Alcatel-Lucent has shot an economic hole clean through fiber to the home technologies, with a demonstration of 10 Gbps broadband speeds over the last 30 meters of twisted pair copper wiring. The demonstration is really about Alcatel proving that there is no need to stop the new telco strategy established for G.Fast, of simply taking fiber closer and closer to the home, without ever going inside.
The last 30 meters includes scheduling with customers, and digging up gardens and drilling holes in walls, and is the most disruptive and expensive step in the creation of high speed broadband to the home in a telco environment. Some observers think it is well over 50% of the total cost.
Recently a G.hn chip from Marvell was used to bring broadband to a series of MDUs in South Korea, carrying a signal of 500 Mbps, operating in a very similar way to how the approaching G.Fast technology is supposed to be used.
Successive telco technologies have required an unceasing move closer and closer to the home with fiber. In much of Europe, where loop lengths have always been on the short side compared with say the US, this has seen DSLAMs move from the exchange some 2.5 Km to 5 Km from each home, to the cabinet just a few hundred feet away, as telcos have prepared to introduce vectored VDSL2.
But while vectoring has been slow to get off the ground, the reason for this is the need to cancel cross talk between all the lines in any given bundle. Coping with this has led to a long pause in the succession of VDSL2 over ADSL2+. It was back in 2005 that chip vendors like Broadcom began to put ADSL2+ and VDSL2 into the same chipsets, and had them automatically distinguish between the signals they received. Since then, moving the modulation point closer and closer to the home has also been slow, as telco networks had to lay fiber closer and closer and create new classes of cabinets to hold smaller, weatherproof DSLAMs.
But once the mechanisms are in place to cancel cross talk, then successive broadband technologies can be installed on telco lines using Vectored VDSL2 to start, G.Fast to follow and potentially something similar to this Alcatel lab version after that, giving measured progression.
The truth is there is little mystery. Components get better at handling DMT (OFDM) signals as their dwindling geometry allows for more complex chips, but this happens at a given rate. As you get closer to a home the higher frequencies on a phone line, which until now have been no good to you, because they attenuate too rapidly, can be brought back into play, and this is what Alcatel is doing.
It initially promises to offer a full 1Gbps symmetrical service (both ways) up and down a copper wire over 70 meters. So once a telco has solved its crosstalk problem, it can simply carry on taking fiber closer and closer to the home, down from the 100 meters specified for 500 Mbps G.Fast, to 70 meters and then on to 30 meters. There is also a second generation of G.Fast that will use 212 MHz that is expected to deliver an aggregate of 1.25 Gbps shared between two directions (about half the speed of the Alcatel experiment) , which also requires the node to be 70 meters from the modem.
But while the economics remain way better than full FTTH, they have many disadvantages which are reminiscent of the small cells problem of cellular carriers. There has to be more stop off points in their network, more and more splitting of fiber to sustain a larger number of points for backhaul, and an increasing number of street cabinets, until it gets to the same point as cable, where there is a box stuck on the outside wall of every home.
The physical amount of extra fiber that needs to be laid is small, but the number of physical elements to the network goes up astronomically. Right now a large DSLAM will look after 384 homes, so one installation for 384 homes. Soon it will be 192, then 48 and in order to be close enough to all the homes in a loop, there will be a new cabinet for 3 or 4 homes and finally one for one.
The Alcatel Bell Labs demo uses a prototype technology called XG-FAST, a simple extension of G.fast, which itself is due in 2015. G.fast uses 106 MHz (compared to a maximum of 30 MHz on VDSL2), and should give broadband speeds up to 500 Mbps over a distance of 100 meters. XG-FAST uses an increased frequency range up to 500 MHz to achieve higher speeds, but this has to be over much shorter distances. The 10 Gbps speed was only achieved over a distance of 30 meters using two pairs of bonded lines, but worked ‘in situ’ on an existing European network (we would guess Belgacom, because it is so friendly with Alcatel).
This is perhaps less important news for the US and Latin America where loop lengths are long because of the vast spaces between housing. MDUs are another problem entirely, although in many cases the same chips, can be used in a different topology, with fiber climbing halfway up an MDU building. That means in Asia Pacific, where the MDU problem is perhaps more urgent, may require slightly different solutions ‘ hence the Korean use of G.hn chips.
But outside of the Americas it looks like Cable operators will shortly, in the 2020 timeframe, be looking at nothing better than parity with Telcos, albeit with vastly differing wiring topologies.
What this announcement does is give confidence to telcos who are preparing to build out suitable fiber proximity for G.Fast, knowing that a simple continuation of that strategy should see most operators into the Gigabyte era and beyond.
It is a strong message from Alcatel because it keeps its name at the forefront of copper broadband and is from a vendor already known for fiber installations, which suggests that it has no particular technology axe to grind.
Marcus Weldon, President of Bell Labs: ‘Our constant aim is to push the limits of what is possible to ‘invent the future’, with breakthroughs that are 10 times better than are possible today. Our demonstration of 10 Gbps over copper is a prime example: by pushing broadband technology to its limits, operators can determine how they could deliver gigabit services over their existing networks, ensuring the availability of ultra-broadband access as widely and as economically as possible.’
Expect companies like Huawei and ZTE to now mount their own alternative demonstrations in the coming months to prove that they too can head down this route, once the technology gets standardized. The expect a war for standardization to begin shortly after G.Fast standards are set in concrete. Alcatel has not yet said anything further about Phantom mode, another broadband augmentation that it showed working back in the 2010 timeframe, when a variety of vendors made statements about increased broadband speeds. Phantom mode works by using two pairs, and sending a third signal across the two combined, without interfering with the original signal, bumping up the upper limit by around 50%. While Phantom mode was shelved while the world sorted out vectoring, it may rear its head at a later point for a simple 50% throughput hike.