Wednesday, 14 October 2009

Optical fibre: Charles Kao and where it has got to today

The standard story of optical fibre for communications invariably identifies a 1966 paper by Kao and Hockham: "Dielectric-fibre surface waveguides for optical frequencies" (Proceedings IEE 1966),as where it all started.

The paper reported the theory of light propagation in optical fibre, and some experiments at Standard Telecommunications Laboratories in the UK.

It is good to see that work now acknowledged in the Nobel prize for Physics awarded to Charles Kao.

Forty-three years later and optical fibre is the backbone of communication networks. It is a long time since copper cables have been used for communication links between towns and cities, or for trans-oceanic communication links: fibre is just so much better. The technology has developed pretty much continuously since 1966, and here are a couple of recent reports:

FCC approves new trans-Pacific fiber-optic cable

By John Boudreau
Posted: 10/08/2009 02:39:07 PM PDT
Updated: 10/09/2009 07:29:56 AM PDT

TAIPEI, Taiwan — The U.S. Federal Communications Commission has given the green light for a trans-Pacific fiber-optic cable funded by an international consortium that includes Google. The new cable, expected to be carrying data traffic by early next year, links the U.S. West Coast and Asia to meet the demand for more bandwidth to handle explosive global Internet communications.
The 6,200-mile cable, costing about $300 million, is being funded by six companies that, in addition to Google, include telecommunications companies Bharti Airtel in India, SingTel of Singapore and Pacnet, a Hong Kong-based deep-sea fiber-optic cable network operator.
The cable will run along the ocean floor from Southern California to Chikura, Japan, dipping as deep as 2,000 feet below the surface, and then connect into other networks. Pacnet will control two of the five fiber pairs in the new cable.
Each pair of fiber cables is capable of carrying up to 960 gigabytes per second, roughly the amount needed for 15 million simultaneous voice calls. ....
Meanwhile, in the labs, more and more bits are squeezed into individual fibres:
Alcatel-Lucent Hits 100-Petabit Optical Milestone

No, it's not warp speed, but Alcatel-Lucent has research technology that can transmit more data faster and further than ever before.

September 29, 2009
By Sean Michael Kerner

New research by Alcatel-Lucent's Bell Labs division has reached a new optical network transmission record: 100 petabits per second.kilometer.

The bit-per-second.kilometers unit of measure is a combination of speed and distance and is considered by Alcatel-Lucent to be a standard measure for high-speed optical transmission.

The researchers used 155 lasers to transmit the data at rate of 100 gigabits per second. The total of 100 petabits per second.kilometer was calculated by multiplying the 155 lasers by the 100-gigabit speed (15.5 Terabit per second) over the entire 7,000 kilometer distance used in the study. In practical terms, that's the distance between Paris and Chicago.

As a result, the Alcatel-Lucent test's speed maxed out at the equivalent of 100 million gigabits per second.kilometer.
This is a typical 'hero' experiment - labs competing with one another to get the headlines - so you have to be a bit wary of interpreting the practical relevance, but on past experience in the field of optical fibre comms, commercial reality might only be a few years behind.

There's some interesting features to this demo that I might write about another time, but for the moment it is interesting to compare the Alcatel-Lucent report with the previous report of the trans-Pacific system.

(My instinct as an OU author is at this point to set it as a 'Self-Assessment activity'... so I'll leave a bit of white space just in case someone reading this wants to do it for themselves before seeing my answer!)
6,200-miles, which is 8 x 6200/5 = 9,920 km
960 gigabytes per second, which is 8 x 960 = 7,680 gigabits/s, or 7.68 terabits/s

Performance in bits-per-second.kilometers, is 9,920 x 7.68 x 1012 = 7.6 x 1016

That is 76 petabits-per-second.kilometers, not so different from the Alcatel-Lucent demo - if I've done my calculation right.


Allan Jones said...

Interesting how fibre has eclipsed satellite. Would anyone have predicted this 20 years ago?

Equally interesting (to me), an apparently knowledgeable commenter at a Guardian technology story below questions whether fibre-to-the-home would really be as good as everyone claims. It's the second comment following the story below. All the comments on this story are good, and strikingly better than you normally get:

David Chapman said...

As we were discussing yesterday, technology predictions seem well-nigh impossible. That commentator is right (and as you say, those comments are quite exceptional compared to your typical newspaper comments) that data rates over copper have risen to meet demand, but no-one would have believed that possible 20 years ago. Interesting points about the limitations of fibre, but I'm not entirely convinced. I don't understand his point about challenges over distance.

Of course, supposing Britain had put in fibre to the home 20 years ago, it would probably have been multimode fibre, and we'd now be complaining about the limitations of MM fibre and talking about the need to change to SM!