Be Careful - Cost Effective Fiber Design Is Not Plug-n-Play! Op-ex versus cap-ex verus Ageing of Fiber over 30 years
April 28th, 2007 by Gordon Cook
Responding to Herman Wagter’s statement on my Symposium mail list: Hendrik Rood added a new layer of complexity: One has to look not only at fiber architecture but also at its interplay with optical equipment at the ends. The spectral capacity of the passive optical network itself is so large that the real carrying capacity is determined by the equipment at the central office and the premises side.
COOK’s Edge: Tim Riordan of Telegeography pointed out precisely this albeit in a different context: “On terrestrial networks, deep declines in the cost of adding high-capacity circuits are prompting carriers to overbuild networks with new generation equipment. This results in massive new amounts of bandwidth at sharply reduced costs.â€
Henrdrik continues: The lifecycle of PON’s passive infra is a similar 30 years. Therefore the real risk in fiber infrastructures is premature aging of fiber from a technical perspective.
Most long distance single mode fiber networks have been built only since about 1985 (before that year it was often multimode fiber, G.651!). Since then G.652 (standard single mode fiber) has evolved from:
- only a 1300 nm window at ca. 0.5 dB/km (the first versions, in non-lab production volumes since about 1984)
- joint 1300 nm and 1550 nm window each around 0.4 dB/km (available since 1986)
- 1300 nm window and improved 1550 nm at around 0.25 dB/km (available since 1989)
- 1300 nm window and very low 1550 nm attenuation <0.20 dB/km, but still with high 1550 nm dispersion (since about 1992 on the market)
- 1300 nm window and 1550 nm window , with high 1550 nm dispersion but low Polarization Mode Dispersion (since 1996)
- 1300 nm window and extended 1550 nm window (from 1490 nm to 1620 nm), with high 1550 nm dispersion and very low Polarization Mode Dispersion (since 2000)
- Reduced and low water peak fiber G.652 (now G.652.d), allowing a single window from 1280 nm to 1620 nm and more space for Coarse WDM (standard since 2005, pre-standard Lucent Allwave fibers available in 2001)
Around 1990, when I entered the industry in optical systems engineering, the talk of the day was the breakthroughs with Erbium Doped Fiber Amplifiers (1988), they allowed a new kind of long-distance transmission: solitons. Before that much R&D effort was spent on exotic crystal (ZBLAN) and hollow fibers with ultra low attenuation. This massive R&D effectively evaporated overnight in 1990.
In response to the new perspective of solitons came G.654 Dispersion shifted fiber in 1550 nm window, which was optimized for Soliton transmission (1992) and allowed for high bandwidth all-optical ocean spanning submarine systems.
However this G.654 fiber was not so fit for another breakthrough at the end of 1980s, the Phasar or Arrayed Wave Guide. So we got the introduction of G.655 Non-zero dispersion shifted fiber in 1550 nm, optimal for DWDM (first pre-standard versions since about 1996).
The huge long distance fiber construction boom at the end of 1990s was with G.655 (or pre-standard versions like Corning’s LEAF), effectively an entire overbuilt of the G.652 fiber networks constructed in the 1980s up to the year the first long-distance fiber glut in 1992 became evident.
I also have noticed Negative Dispersion Fiber and some other exotic types (like radiation hardened fiber and polarization maintaining fiber).
So while we have had a rapid improvement in fiber types, I have also witnessed several overbuilds and removal of older generations because of their aging properties.
I also have directed field efforts to replace glue spices with fusion splices (glue splices worked only for 1300 nm transmission, but reflected at 1550 nm)
There are some reasons to consider current G.652.d fiber as a more-or-less finished product in engineering terms. In many of its properties, it is nearing fundamental physics limits and material limits. However the above synopsis of Single Mode fiber evolution should give a stern warning to everyone not to overestimate the lifecycle of fiber plant currently put into the ground.
I would neither bet on the fitness of today’s G.652.d plant to last 30 years for point-to-point as well as point-to-multipoint passive deployments.
P.S. Phasar was the original device name, developed by Meint Smit, the inventor of this passive core component for DWDM optical multiplexing in InP semiconductors. I, with fellow students at Delft University, who were then doing our Master thesis research in the submicron semiconductor technology group, often were not that happy, when Smit, then a PhD student of the optics group, came in on Friday afternoon claiming our Electron Beam Pattern Litography pre-processing computers for the weekend for his jobs to execute.
Normally the weekend was graduate student time. Well let’s say we were not that well aware where he really was at. Only when I entered the optical networking industry I started to grasp the impact of his work. It is really fun afterwards, to understand that you were just observing breakthrough work and remember that your main thoughts were: “could that guy not stay away with his time consuming computer jobs”.
Now, some of the students of Smit have joined
One of the prime points to make is that Infinera’s integrated 10×10 Gbit/s transceivers make excellent master head end electro-optic devices for a WDM-PON at a far lower cost than any of the solutions requiring 10 separate 10 Gbit/s transceivers.