The co-founder and CEO of Sckipio Technologies, David Baum, has suggested to ISPreview.co.uk that BT’s plan to roll-out 300Mbps G.fast broadband technology could save as much as £980 per home vs a pure fibre optic (FTTH/P) network and even faster speeds should be possible in the future.
Israel-based Sckipio knows a thing or two about the new G.fast (ITU G.9700/9701) technology because they helped to create about 20% of the technical standard. On top of that they’ve also assisted with Openreach’s (BT) trials of the service, which is important because the operator hopes to start rolling out across the United Kingdom in 2017.
BT currently intends to make G.fast available to 10 million premises by 2020, with “most of the UK” likely to be done by 2025 (we predict around 60% UK coverage). Initially G.fast will only offer top download speeds of ‘up to’ 300Mbps (50Mbps upload), but there is talk of later increasing this to 500Mbps.
According to David Baum, the “strategy is to build a copper technology that feels like [a pure fibre optic connection] to the user“, which they say means delivering ultra-fast speeds, very low latency, breakthrough stability, very fast recovery times and very fast training times.
But unlike a pure fibre optic connection, such as FTTH/P, the hybrid-fibre approach adopted by G.fast still depends on reusing a stretch of existing copper cable and so should be significantly cheaper and faster to deploy than pure fibre. Baum suggests a saving of as much as $1,300 (£980) per household, but experiences will vary.
However others fear that the bold ambition of G.fast’s performance may struggle to deliver in the real-world environment. Naturally ISPreview.co.uk wanted to know more and so we caught up with David Baum to ask him a few additional questions. But before that, here’s a simplified explanation of how the new technology works..
The technology itself (see our technical summary) works in a roughly similar way to the current ‘up to’ 80Mbps capable Fibre-to-the-Cabinet (FTTC) service that is available to most of the UK, where a fibre optic cable is run from an exchange to your local Street Cabinet and then the remaining copper line to your home uses VDSL2 to deliver the service.
However G.fast requires significantly more spectrum (G.fast 106MHz+ vs VDSL2 17MHz+) and thus it can only deliver its best speeds over much shorter copper lines (less than around 300-350 metres for 300Mbps+). Furthermore Openreach now intends to roll-out G.fast via an extension pod on their existing PCP street cabinets (here), but they can also deploy it from smaller nodes.
The smaller nodes could be built either underground (manholes), inside big buildings or placed on top of nearby telegraph poles, usually with a small power supply cabinet sitting nearby. The smaller nodes would most likely deliver even faster speeds because the fibre optic cable would often come closer to homes via a more expensive Fibre-to-the-distribution-point (FTTdp) network .
In theory it’s possible to pull Gigabit (1000Mbps+) speeds out of a G.fast network, but this is only possible over a few tens of metres (copper cable) and therefore does not reflect most real-world networks where the copper lines from a street cabinet can be hundreds of even several thousand metres long. An aggressive FTTdp roll-out would be needed to deliver such speeds.
1. Sckipio is a company in Israel that seems to have come out of almost nowhere in 2012 and has since dedicated itself towards helping to develop specifications and hardware for the ITU’s new G.Fast broadband technology (we believe you helped to create about 20% of the technical standard).
Can you tell us a little about the story of how Sckipio came about and what drew it to focus on G.fast so exclusively?
I have been involved in the wired networking space for 20 years. At my last company, we were a leader in wired home networking. I sold that company successfully, and when I left the company, I started to think about doing something new. I realized that there was enough bandwidth in the home. But, bandwidth to the home was now the bottleneck. It was very challenging to receive 1Gbps bandwidth to the home.
I wondered, “What is the next step for broadband access?” What are the telcos asking for? I went and visited the FTTH show in Florida and I started to talk to installers. I realized that fibre wasn’t the best solution. It was too expensive to install. The cost of trenching, the scheduling of appointments, the in-home installation would take forever. It’s not that the technology is expensive – it’s the physics, the logistics. That’s never going to change.
