Friday, March 13, 2015

Is DSL Progress at a Limit?

With increasing stress on some service provider business models related to fiber to the home investments, the logical question is how much more can be done to prolong the life of the copper access infrastructure.

For 25 years, the answer has been "lots more." And some believe the progress is not yet at a limit.

The ability to deliver higher speeds over digital subscriber line or hybrid fiber coax,  for example, translates into lower retail costs, since the full replacement of the access network is avoided.

In that regard, it is worth noting that digital subscriber line has surprised even its would-be supporters. There was a time when the chief technology officer of a tier one global infrastructure supplier could say, privately, that “DSL won’t work.”

But smart people threw effort at the problem, and DSL did work. The big tradeoff has been distance versus bandwidth, so shorter access loops mean higher speeds are possible.

The other persistent limitations include line noise and the availability of spare wire pairs to bond.

Oddly enough, in many cases the abandonment of voice services by a majority of former users means there are more available copper pairs to be bonded for the remaining potential customers.

Roughly speaking, 50 percent fewer customers also means roughly 50 percent more copper pairs available to be bonded into high speed access lines.

Within some distance parameters, it is believed possible to push beyond 10 Gbps. The issue is the distance at which this is possible, how well the existing physical plant corresponds to ideal laboratory conditions, and the service provider’s ability to pull fiber deep enough into a serving area to support short access loops.

At least as Alcatel-Lucent’s Bell Laboratories sees matters, progress is not at a limit. That is a fundamental change from some thinking in the early 1990s.

In principle, and with the caveat that distance matters, speeds ratcheting to 40 Gbps are conceivable.

The lesson here is clear. When we say “something cannot be done,” that is a conditional statement. We mean “something cannot be done today, by us, at a cost that allows commercial operations and business models.”

Conditions can change.

There are other implications. When one tries to measure “quality,” one has to pick quantifiable metrics. Among the issues is whether the metrics one chooses represent quality as viewed by the supplier, or quality as seen by the customer.

In the end, the only metric that matters is the customer’s metrics for quality. So even if we believe fiber to the home is "better," and represents "better quality," that might only partially match customer perceptions.

The old adage that "customers don't care" about how you provide a service, only that you do so" is germane. All marketing hype aside, tests tend to show that beyond about 10 Mbps, any single user is unable to perceive an advantage compared to access at higher speeds.

Those requirements will grow over time, as they have since people began using the Internet widely, but the principle remains: end user experience increasingly is not dictated by the bandwidth of the local access loops.

Nevertheless, supplier choices do matter, because end user assessment of quality includes a “price” evaluation, not simply a “performance” perception.

That matters since all supplier costs inevitably are reflected in end user retail prices. If faster DSL allows lower retail prices, that is a "better" choice for many ISPs than ripping out all access network copper and substituting fiber to the home.

To a startling degree, headline speeds are a marketing necessity, but not necessarily an end user perceivable value. Beyond a certain point, headline speed increases yield no discernible end user advantage. 

That might not matter. Marketing concerns often dictate the ability to sell faster speed service, even if such enhancements do not lead to higher end user quality of experience.

No matter. ISPs will seek to deliver high speeds, because that is what successful marketing requires.

Bell Laboratories Technology comparison
Technology
Frequency
Maximum aggregate speed
Maximum Distance
VDSL2
17 MHz
150 Mbps
400 meters
G.fast phase 1
106 MHz
700 Mbps
100 meters
G.fast phase 2
212 MHz
1.25 Gbps
70 meters
Bell Labs XG-FAST
350 MHz
2 Gbps (1 Gbps symmetrical)
70 meters
Bell Labs XG-FAST with bonding
500 MHz
10 Gbps (two pairs)
30 meters

No comments:

Generative AI Will NOT have the Impact Many Expect

Generative artificial intelligence, to say nothing of machine learning or neural networks (and eventually general AI), might collectively re...