Raising the Bar

Full Vectoring More Than Doubles Data Rates, Reach and Revenue Potential from Service Provider Networks

Next-Generation Networks Demand Full Not Partial Vectoring to Deliver the True Benefits of the Technology and Maximum Revenue from New Bandwidth-Intensive Services Low deployment costs and high potential data rates have allowed Fiber to the Node (FTTN) networks using traditional copper twisted pair for the last mile link to become the architecture of choice for delivering voice, video and advanced data services to a bandwidth hungry world. This topology allows for the rapid deployment of lucrative triple play services at a fraction of the cost of an all fiber network. However, broadband equipment vendors must overcome the challenge of crosstalk in order to utilize copper to its fullest. International Telecommunications Union (ITU) standard G.993.5 (G.vector) plays a critical role in cancelling crosstalk. And by doing so, the technology can drive throughput of 100 Mbps and beyond for next-generation FTTN networks. But to truly guarantee that performance level and fully utilize the bandwidth available, service providers must have systems capable of full vectoring, not the partial cancellation that has been discussed by some in the industry. Partial cancellation dramatically diminishes the value of vectoring and cuts network performance in half in many real-world scenarios. To find out more, read on.

Crosstalk in Cable Bundles The Critical Performance Challenge

VDSL (Very high speed Digital Subscriber Line) is the most relevant technology to deliver very high bandwidth (100Mbps and beyond) to users over copper lines that complete the last mile link from neighborhood network nodes. However, copper lines are subject to a variety of noise sources, both internal and external, which can significantly degrade system performance. Crosstalk is one type of radiated noise inherent to any copper conductor carrying high-speed data signals, including xDSL, T1, etc. It is injected via capacitive or inductive coupling into neighboring copper lines. In VDSL applications, far-end crosstalk, or FEXT, is a primary performance limiter and is generated from virtually every copper pair within any cable bundle. Its impact is especially high on shorter loops. This article describes how noise cancellation technologies in particular ITU-T standard G.vector using full cancellation or full vectoring eliminates this crosstalk and enables 100 Mbps performance in VDSL applications. In an FTTN environment, fiber and copper are utilized to provide high bandwidth services. Although FTTN cable bundles tend to be smaller in diameter than those from traditional central offices, a typical cable bundle often includes between 200 and 400 pairs. The large cable bundle (or binder) is comprised of multiple smaller bundles (see a typical 300 pair bundle in the illustration at right, consisting of twelve 25-pair binders). From the street cabinet, this large bundle is extended anywhere from 50

feet to 500 feet (15 to 200 meters) before sub-bundles fan out into the neighborhood. Therefore the crosstalk challenge must be addressed within the large bundle since binder management is typically prohibitively expensive and unrealistic, and crosstalk coupling is not limited to individual unshielded binders within the cable. In typical xDSL deployments, crosstalk between cable pairs inevitably drives down network performance. The closer in proximity the cable pairs are, the greater the potential crosstalk. This effect is magnified when looking at the inner binders, which can be affected by copper pairs on every side. The problem is further amplified by distance. The longer a copper pair is in proximity with another, the greater the impact of crosstalk across those pairs and the greater the performance degradation. The color-coded FEXT intensity map at right illustrates the measured crosstalk coupling for cable pairs spanning two binders. Testing indicates that the highest levels of crosstalk experienced by a cable pair generally originate from within its own binder, however crosstalk from nearby binders is also too large to ignore. Squares adjacent to the blue diagonal (i.e., lower-right corner to upper-left corner,) represent cable pairs within the same binder that are in physical proximity with one another. As shown, there is a greater concentration of orange and red squares along this diagonal, indicating increased levels of crosstalk due to these adjacent/neighboring disturbers. Squares located in the upper-right corner and lower-left corner represent cable pairs located in a different cable binder and are (relatively) physically separated from one another. As expected, there is a greater concentration of yellow and green squares, indicating lower measured crosstalk levels due to these remote disturbers. Although crosstalk coupling is typically lower between binders, coupling can still be higher in some situations as represented by the yellow boxes in the upper-right and lower-left corners of the illustration. Since binders are not usually insulated, crosstalk coupling between physically adjacent cable pairs spanning two binders is common. In addition, it is common for service provider personnel to reorganize
Less Crosstalk More Crosstalk

one or more cables in the field at splice points, thus introducing additional randomization of the FEXT intensity map for the system. In summary, it is extremely difficult to accurately predict which cable pairs will interfere with each other, and to be effective, crosstalk cancellation techniques must take the entire cable into account, including any splice points that may have been introduced by the service providers personnel in the field.

