spectrum

Required to support

1 Gbps (MHz)

QPSK 2 15 12 9 667

8QAM 3 18 15 14 444

16QAM 4 21 18 18 333

32QAM 5 24 21 24 255

64QAM 6 27 24 28 212

128QAM 7 30 27 33 182

256QAM 8 33 30 38 159

512QAM 9 36 33 42 141

1024QAM 10 39 36 49 122

2048QAM 11 42 39 54 111

4096QAM 12 45 42 59 102

8192QAM 13 48 45 64 94

16384QAM 14 51 48 69 87

Docsis 3.0 standard for Upstream

Docsis 3.0 standard for Downstream

Docsis 3.1 proposed standard for both Upstream and Downstream

Data Rates are +/- 10% for differences in

overhead Upstream vs. Downstream and Reed

Solomon vs. LPDC

Single Channel

QAM RS UpstreamOFDM LDPC Upstream

< 35% improvementSingle Channel QAM RS

Downstream

OFDM LDPC

Downstream 28%

Improvement


• The CNR can vary by at least 8 dB on a good plant.

Equivalent to ~3 orders of modulation

• D3.1 will sort CMs into different profiles

MCS = Modulation and Coding Scheme

Not one MCS per CM. No unicast.

• 4 profiles should suffice

A: Best Case (e.g. 4096-QAM)

B: Better Case (e.g. 2048-QAM)

C: Good Case (e.g. 1024-QAM)

D: Common channel (e.g. 256-QAM)

Worst

Case

Average

Case

Best

Case


ForwardingEngine

✓Tags packet to both L3 and L2 queues

L3/L2Queuing

CL Buffer

Channel A

CL BufferChannel B

CL BufferChannel C

CL BufferChannel D

MuxBurst

Builder

✓Builds sequential

bursts

✓FEC

✓MCS

✓MAC Domain

✓QoS

✓HFQ

✓Rate Shaping

✓Service Flows

iFFT

CL PHYControlChannel

MessageBlocks,

PreambleCPU

✓Generates MMM ✓Convergence Layer - Framing, Mapping


A B B C A AD

A B B C AD

= sub-carrier group

for one symbol time

= FEC block begin/end

A = DS Channel with a

unique OFDM profile

f

t

A• FEC blocks are mapped to sub-

carrier groups (N sub-carrier for one symbol time) vertically (in the frequency domain, one symbol wide) within a FFT block.

• This in effect creates a serial byte stream.

• Multiplexing of profiles is done within this byte stream.

• It is okay if the FEC block crosses a FFT block boundary in consecutive (time-domain) FFT blocks.


OFDM D/S PHY

channel

Future OFDM

D/S bandsOFDM D/S PHY

channel

Bonded Group

D/S

Frequency

Downstream Band

Le

ga

cy

Le

ga

cy

Le

ga

cy

Slightly

Modified

New

LegacyG

ua

rd B

an

d

750MHz 1000MHz

Le

ga

cy

Le

ga

cy

Le

ga

cy

5 Gbps > 7 Gbps > 10+ Gbps

Future OFDM

D/S bands

Le

ga

cy

42/54

MHz85/105

MHz230/300

MHz

• Analog

Reclamation

• Node Splits

• Improved bits/Hz

(OFDM)

• Broadcast reduction


OFDMA

U/S PHY

channel

OFDMA

U/S PHY

channel

OFDMA

U/S PHY

channel Le

ga

cy

Bonded Group

PHY

MAC

Le

ga

cy

U/SUpper

Layers

Upstream Band

Gu

ard

Ba

nd

Le

ga

cy

Typical 42 MHz US Plant at 64QAM is

limited to 4 channels at about 100 Mbps

With OFDMA the capability could be

stretched to ~ 150 Mbps (250 Mbps with no

QAM)

Upstream spectrum upgrade to

85 MHz:

• Move/remap 5 DS channels

• 300 Mbps w/ 64QAM

• 500 Mbps w/ OFDMA

42/54

MHz85/105

MHz

230/300

MHz

Upstream spectrum upgrade to 230 MHz

• Move / remap channels 2 – 36

• Aggregate data rate 1.2 – 1.7 Gbps depending on CNR

Bonded Group

100 Mbps > 150 Mbps > 500 Mbps > 1200 Mbps


OFDM D/S PHY

channel

Future OFDM

D/S bands

OFDMA

U/S PHY

channel Le

ga

cy

OFDM D/S PHY

channel

Bonded GroupBonded Group

PHY

MAC

Le

ga

cy

U/SD/S

Upper

Layers

Frequency

Upstream Band Downstream Band

Le

ga

cy

Le

ga

cy

Le

ga

cy

Slightly

Modified

New

Legacy

Gu

ard

Ba

nd

750MHz 1000MHz

Le

ga

cy


• Both OFDM and OFDMA work by separating a single signal into subcarriers, or, in other words, by dividing one extremely fast signal into numerous slow signals that optimize mobile access, as the subchannels can then transmit data without being subject to the same intensity of multipath distortion faced by single carrier transmission. The numerous subcarriers are then collected at the receiver and recombined to form one high speed transmission.

