Boosting VDSL2 Bit Rates with Vectoring
By providing cutting-edge noise-cancellation, vectoring enables VDSL2 lines to approximate their theoretical maximum speed in real-world conditions — offering cost-effective high bandwidth and very fast build-out. Because widespread installation of fiber is a lengthy process, service providers need additional technology options for delivering high bandwidth over the next 5 to 10 years. Very-high-speed Digital Subscriber Line 2 (VDSL2) vectoring offers one powerful, cost-effective way to satisfy individual subscriber demands and meet government goals for universal broadband access. It takes full advantage of existing resources by making conditions in the field as close to ideal as possible, so each VDSL2 line can operate at its best — and deliver higher bit rates. Downstream speeds of 100 Mb/s can be achieved at distances of up to 400 m, and 40 Mb/s can be supported with loops as long as 1000 m. In field trials since 2010 with leading service providers including Belgacom, A1 Telekom Austria, Swisscom, Orange, P&T Luxemburg and Türk Telekom, vectoring improved previous downstream bit rates by 90% to 150%. Vectoring enables these gains by canceling interference between copper lines, which is one of the most significant factors limiting the achievable bit rate. In a dynamic process, vectoring continually measures and cancels this “crosstalk,” so that all lines can operate at much higher capacity, as shown in Figure 1.
Figure 1: Typical vectoring gains
Because sophisticated noise cancellation is Central Processing Unit (CPU)-intensive, vectoring works best for the smaller number of lines (few hundred) typically found in Fiber to the x (FTTx) deployments — and where measurements are available from all lines. This means that the lines all need to be under full control of a single service provider for optimal performance gains. In these scenarios, vectoring technology allows service providers to deliver significantly higher speeds, with faster time to market and faster return on investment.
How vectoring works
Vectoring is not a method for raising the theoretical maximum transport speeds. Instead, this noise-cancellation technology addresses the gap between the theoretical maximum rate and the speeds that service providers can deliver in typical field conditions. In most deployments, telephone lines that carry VDSL2 signals are part of cables (sometimes partitioned in smaller cable bundles) that contain 10 to a few hundred lines positioned very closely together. This close proximity results in crosstalk, and the higher the number of lines in a cable (bundle), the more crosstalk is generated. Crosstalk is the main reason why lines in the field perform significantly lower than their theoretical maximum. Vectoring enables each line to perform as if it is alone, that is, without crosstalk. In concept, vectoring is comparable to the noise-cancelling technology used in headphones. It produces a clean signal for each line by measuring the crosstalk from all other lines and generating anti-phase signals to cancel the crosstalk signals out, resulting in almost zero noise. This concept sounds simple, but its execution can be highly complex, depending on the environment in which it is deployed. The noise cancellation technology in headphones, for example, has few variables to address as it suppresses interference in the audio band (20 kHz). Vectoring, on the other hand, often deals with the crosstalk generated by a few hundred VDSL2 lines and signals that span a wide frequency band (e.g. 17.6 MHz for 17a VDSL2).
The calculations needed for vectoring require enormous computing power, because vectoring dynamically measures and cancels interference across hundreds of VDSL2 lines over the full frequency spectrum occupied by the VDSL2 signals. The processing occurs in the frequency domain by subdividing the spectrum in up to 4096 narrow frequency bands – known as ‘tones’ or ‘carriers’ – and processing each tone independently. The tones are represented as ‘complex’ numbers with a ‘real’ and ‘imaginary’ part. The calculations are repeated per VDSL2 symbol whose duration is 250μs (so 4000 symbols per second). All VDSL2 lines are processed simultaneously and the results are used in real time to develop anti phase compensation signals for each line, based on the actual signals transmitted on the other lines. The frequency dependent crosstalk coupling between each pair of VDSL2 lines is automatically and dynamically measured and updated. For example, to cancel the interference in a cable with 200 active VDSL2 lines, 200 (VDSL2 ‘victim’ lines) x 200 (VDSL2 ‘disturber’ lines) x 4096 (tones) x 4 (real MAC operations per multiplication of 2 complex numbers) x 4000 (VDSL2 symbols per second) = 2621 billion MAC (multiply–accumulate) operations per second are required. The scale, speed and dynamics of this process are challenging, which is why VDSL2 vectoring only emerged as a viable option for service providers when recent advances in silicon made such sophisticated computing possible. It’s also why vectoring is best suited for small nodes with less than 400 lines, like those typically found in FTTx deployments.
