The Numbers are in: Vectoring 2.0 Makes G.fast Faster
Vectoring 2.0 proves its value
The emerging G.fast standard holds promise but is even more susceptible to crosstalk, or noise among the lines, than VDSL2. Vectoring 2.0 will make G.fast live up to its performance promise, just as vectoring does today for VDSL2. G.fast is a standardization initiative in the International Telecommunications Union (ITU). It can increase aggregate — upstream plus downstream — bit rates over copper loops shorter than 250 m to fiber speeds of more than 1 Gb/s. It also offers a cost advantage over deploying fiber directly to the home. The International Telecommunications Union (ITU) G.fast project covers the potential for very strong crosstalk coupling and the need for vectoring But it took testing and development to fully understand the impact of crosstalk on G.fast. Through research at Bell Labs, Alcatel-Lucent identified a number of factors that increase the complexity of vectoring with G.fast. This complexity is driving new innovations that will result in Vectoring 2.0. Recent trials at Telekom Austria using prototype technology from Bell Labs confirm the value of Vectoring 2.0. Tests of G.fast on a good quality cable achieved aggregate speeds of:
- 1.1 Gb/s over a single pair of 70 m lines
- 800 Mb/s over a single pair of 100 m lines
Tests on the older, unshielded cables found in many buildings in Austria achieved speeds of 500 Mb/s over 100 m when a single line was active. However, when a second G.fast line was added, crosstalk dropped G.fast speeds from 500 Mb/s to just 60 Mb/s. G.fast with prototype Vectoring 2.0 technology removed the crosstalk and brought the speed back up to 500 Mb/s over 100 m.
The need for G.fast
With demand for high bit-rate services straining access networks to their limit, the industry is looking for a solution to the next bandwidth bottleneck. G.fast promises to be that solution. Fiber to the home (FTTH) was initially seen as the only long-term solution to the bandwidth problem. But VDSL2 vectoring changed this perception. With a single innovation, the market shifted. Copper became a valuable commodity again as operators began using their copper assets to deliver fast broadband speeds faster. As Figure 1 illustrates, VDSL2 today uses up to 17 MHz of spectrum in outside plant applications. The G.fast standard will allow for 106 MHz and 212 MHz profiles, providing a significant increase in bandwidth. To keep implementation complexity manageable, bit loading in G.fast is limited to 12 bits per frequency carrier; VDSL2 carries up to 15 bits per frequency carrier. Due to the loop attenuation that increases with frequency, these extremely high frequencies are only available for bit loading on very short loops. This is why G.fast is seen as a technology that is intended to operate over loops that are less than 250 m on 0.5 mm cable.
Figure 1. G.fast uses very high frequencies to significantly increase bit rates on very short loops
Innovations target crosstalk challenges
With VDSL2, interference among multiple active lines in a copper cable significantly reduces performance compared to the case where a single line is active. The impact of this interference — known as far-end crosstalk (FEXT) — is unpredictable and causes VDSL2 to perform well below the promise of 100 Mb/s. Vectoring allowed VDSL2 to reach its full potential.
Figure 2. Crosstalk among lines in the same cable reduces performance
Crosstalk also degrades performance when multiple G.fast lines occupy the same cable binder. Bell Labs studies indicate that the effects of crosstalk are much greater with G.fast than they are with VDSL2. The very high frequencies that G.fast uses are at the root of the crosstalk challenges. At these frequencies, it is not uncommon to see crosstalk on a G.fast line that is similar in strength to the actual signal. One challenge is to create a compensating signal that eliminates crosstalk without exceeding the Power Spectral Density (PSD) mask. More advanced algorithms are required to compensate for these high crosstalk levels. The broad frequency range used by G.fast — 6 to 12 times that of VDSL2 17a — adds a factor of scale. A wider frequency range means more calculations per second for the vectoring engine. Figure 3 provides an example of the gains that can be achieved when vectoring is used with G.fast on a high-crosstalk cable.
Figure 3. G.fast with vectoring eliminates crosstalk to keep speeds high
Actual gains will depend on loop length and line quality. The blue line indicates the performance when a single line is active in the cable. Once additional G.fast lines are activated in the cable, performance drops significantly (red line). The blue and red lines are benchmarks by which all vectoring performance is measured. Activating G.fast vectoring causes a significant increase in performance (green line). In Figure 3, vectoring increases the aggregate rate at 50 m from 250 Mb/s without vectoring to 650 Mb/s with vectoring. The attainable bit rates in each network will vary based on network conditions.
