"Hybrid" Access Networks Continue to Make Sense
Historically, times of technology transition result in “hybrid” approaches. Once upon a time, sailing ships were outfitted with steam boilers as well.
Decades ago, building on pair gain technologies originally developed to support what we know as “T1 or E1” service, engineers began experimenting with Asymmetrical Digital Subscriber Line, essentially a modification of the method used for T1 service.
Initial results were not so promising. Some thought it never would be possible to make ADSL workable in real-world operating environments, even if some techniques worked in the laboratory.
Of course, once upon a time, engineers at Bell Laboratories also believed it was impossible to deliver 20 channels of analog video using a single pair of lasers. But systems delivering at least 80 channels of analog video eventually were developed.
Once upon a time, engineers doubted a high definition TV signal could be delivered in just 6 MHz of bandwidth. At the time, 45 Mbps or so was thought to be necessary. Today, standard broadcast TV uses a 6-MHz signal.
The point is not simply that hybrid approaches are a common bridge between technology eras, but also that “what is possible” can change, dramatically.
So it is that BT reports G.fast trial results suggesting that “1 G.fast” technology could deliver 700 Mbps speed downstream and 200 Mbps upstream, if G.fast was deployed, to about 80 percent of locations now served by fiber to the curb networks (FTTC).
That would have been unthinkable two decades ago, and improbable a decade ago. But as with earlier transformations related to multichannel analog video using a single laser or HDTV bandwidth, additional effort has wrung unexpected output from older technology platforms.
The BT trial is part of development efforts seeking to reach gigabit speeds over hybrid networks using copper drops and optical fiber distribution.
In its recent G.fast tests, BT has been able to deliver download and upload speeds of 786 Mbps downstream and 231 Mbps upstream over an FTTC line using a 19-meter (about 62 feet) copper drop cable.
That is significant since many fixed networks in suburban or urban areas feature copper drop cables of perhaps 100 feet to 150 feet long. To achieve 700 Mbps, a service provider would have to use a “fiber to the telephone pole” or “fiber to a cabinet” network that places cabinets about as densely as a telephone pole network does.
BT also tested G.fast with a 66-meter (216 feet) copper line and found that the download speed fell to 696 Mbps while top uploads weighed in at 200 Mbps.
In other words, “close to gigabit speeds” might be possible over relatively standard drop cable networks.
The International Telecommunication Union is working on standards for G.Fast, so commercial deployment will have to wait a bit. Some think the standard could be finished by 2015.
But the development shows how much value a hybrid approach can yield during a time of technology transition. Decades ago, few believed such performance was possible, using such pair gain technologies.
But industries have very high incentives to do something dramatically better in the context of a legacy technology platform.
Broadcasters “needed” the ability to deliver HDTV in the same bandwidth as analog TV. Cable operators “needed” to preserve transparency of video format (no transcoding) as they added optical fiber backbones.
Telcos “needed” to preserve the value of expensive copper drop cable networks as they have progressively boosted speeds. But hybrid approaches abound elsewhere as well.
Hybrid computing environments, partly using cloud and also premises-based data centers, now are relatively common. Mobile service providers use both their own networks and Wi-Fi to support communications.
So work on gigabit access networks using hybrid approaches is not unusual.
Bell Labs, the research arm of Alcatel-Lucent has achieved 10 Gbps speeds over very-short distances as well.
Bell Labs XG-FAST, an extension of G.fast technology, has been shown to reach gigabit speeds at distances of about 70 meters (229 feet).
By way of comparison, G.fast has been shown to deliver 500 Mbps over a distance of 100 meters (328 feet).
Bell Labs also achieved 1 Gbps symmetrical service over 70 meters on a single copper pair. 10 Gbps was achieved over a distance of 30 meters by using two pairs of lines (by “bonding” two pairs of wires.
“Fiber to where you can make money” is an aphorism illustrating the economics of optical fiber access network deployment, and the aphorism remains apt.
Internet service providers have high incentives to maximize the use of existing copper access facilities in many scenarios, even if fiber to the home makes more sense in new installations.
Maximum aggregate speed
G.fast phase 1*
G.fast phase 2*
Bell Labs XG-FAST**
2 Gbps (1 Gbps symmetrical)
Bell Labs XG-FAST with bonding***
10 Gbps (two pairs)
* Industry standard specifications. G.fast allows for upload and download speeds to be configured by the operator.
** In a laboratory, reproducing real-world conditions of distance and copper quality.
*** Laboratory conditions