Challenge of Full Vectoring vs. Partial Cancellation

Crosstalk experienced by each cable pair originates from all other cable pairs within the bundle. The challenge of cancelling all disturbers in a cable bundle becomes overwhelmingly complex when the dynamic nature of crosstalk is combined with the very high-speed data rates in the VDSL physical layer. For example, consider a typical service provider network node of 384 ports or subscriber lines. Such a system presents the challenge of a 384 x 384 cancellation matrix across up to 4096 tones for each VDSL line over a 17 MHz or 30 MHz spectrum. Without special techniques, full cancellation across this node would require Gigabytes of high-speed memory and very high implementation complexity to achieve the benefits of full cancellation. One option being explored by some vendors is to identify the top disturbers to be cancelled (e.g., the top 25%). Although this approach does reduce the system overhead required for disturber cancellation, it also increases the complexity of the logic required in order to dynamically identify and track the top disturbers within a bundle. While this partial cancellation is less complex, its also much less effective. System level costs are similar between the two implementations, if done in ASICs, but the benefits of vectoring are dramatically reduced with partial vectoring. Performance benefits for a partial cancellation system can be <50% of what is achievable with full vectoring. Detailed simulations illustrate this in great detail.

In Ikanos tests, a 300 pair 1000 foot (300 meter) cable bundle was used to illustrate the point (see illustration above). To emulate a larger vectoring system, a 32 port prototype was used with 24 ports placed in a victim binder and eight ports were used as roving disturbers to generate crosstalk from each adjacent bundle. All crosstalk into the victim binder was recorded and processed offline to project achievable data rates for various scenarios.

Guaranteed Performance vs. Lowest Common Denominator

Ikanos ran detailed analysis using the 300 pair cable for both a partial cancellation scenario and a full vectoring implementation. For partial cancellation, the top 96 disturbers across each frequency band were identified and actively cancelled. In the case of full vectoring, all disturbers across all bands were identified and actively cancelled. These tests were performed for both downstream and upstream directions and data rates were gathered and compared. The chart at right illustrates the wide disparity in downstream data rates between the partially cancelled system and the system employing full vectoring. For the 1000 foot (300 meter) cable it was possible for the full vectoring system to deliver consistent performance well over the 100Mbps level (the target performance for a 500-700 meter cable). In fact, all lines were able to support consistently high performance essentially a guaranteed level of throughput approaching 140Mbps. In contrast, the performance of the partially cancelled system varied widely from line to line with the low being approximately 70Mbps or roughly half that of the fully cancelled system. Only 40% of the lines were able to achieve performance approaching that of the system using full vectoring. Because of this wide variation in performance on a line-by- line basis, the guaranteed service level for subscribers serviced by this node would drop back to the lowest common denominator rate, in this case is less than 70Mbps.

The same scenario plays out when comparing upstream data rates. Over time, given the move to cloud computing and other interactive applications, upstream data rates will become increasingly important to consumers and demand will approach 50Mbps and above, compared to sub-10Mbps today. Again, the full vectoring system delivered consistent performance across all lines up to 60Mbps. Similar to the downstream scenario, the partially cancelled systems performance varied widely. Less than 10% of the lines achieve data rates approaching that of the full vectoring system, far worse than in the downstream case. And the actual guaranteed upstream data rate to the subscriber would be closer to 25Mbps approximately half the guaranteed data rate achieved by the full vectoring system.

Full Vectoring Delivers Dramatic Deployment Cost Reduction Eliminates Need to Condition as Much as 85% of Subscriber Lines
Line conditioning is one of the most costly components of a VDSL deployment. Service providers can spend as much as $100 or more for each subscriber line to identify and eliminate grounding problems, bridge taps and other issues that contribute to crosstalk and negatively impact performance. Without full vectoring, most service providers today opt to proactively condition all lines as a standard operating procedure when installing new network nodes for a cost per node approaching $40,000.

Full vectoring, with its ability to cancel all crosstalk, recovers the performance potential of these lines without the need for line conditioning across the entire node. Also, once installed, the full vectoring system can be used to pinpoint faulty lines that require conditioning, allowing the service provider to perform reactive conditioning on only those lines. The chart above shows the cost of proactive conditioning vs. reactive conditioning made possible by full vectoring across a network node of 384 ports. In the case of proactive conditioning, 100% of the lines must be conditioned at a cost of $100 per line or $38,400. With full vectoring, if only a few lines (less than 60 out of the 384, for instance) require conditioning, then the total cost of conditioning would be just $5,800 thats an 85% savings over the nearly $40,000 cost per node of proactive conditioning.