• The difference between OFDM and OFDMA is that OFDMA has the ability to dynamically assign a subset of those subcarriers to individual users, making this the multi-user version of OFDM, using either Time Division Multiple Access (TDMA) (separate time frames) or Frequency Division Multiple Access (FDMA) (separate channels) for multiple users. OFDMA simultaneously supports multiple users by assigning them specific sub-channels for intervals of time.


• The OFDM MAC will be based upon the SCDMA MAC which is similar to the ATDMA MAC.

Minislot = X sub-carriers forY symbol times.

• All three MACs use mini-slots with upstream scheduling.

ATDMA: minislots map to time

SCDMA: minislots map to time and a group of codes

OFDMA: minislots map to time and a group of tones


� OFDMA Convergence layer is capable of multiplexing ATDMA, SCDMA, and OFDMA PHYs.


• Upstream

ODFM and ATDMA/SCDMA can share the same spectrum

Bonding between OFDMA and ATDMA/SCDMA is possible

• Downstream

Bonding between OFDM and SC-QAM is supported.

• This allows a gradual and evolutionary introduction of DOCSIS 3.1.

This is a distinct competitive advantage that DOCSIS has over other non-DOCSIS solutions such as EPOC.

• DOCSIS 3.0 will get capped.

The target cap is 16x4 or 24x8.


• DOCSIS 3.1 plans to state:

D3.1 CM MUST support SCDMA.

D3.1 CMTS MAY support SCDMA.

• It is generally agreed that OFDMA with LDPC will be able to replace the role that SCDMA and ATDMA perform today.

• Thus, support for SCDMA is for legacy D3.0 and below CMs.

• Long term use of SCDMA really depends upon if and how much of SCDMA gets deployed prior to D3.1 being available.


Time

Up

Do

wn

DOCSIS 1.0 - 3.0

DOCSIS 1.0 - 3.0

(HSD, VoIP, & IP Video)DOCSIS 3.1 OFDM/LDPC

(HSD @ PON Speed,

Video over IP, Ultra HD, &

un-discovered apps)Legacy

Video EQAM

(Digital Video)

DOCSIS 3.1 OFDMA/LDPC

3.0 CMTS

& EQAM3.1 CCAP3.0 CCAP

Phase 2

Phase 3

A: Initially run D3.1 CMs in D3.0 mode (avoiding RF Data Simulcasting Tax)

C: Increase D3.1 CM count in SG. Enable some DS OFDM channels & bond with legacy D3.0

D: Use existing passives with OFDM & 1.2+ GHz electronics as required

E: We Could End Up With One Advanced PHY (OFDM/LDPC)

For The Entire Spectrum as our Target Architecture

A

B

C

D

E

B: For US, enable OFDMA and perform channel bonding with legacy D3.0

Phase 1Now

* from DOCSIS 3.1 Deployment Scenarios “Joint Suppliers” Presentation

hosted by Cisco, Moto, Intel and Harris July 12th 2012


• The legacy migration concerns with mid-split and high-split such as analog TV, RF interference, ADI and OOB, have workable solutions.

• Analog TV can be reduced, removed, or remapped.

• Interference with specific OTA signals can be managed by attenuating specific OFDM tones.

• ADI = Adjacent Device Interference

HPF needed on coax in same house as mid/high-split HGW

Adjacent home should be okay if coax design is good.

• OOB can be replaced by DSG on most devices

• A Legacy Mitigation Device (LMA) can be used to fix OOB and ADI concerns if and when they occur.


• DOCSIS is defined by:

market requirements,

the HFC environment,

available technology, and

the will and creativity of the DOCSIS community.

• DOCSIS is the most successful Ethernet over Coax technology to date.

• DOCSIS can be anything the DOCSIS community wants or needs it to be.

• DOCSIS 3.1 is intended to scale the delivery of all IP services over the HFC plant and do so in a manner that is competitive with FTTH or any other broadband technologies.


✔✔✔✔ Date Milestone

✔ 2012 -

2012

AMP exploratory committee at CableLabs to

determine technology options.

✔ 2012-07 MSO CTO Meeting to determine D3.1 direction

D3.1 Committee has its first meeting

2013-02 PHY Spec W01 – Downstream only

2013-03 MAC Spec W01 – Downstream only

TBD MAC and PHY Spec W02 – Upstream included

2014 CM Silicon available. System integration and

test.

2015 DOCSIS 3.1 CM Product Availability

2015+ DOCSIS 3.1 CMTS Product Availability

� NOTE: Final vendor schedules may differ.


The Key Takeaways of this presentation were:

• DOCSIS 3.1 will scale to 10 Gbps x 1 Gbps

• DOCSIS 3.0 can do 1 Gbps in 2013

• Cisco is helping to lead this effort.

• PHY layer is OFDM and LDPC


� IEEE standardization in process (as compared to Docsis 3.1 standardization in process through

� Recent agreement that EPON and Docsis 3.1 will utilize the same PHYQ OFDM with LDPC

Currently EPoC yields greater spectral efficiency when compared to DOCSIS 3.0. This is not the case when evolving to DOCSIS 3.1

� MAC and PHY for EPOC will be isolated at the CMC locationQ no QAM sharing across service groups.