Addressing migration challenges
Although most of the processing and necessary intelligence for vectoring resides in the Digital Subscriber Line Access Multiplexer (DSLAM), minimal support is needed at the Customer Premises Equipment (CPE) for the efficient estimation of the crosstalk from the line into the neighboring lines and vice versa. This additional functionality at the CPE side is defined by the International Telecommunication Union (ITU) vectoring standard, G.993.5 (G.vector). In order to achieve the full vectoring gain, all VDSL2 lines in the cable need to participate in the crosstalk estimation. Otherwise, the crosstalk from some lines will remain un-cancelled, reducing bit rates on vectored lines. The ultimate situation is where all VDSL2 lines operate in G.vector mode. This requirement can present a challenge when service providers with an installed base of VDSL2 CPEs migrate to vectoring, because migration to a new technology often occurs incrementally. Some subscribers may be satisfied with the existing performance of their VDSL2 line and choose to maintain the same service – so are not willing to go through the hassle and take the cost of a CPE replacement. Others may be eager to take advantage of higher bit rates — and are willing to purchase a new CPE that supports vectoring. Service providers must consider how to address the disparity in service levels and CPE, since both types of subscribers may coexist in the same cable (bundle). If some CPEs are not vectoring-capable, subscribers who have opted to pay for vectoring may have their bit rates reduced by any nearby lines still connected to legacy CPE. Luckily, most of the existing VDSL2 CPEs in the field can be software upgraded to support vectoring, or to be at least “vectoring-friendly”. The latter has recently been defined by the ITU in Annexes X and Y of the VDSL2 standard (G.993.2) and allows the crosstalk from the legacy line into the neighboring vectored lines to still be measured. Annex X defines requirements for downstream friendliness such that the crosstalk from the legacy line into the neighboring vectored lines can be estimated and cancelled in downstream direction only. Annex Y defines requirements for full friendliness, allowing estimation of crosstalk from the legacy line into the neighboring vectored lines in up- and downstream direction. In principle, ‘friendly’ customers do not benefit from vectoring gains but their equipment no longer impairs vectoring for subscribers who are paying for this enhancement.
The impact on regulation
Migration to vectoring involves VDSL2 lines from a single service provider, who has sole control over all VDSL2 lines in the cable. But what happens in an unbundling scenario, when the service provider must give alternative operators access to its lines? Although Asymmetric Digital Subscriber Line (ADSLx) Local Loop Unbundling (LLU) from the CO – widely used today – has little impact on vectoring gains, Sub-loop Unbundling (SLU) of the VDSL2 lines used in FTTx does have an impact on vectoring. It reduces bit rate gains, because full coordination cannot be achieved over the entire vectored system when a competitor connects subscriber VDSL2 lines to a different DSLAM. Even a single VDSL2 line terminated on another DSLAM can significantly impair vectoring performance, if that line is a dominant disturber. Currently, SLU is uncommon in most countries because of its complex business case and the collocation of different service provider DSLAM equipment in the same or separate street cabinet causes practical issues. However, its effects could become far reaching, since vectoring technology offers one valuable option for achieving universal broadband goals. When operating most effectively, VDSL2 vectoring provides a fast, cost-effective way to deliver high bandwidth over existing infrastructure. But if the achievable bit rates are reduced by unbundling, subscribers in many regions will have to wait longer to get their high-bandwidth services. From a technology point of view, the best way to address this issue is to allow the first FTTx service provider to terminate all the VDSL2 lines in a bundle, then offer bitstream access to competitors. This approach enables all service providers to benefit from the higher bit rates. Of course, this is a topic for regulatory discussion.
The benefits of next-generation xDSL
As service providers make the gradual transition to fiber, they can benefit from reliable, cost-effective options that allow high bandwidth to be delivered on a faster schedule. As shown in Figure 2, next-generation x Digital Subscriber Line (xDSL) technologies, such as VDSL2, bonding and vectoring, are providing a growing number of ways to satisfy subscriber demands for bandwidth for the foreseeable future, using existing copper resources. With more than 1.2 billion of the world’s households currently connected to a copper line, these technologies can also help governments meet their goals for universal broadband, which is critical for e-health, e-learning and socio-economic development.
Figure 2: Delivering higher speeds over copper — today
Next-generation xDSL advances began in 2007, when FTTx deployments combined deep fiber with VDSL2 to provide speeds up to 40 Mb/s at distances up to 400 m. In 2010, VDSL2 bonding combined 2 copper pairs to deliver twice the bit rate (80 Mb/s over 400 m) or serve longer loops (40 Mb/s over 1,000 m). Now vectoring has pushed the rate even further, up to 100 Mb/s over 400m. Recent research has produced even higher speeds — 300 Mb/s over two 400 m lines in the Alcatel-Lucent lab - by combining VDSL2 bonding, vectoring and Phantom mode. Fiber deployment is now underway, but it will require years to complete — while the copper infrastructure is widely available. These existing resources can be leveraged to help many countries meet their timelines for universal broadband, and service providers can use the copper infrastructure to deliver higher speeds, in less time, with faster return on investment. To contact the authors or request additional information, please send an email to firstname.lastname@example.org
- ITU-T standard G.993.5 (04/2010), ‘Self-FEXT cancellation (vectoring) for use with VDSL2 transceivers’.
- ITU-T standard G.993.5 (09/2011), ‘Very high speed digital subscriber line transceivers 2 (VDSL2)’.
- World Development Indicators, The World Bank