'Routing Will Move from Electronics Domain to Fibre Optics,' - Dr. Alastair M Glass, director, Photonics Research Lab, Lucent Technologies Bell Laboratories.

VoicenData Bureau
New Update
src="glass.gif" width="200" height="244" alt="glass.gif (31839 bytes)"> "Routing Will Move

from Electronics Domain to Fibre Optics,"

name="Dr Alastair M Glass, director, Photonics Research Lab, Lucent Technologies Bell Laboratories."> face="Arial">Dr Alastair M Glass,

director, Photonics Research Lab, Lucent

Technologies Bell Laboratories.

face="Arial" SIZE="2" COLOR="#016077">Dr Alastair M Glass COLOR="#000000" face="Arial"> is the director, Photonics Research Lab of Lucent

Technologies Bell Laboratories. He heads all Bell Labs research on optical fibre and

photonics systems.

He was in India recently to address the plenary

session in Photonics ’98 and deliver the keynote address in the ceremony for the

Fibre Optics Person of the Year award.

He speaks on the directions in fibre

optics research and usage while giving some interesting insights into the paradigm shifts

in the capability of optical fibres to
COLOR="#016077">Shyamanuja Das. SIZE="2">


Today, you have data packets

with intelligence on how they will reach next destinations. With WDM and multiple

channels, you can mark those packets with the final destinations.


do you see the capability of fibre optics 10 years from now?

That is a long period.

Fifty to hundred terabits per second is what one can imagine. But that is also not

practical. Roughly, the capacity increase will be growing by a factor of two every year.

That way, you get 128 terabits per second in seven years. But I don’t think that will

happen. After the initial few years, it should slow down. I think 10 terabits per second

is most likely.

And I am talking of long-distance. Because

the capacity also depends on distance. A big trend today is the metropolitan area

networks, where fibre is being deployed. It started with long-distance. It is coming to

metropolitan networks. It should come to LANs. But there, 10 gigabits is what will get


You said a factor of two annually. But electronics, if we

go by Moore’s law, will grow by a factor of two in 18 months. That means there is a

lag. Also, fibre technology is already ahead. Don’t you think there will be a problem

if electronics does not keep pace with the developments in fibre and photonics?
SIZE="2" COLOR="#000000">

There is already a problem. Even with 10

gigabits per second, if you use TDM, it means your electronics equipment have to switch on

and off at that speed.

But WDM does not require that switching

speeds in electronics. In fact, WDM is actually an answer to that problem. You are not

constrained by the speed of electronics.

But do we need that kind of capacity? SIZE="2" COLOR="#000000">

Of course, in the short run, we do. But

after some time, we will have reached a point where we will have excess capacity. Then

there will be a wastage of bandwidth. Today, do you bother about how much hard disk space

lies unused? Does anyone care about efficient management of that resource? When technology

matures, it is available in plenty and people start wasting it. That will happen with


Today, there is so much concern, because

bandwidth is scarce. What I am not suggesting is we should not worry about bandwidth.

Today, everyone knows we require bandwidth. But there is a limit. A time will come when it

will be wasted. Say, when every person gets a gigabit of it (laughs)!
SIZE="2" COLOR="#016077">

But the advances in fibre optics research is only for

higher capacity, higher transmission speeds. Do you think any functionalities of

electronics will come to fibre?

Yes. Routing will come to fibre. Today,

you have data packets with intelligence on how they will reach next destinations. With WDM

and multiple channels, you can mark those packets with the final destinations. And they

will straightaway travel there. All destinations will virtually be point-to-point

connected. The network will become a virtual mesh.

Is that just a concept ... COLOR="#000000">

No. We are working on it in our labs,

though a field testing is yet to be done. And it is working in the labs. We are trying to

convince the carriers. I think it will happen soon.

OK. Let’s now come to the most important thing. The

cost. What are you doing to bring the cost down?

Today, for any researcher that is the

driving factor. In fact, large scale deployment–fibre-to-the-office and

fibre-to-the-home–is possible only when cost comes down drastically. We are working

to reduce the cost of the fibre networks in many ways.

These fibre-based data networks that I

talked about is much cheaper. WDM will make things cheaper as you do not spend on costly

electronics. Greater automation in manufacturing will also help to bring the

costs down. In fact, fibre is already cheap. It is the connectors, filters, the laser
devices which are costly. Fibre is actually cheaper than copper ...
SIZE="2" COLOR="#016077">

Well, I do not think that is true today. Optical fibre

cables are much costlier than copper cables ...

Well, I do not understand that. Fibre

should be cheaper than copper. I would like to know why it is not. You should ask these

questions to the people who are selling them. All the companies. As a researcher, I do not

know the dynamics of pricing. But I would definitely like to know why is fibre not cheaper

than copper. I do not see any reason.

Yes, when I say fibre, I am talking of

single-mode fibres which the telcos use. Multimode fibres which are used in corporate

networks are costlier.

I think it has to do with volumes. Fibre volumes are still

much lower compared to copper ...

Probably that is the reason. It is still

tens of thousands, not multi millions.

When do you think developing nations like India will go

for technologies such as WDM?

They should plan now. They can start

small. They do not need to go for 80 wavelengths to start with. They can start at two and

go on adding more boards as and when they need more capacity. But deploying fibre now

makes a lot of business sense, as you will protect investment.

In fact, I think even for India, not

investing now is not sensible. We know the same thing will happen in India what has

happened in the US, when Internet usage grows. When you have time and not much legacy, it

is better to make use of the situation. India can actually avoid the bottlenecks others

are facing. I think India will do that.

Are you hopeful? COLOR="#000000">

Yes. I met Chandrababu Naidu, the chief

minister of Andhra Pradesh. He is planning to deploy a high capacity network in Hyderabad.

He is passionate about his plans.

When do you think the major transition from copper to

fibre will happen globally?

It is beginning now. I think next year,

there will be a lot of activity. And the subsequent year will see that change. I feel the

next two years are very, very important. The diminishing boundaries between voice and

data, between the telco business and enterprise business, will be the catalyst.
SIZE="2" COLOR="#016077">

To conclude, what are the major challenges before the

fibre optics research community?

I see three major challenges. One, of

course, is more capacity. That means faster electronics for TDM. Efficient use of the

technologies. What I mean is the right combination of TDM and WDM.

Second big challenge is wavelength

agility. Being able to convert from one wavelength to another easily. Today, we are

beginning to do that in labs. It is still not possible commercially.

Third and the biggest challenge is optical

cross connect. Efficiently being able to do that will take us a long way. Today you have

80 channels, and if you use eight fibres, it is 640. Being able to cross connect them is


And above all, the

underlying objective, the most important one, as you also said, is cost. Bringing the cost

down. That is the single most important challenge for the whole fibre optic community.