Evolve with VDSL2 and G.fast
Because G.fast is designed for ultra-high speeds and very short loop lengths, it is an ideal evolution path for deep-fiber deployments. As illustrated in Figure 4, small nodes are deployed very close to the subscriber, in any location where the operator has access to the copper cable. The location could be at the curb, in a building, on the outside wall of a home, or anywhere in between.
Figure 4. G.fast supports multiple deployment models
These deployment models are referred to with FTTx terminology such as FTTCurb, FTTBuilding, FTTWall, or by the generic name Fiber to the Distribution Point (FTTdp). The deployment models share the same characteristics — small nodes, very short loops, a small number of subscribers (tens or fewer), and very high bit rates. These characteristics are basically the fixed networks equivalent of wireless small cells. Typical applications for G.fast and VDSL2 vectoring will depend on the loop length and the number of subscribers:
- >200 m: In deployments with longer loops, VDSL2 vectoring is, and will remain, the technology of choice. G.fast is simply not optimized for these loop lengths. The G.fast standard targets 150 Mb/s aggregate speed at 250 m on a 0.5 mm line. VDSL2 vectoring can deliver 140 Mb/s to 150 Mb/s aggregate speed at up to 400 m.
- <200 m, multiple subscribers: In deployments with multiple subscribers and short loops, VDSL2 or VDSL2 vectoring can be used today. G.fast provides an evolution path, but vectoring will be required to achieve high G.fast speeds.
- <200 m, single subscriber: Deployments with a single subscriber are a very good fit for G.fast due to its high bit rates and ease of installation. VDSL2 will be used in these scenarios until G.fast solutions become available.
Choose the right FTTx model
Different FTTx deployment models offer different advantages. Most operators have FTTH as their long-term strategy, but FTTH requires significant investment and significant time to deploy. Operators should select their deployment model(s) based on the required investment, time-to-market and required bit rates. Figure 5 compares the cost of various FTTx deployment models.
Figure 5: Operators should consider multiple factors when choosing an FTTx strategy
As illustrated in Figure 5, FTTH is about 15 times more expensive than offering ADSL from a central office. Most of this cost can be attributed to civil works; digging up every street and going into every home to install new fiber infrastructure. FTTN with VDSL2 vectoring rates about 4 to 5 on the cost scale, around 3 times cheaper than FTTH. Reusing the copper infrastructure in the last mile significantly reduces the civil works cost. The cost of deployment models in between FTTN and FTTH varies with the proximity to the end user. The closer the deployment gets to the end user, the closer the cost comes to that of FTTH. For example, FTTB rates a 10 on the cost scale which is about 30% cheaper than FTTH, while FTTwall is comparable to FTTH. But cost is not the only decision criterion. Time-to-market is equally important. Deploying FTTH nationwide can easily take 10 to 20 years; many governments, operators and end users simply can’t afford to wait that long. FTTx deployments can speed up roll-outs because operators can skip the last mile. Other factors can also impact the choice of deployment model. For example, deploying aerial fiber rather than buried fiber makes FTTH deployments much faster and cheaper. On the other hand, entering the home can be a major headache, and can drive up the cost and time required to deploy FTTH. FTTWall with VDSL2 in the short term, or G.fast in the future, allows operators to avoid entering the home, saving time and money. Most operators deploy a mix of FTTx and FTTH technologies. This allows them to select the best suited deployment model for a given area so they can connect more subscribers, quicker and cheaper.
VDSL2 today, G.fast tomorrow
Both vendors and operators are contributing to the maturing G.fast standard. Many key items are already agreed upon while other aspects are still being studied. The ITU goal is to have standard approval in 2014. Once this milestone is achieved, vendors can begin work on standards-based chipsets. If standard development cycles are followed, the earliest availability of G.fast products is the end of 2015. Meanwhile, pre-standard prototype implementations will allow vendors to continue developing and testing the technology. While G.fast is not yet standardized and won’t be commercially available for several years, it is a natural evolution of VDSL2. In the meantime, operators can rely on VDSL2 vectoring to serve their customers in a cost-effective way.
Editor’s Note: The authors would like to thank Keith Russel for his contribution to this article.
To contact the authors or request additional information, please send an e-mail to firstname.lastname@example.org.