Unlike EPoC, DOCSIS QAM’s can be shared across multiple service groups

� EPOC will not be backwards compatible to SC QAM Docsis

The current investment in DOCSIS QAM’s can be leveraged as we evolve to DOCSIS 3.1

� Requires dedicated channels and RF spectrum for each technology (Docsis and EPOC)


�The DOCSIS 3.0 and EPoC analysis is focused in four primary areas

1. The similarities and differences of coexistence and backwards compatibility with DOCSIS 3.1

2. Spectrum efficiency of each technology based on assumed bandwidth requirement

3. Physical topology of the fiber and coaxial infrastructure

4. CAPEX estimates based on the assumed evolution of technology and future price declination


Optical Line

Terminal

Chassis

Legacy

Transmitter

/ Reciever

1310 or 1550nm

EPON OLT

Optical

Network

Unit

Coaxial

Media

Converter

CAT-5DS US

EPOC RF

Fiber Node

Op

cti

cal S

plitt

er

Op

cti

cal S

plitt

er

Node 1

Node 2

Node 3

Node N

Fiber Network• Fiber plant is a parallel network using

standard EPON equipmentQ does not

necessarily require an additional fiber

(does require some wavelength planning)

• If DPoE is utilized then EPOC can share

CMTS chassis and use common

provisioning tools.

Coax Network• Coax Network requires outside plant

changes to insert RF signal at the Node

and make room for RF signals.

• Coax network must share RF spectrum

with HFC and current services

• Multiple possible RF spectrum solutions,

(i.e. Top Split, High Split)

Coax signals must coexist

with legacy HFC signalsQ

multiple possible solutions

Gateway /

STT

Cable

Modem

EPOC

Coax

Network

Unit

Home

CMTS7600

Router


8 Wavelength DWDM with dual fiber

DOCSIS+EPOC RF

Optical Line Terminal

Chassis(Serves up to 16 nodes)

(requires new RF combining modules in existing node)

Serves up to

32 Fiber Node’s

Hub Outside Plant

Downstream fiber

Upstream fiber

1x2OLT

PORT 1

OLT

PORT 21X2

1x2

FN’s 1-8

FN’s 9-16

1x8 US DWDM + DS Ethernet

Expansion

Ports

1310/1490nm

ONU/CMC

EPOC RF

750 – 1.125 GHz

Fiber Node

New

1550nm

Transmitter 1x8 D

S M

ux

1490nm

US

1310nm DS

1x8

US

Deux

1x8 U

S M

ux

1x8 D

S D

em

ux

1490nm

US

1310nm DS

Expansion

Ports

Each node is

segmented up to 8x

(~128 hhp)

1x8 DS DWDM + US Ethernet

� Initially eight fiber nodes are provisioned per 10G OLT

� Provisioning EPoC still allows the use of all 155x nm wavelengths

� Each “set” of 16 nodes requires an additional 10G OLT


� The first and second node segmentations occur in 2015 and 2019

� Node segmentation occurs prior to reaching the 1 Gbps threshold

� 1 Gbps per downstream service group determines the timing for node segmentation

� The timing of downstream node segmentation causes under-utilization of the upstream spectrum

� Increasing the downstream threshold would better utilize spectrum in both directions

1000 hhp 250 hhp 125 hhp

1000 hhp 250 hhp 125 hhp


� The initial EPoC deployment requires a significantly greater amount of Capital investment

The DOCSIS deployment is able to leverage the existing infrastructure

The NPV calculation highlights the value of the EPoC investment


� Since EPOC and DOCSIS 3.1 will use the same PHY, there will be no difference in RF Spectrum efficiency between DOCSIS 3.1 and EPOC

� EPoC is not backwards compatible with DOCSIS and therefore cannot bond with SC QAM (Single Channel QAM)

� The coexistence of EPoC and DOCSIS requires segregated RF spectrum for both technologies

� When provisioning 1 GHz EPoC, none of the previous investment in DOCSIS QAM’s can be leveraged

� The initial investment in EPoC is much higher than scaling DOCSIS 3.0 or the evolution to DOCSIS 3.1

� Although the total CAPEX for DOCSIS 3.0 and EPoC is similar, the evolution to DOCSIS 3.1 and the value of Capital over time indicates DOCSIS 3.1 is a wise investment

Thank you.Thank you.


• November 15th 2012

• Review with SCTE materials on D3.1 and D3.1 vs EPOC Case Study

• September 5th 2012

• Flagged “INTERNAL ONLY”

• August 30th 2012:

• Slide 12: Added target to the increase in spectral efficiency

• Slide 13: added remark related to use of earlier D3.0 DS LC

• Slide 16: Inserted first targets around timing

• August 24th 2012:

• made correrctions and added comments to slide 9 and 10.

• Added source reference for figure on slide 14

• August 22nd 2012: Initial post

Download - 3. DOCSIS 3.1

Top Related