Full Vectoring Drives Double the Revenue Potential of Partial Cancellation

Vectoring has the potential to deliver the performance of full fiber-to-the-home networks up to 100Mbps data rates and beyond at a fraction of the cost of implementing a network with FTTH architectures. Industry estimates for passing and connecting a home via FTTH can run above $2500 per residence while a fiber-to-the-node architecture using vectored VDSL can cost only a few hundred dollars per residence. Ikanos estimates that the cost of implementing a FTTN networking using vectoring either full vectoring or only partial cancellation will be roughly equivalent, if the solution is realized by ASIC implementation. However, there is a massive disparity between the two implementations when comparing a service providers ability to deploy high-speed services and extract value and revenue from their investment. As Ikanos tests demonstrate, full vectoring systems can consistently deliver guaranteed service levels of 100Mbps downstream and 50Mbps upstream over loop length more than 500 meters. Our estimate for partially cancelled systems over the same loop length is 50Mbps and 25Mbps respectively. With the higher guaranteed data rates of a full vectoring system, service providers can offer a wider range of voice, video and data products at higher prices than those employing only partial cancellation. Average revenue per user (ARPU) for the higher levels of service made possible by full vectoring are estimated to be $125 per month vs. $70 per month for partial cancellation, or approximately a 50% premium for the higher speed fully vectored offerings. In addition, the all-important take rates the percentage users passed vs. those that actually purchase these new services will be greater since those made possible by full vectoring are more competitive with cable MSO offerings and those provided by carriers using FTTH architectures.

The chart at right illustrates the three year revenue potential of full vectoring system vs. one that employs partial cancellation across a customer base of 2.8 million households. In this scenario, a service provider using full vectoring in their network is able to offer two primary tiers of service: 100/50 downstream/upstream data rates and a 50/25 downstream/upstream product. Using partial cancellation, the two tiers of service would be a maximum of 50/20 downstream/upstream and 25/3 downstream/upstream because of unpredictable performance. Given the higher data rates of full vectoring, we expect the service provider to achieve take rates in year one of 10% increasing to 20% over the three year period. Service providers using only partial cancellation will have limited take rates of approximately 10% per year, and potentially declining given the lack of competitiveness in their offerings. Over the three year period using the scenario just described, a service provider using full vectoring could achieve average annual revenue over $500 million, or double that of the service provider using only partial cancellation. In addition, the service provider using full vectoring will have achieved complete payback of their network investment while its counterpart using partial cancellation will be looking at many more years before achieving payback. In other words, the economics of full vectoring vs. partial cancellation are clear if you are looking to maximize the value of your network investment, full vectoring provides the most capable solution.

Ikanos NodeScale Vectoring

Ikanos provides a unique solution that extracts the maximum benefits of vectoring by achieving full vectoring across an entire node. Historically, to achieve full vectoring across a node, vast amounts of memory and excessive interface bandwidth were required. However, Ikanos smart compression technology eliminates this bottleneck by achieving near lossless compression through implementation of full Nodescale Vectoring in scalable ASIC silicon to satisfy market deployment scenarios with little or no added complexity.

 2011 Ikanos Communications, Inc. All Rights Reserved. Ikanos Communications, Ikanos, the Ikanos logo, the Bandwidth without boundaries tagline, Fusiv, Fx, FxS, iQV and Ikanos Accelity, Ikanos Capri, Ikanos ISOS, Ikanos Maxtane, Ikanos NodeScale, Ikanos Orion, Ikanos Solos, Ikanos Velocity, Ikanos Vulcan are among the trademarks or registered trademarks of Ikanos Communications.

All other trademarks mentioned herein are properties of their respective holders. This information is protected by copyright and distributed under licenses restricting, without limitation, its use, reproduction, copying, distribution, and de-compilation. No part of this information may be reproduced in any form by any means electronic, mechanical, magnetic, optical, manual, or otherwise, without prior written authorization of an authorized officer of Ikanos Communications, Inc (Ikanos).

Disclaimer
This information is furnished for informational use only, is subject to change without notice, and should not be construed as a commitment by Ikanos. Ikanos assumes no responsibility or liability for any errors or inaccuracies that may appear in this material. Ikanos makes no representations or warranties with respect to the design and documentation herein described and especially disclaims any implied warranties of merchantability or fitness for any particular purpose. References in this document to an industry or technology standard should not be interpreted as a warranty that the product or feature described complies with all aspects of that standard. In addition, standards compliance and the availability of certain features will vary according to software release version. For further information regarding the standards compliance of a particular software release, and the features included in that release, refer to the release notes for that product. Ikanos reserves the right to revise the design and associated documentation and to make changes from time to time in the content of this document without obligation of Ikanos to notify any person of such revisions or changes. Use of this document does not convey or imply any license under patent or other rights. Ikanos does not authorize the use of its products in life-support systems where a malfunction or failure may result in injury to the user. A manufacturer that uses Ikanos products in life-support applications assumes all the risks of doing so and indemnifies Ikanos against all charges.

For more information, contact Ikanos. Ikanos Communications, Inc. 47669 Fremont Boulevard Fremont, California 94538 www.ikanos.com
P

+1 510.979.0400 +1 510.979.0500