Doing the math of how much telcos can save using copper instead of fibre, it was clear there was a business case. We realized the opportunity was there, the market was there, the requirement was there and there was little competition. That’s why we made the investment.
2. G.fast appears to deliver most of its performance improvements vs today’s loosely similar VDSL broadband technology by harnessing more spectrum frequency, using shorter runs of copper cable, adjusting its power and adopting Vectoring as standard to help reduce cross-talk interference. Can you tell us about any other ways in which G.fast as a technology has improved upon VDSL?
G.fast is not an xDSL technology, but it does leverage best practices and experiences learned in xDSL, G.hn, WiFi, HomePNA, and many other communications technologies. It has several key attributes.
Our strategy is to build a copper technology that feels like fibre to the user. This means delivering very low latency, the fewest dead zones, breakthrough stability, very fast recovery times, and very fast training times. Plus, it uses TDD to deliver more flexibility in the upstream and downstream ratio.
3. In the UK, Virgin Media has recently committed to ensure that over 1 million premises are reached by using ultrafast Fibre-to-the-Premise (FTTP) technology and BT have similarly pledged to cover 2 million premises.
Cost is often highlighted as one of the primary reasons why operators with a lot of copper cable in their network prefer hybrid-fibre approaches like VDSL and DOCSIS.
Considering the full distance of a connection from exchange to the end-user’s home socket, how do the costs between rolling out FTTP/H vs G.fast compare?
We have found the cost savings of using G.fast instead of fibre can be as much as $1,300 (£980) or more per household. That said, each service provider has a different architecture and different costs in terms of labour, methods of bringing fibre closer, etc. So specific rates will vary.
4. Some of Sckipio’s hardware has been used in BT’s G.fast trials. Can you tell us a little bit more about how you assisted in the trials and what you learnt from taking part?
We have performed testing both in the BT labs and in the field. This has given us a broader range of scenarios and experiences – which will make it possible for us to optimize the technology further and handle all kinds of situations in a much quicker way. This knowledge is hard to replicate and enhances Sckipio’s DPU leadership versus rivals.
We learned about different kinds of wires and different cross-talk conditions, how to optimize our performance and handle things we don’t have in the lab. This has helped us verify our stability performance and identify ways to make our stability even stronger.
We also discovered the real-world Sckipio performance over long distances is outstanding. This is appreciated by all kinds of service providers who are looking to utilize existing infrastructure to improve their business case and improve the penetration rates with minimal incremental investments.
5. Openreach (BT) has hinted that they should be able to deliver G.fast broadband speeds of 300Mbps download and 50Mbps upload over a single (NOT bonded) copper line distance of 350 metres in the United Kingdom. However their recent trials struggled to deliver this at the desired distance (here), although a number of future improvements were suggested that could make it achievable.
BT’s Planned G.fast Enhancements
* Enable higher bits per tone (12>14)
* Improve the receiver sensitivity (<-150dBm/Hz)
* Increase the transmit power (4>8dBm)
* Optimise the frequency usage with VDSL
* Increased vectoring group sizes (>96)
Does this sound about right to you and what other changes might be needed in order to deliver upon BT’s aspiration of 300Mbps at 350 metres?
BT is an active participant in the ITU and brought many of these requirements to the working group. As a result, most of these features are now part of the G.fast standard. All of these features will contribute to higher performance at longer distances.
It is true that as the distance grows, more subscribers can be reached from a single central office. Therefore, there is a need for vectoring with higher port counts and we should expect higher densities in the near future. Already, Sckipio is shipping 24-port vectoring and demonstrated 32-port vectoring last fall. The ITU has made some modifications to the standard to support a large number of ports (e.g. 96).
In summary, increasing the number of bits per carrier helps mainly for shorter lines, where the SNR is relatively high. On the other hand, lower receiver noise floor and higher transmit power help improve performance over long lines.
Better utilization of the VDSL band also helps improve performance over long lines as it allows G.fast to use lower frequencies, where the line attenuation is lower.