Trends in

Fiber Optic Communications

Jan Conradi
Director, Strategy
Corning Optical Communications

2001 ICF Congress, Berlin

Good morning Ladies and Gentlemen,

It is great to be in Berlin again.
My first visit was many years ago, when I took a train from West to
East via a number of boarded up railway stations. I emerged
somewhere very near this spot, to see the Brandenburg Gate. It is
really wonderful to see now what has happened meanwhile!
Now, let us move to the first slide of my presentation.

ICFC 2001 - Networks Telecom - Conradi - page 41

Outline
Drivers
Drivers
Bandwidth
Bandwidth Demand
Demand

Technology
Technology
Driver
Driver &
& Enabler
Enabler

Summary
Summary Outcomes
Outcomes of
of Technology
Technology Evolution
Evolution

This is about Bandwidth from both a historical perspective, as well as

a technology and a service perspective.
The demand for increased bandwidth, as also explained by Ove
Alm, is of course the primary driver today. But there is more to it.
Back in 1968, at Standard Telecommunications Lab, UK, Kao and
Hockham suggested the potential of long distance transmission by
using light. The pace at which the physical technology has evolved
since that date, has really made technology a driver in the evolution
of the network. Particularly in terms of bandwidth evolution. The
resulting network capabilities in turn enable economic provisioning of
broadband services.
Also, it isnt just physical technologies such as optical fibers, lasers,
and other optical components that form part of the kit of enabling
technologies, but we need to consider networking technologies as
well.

ICFC 2001 - Networks Telecom - Conradi - page 42

The Growth in Telecommunication

Capacity/Capability
1010
108,000 VC

Bit Rate
(Bits/s)

108

Fiber Optic
Systems

3600 VC

106
Coax-60 VC

104

1st Telephone

102
1

Oscillating Needle Telegraph

1850

1900

1950

2000

Year

Bandwidth increase has been exponential for a century and a half.

In the very beginning, first telegraphs and the morse code, there was
exponential growth to some extent. However I wish to draw your
attention to the period 1930 and onwards. The capability of coaxial cable
made the telecom capacity grow at an exponential rate, not very much
different from what we have seen for fiber optics since about 1975.
There were installed coaxial systems by Nortel between Montreal and
Toronto, carrying 274 Mb/s on a single coax. So in a broad perspective,
bandwidth growth has been exponential through several generations of
telecom technologies.
A more focused concentrated perspective starts in 1985 where the early
speed increases in optical systems was more due to electronic speeds
than anything else. The early fiber optic systems operated at 45 Mb/s on
a single fiber increasing progressively to 135 Mb/s, and 565 Mb/s prior
to the introduction of SONET and SDH; then to 622 Mb/s, 2.5 Gb/s, and
10 Gb/s.
The optical amplifier came along in the early 90s and allowed longer
distances as well as more bandwidth by adding a number of wavelength
channels that could then be simultaneously amplified.
Slide courtesy of my colleague and co-inventor of the modem fiber - Don Keck.

ICFC 2001 - Networks Telecom - Conradi - page 43

Demonstrations of Optical Fiber Capacity

Growth
100000

30Tb/s
New Approaches:
PDM, S-Band, VSB

Capacity [Gbit/s]

10000

Rapid Conventional
Population of C+L

1000

Final Optimization of
Conventional Approach

100

10

Early WDM
Development

1
1985

1990

1995
Year

2000

This slide gives the evolution of demonstrations.

availability tends to lag by about 3-5 years.

2005

Commercial

Top of commercial systems is a 160 wavelengths at 10 Gb/s each.

This illustrates how various technological improvements each gave
rise to significant increase in total transmission capacity.
The conventional band or C-band 1532-1565 nm, the Long wave Lband up to ~1625 nm and the Short-wave S - band down to 1420 nm.
The optical people discovered - like the electrical engineers before that various modulation techniques can be used to
improve
efficiency. Such new approaches for optics are explored in recent
years:
PDM - Polarization
transmission) and

Division

Multiplexing

(used

in

satellite

VSB - Vestigial sideband (originating from cable TV).

The red line at the top represents an approximate limit from
information theory standpoint for binary signals in the low loss fiber
window between 1420 & 1670 nm. That is in the order of 30 Tb/s for
one fiber.

ICFC 2001 - Networks Telecom - Conradi - page 44

USA Traffic Demand Growth

10,000,000
1,000,000

Voice
8%/yr domestic
17%/yr international
Transaction data
34%/yr domestic
52%/yr international
Internet
157%/yr
x Total traffic

Traffic demand is
related to the number
of users and their
required bandwidth

100,000

Gb/s

10,000
1,000
100
10
1

1995

2000

2005

2010

2015

Year

Lets turn our attention to actual traffic demanded of the network as in

the USA. This slide on a logarithmic scale is similar to what Ove Alm
just showed.
For voice even with the proliferation of cell phones, the growth rate of
aggregated voice traffic is still only 8%/yr compared with 34 to 55%/yr
for conventional data traffic and then 150%/yr for Internet. This is
closely tracked; most recent data is July of this year. No saturation is
seen at all for internet traffic. This growth is due to four factors; which
when multiplied together gives the rate at which total traffic demand
grows: the number of users x a growing number of hosts x the time
on line x the bitrate that an individual user has access to. Also here
we find that the higher the access bitrate, the longer a user tends to
stay on line.
You dont have to be a rocket surgeon, as a Canadian hockey
commentator once said, to realize that Internet is rapidly beginning to
dominate.
Similar growth rates exist in other parts of the world, but from
different starting points. This means that similar curves develop there
but with some what later intersection points.

ICFC 2001 - Networks Telecom - Conradi - page 45

Traffic Demands

Metropolitan

USA

International

7
6 Traffic Growth
Internet
5
Data
4
3
2
Voice
1
0

1995 2000 2005 2010

To consider traffic & technology we need to consider how the traffic is

distributed over the world wide network.
This slide is a US centric view for which I apologize, but it does
illustrate a point.
Voice traffic is usually clustered around metropolitan networks or
other relatively small regions - we speak somewhat more to
neighbors than to somebody across the ocean.
Corporate or transaction data is a bit more spread out, but limited to
clusters where the businesses are established.
Internet data in almost totally diffuse and distance in-dependent.
This has dramatic impact on the nature of the network, where now
data calls are almost independent of distance. The traffic growth here
is very rapid and the distances are of global nature. That means that
the cost and revenue mix is dramatically different for the network
operator when compared to voice or conventional data services.
Most of the revenue still comes from the services which are clustered,
but most of the costs now come from the demand for long distance
capacity. This explains the strong requirement that the costs for long
distance communications come down dramatically in order to provide
for the growth.

ICFC 2001 - Networks Telecom - Conradi - page 46

Network Segmentation
Long Haul Transport
Regional Transport

Residential
Access
Business
Access

If we break the terrestrial network down to a slightly different level, we

can clarify different portions according to their geographic extent.

ICFC 2001 - Networks Telecom - Conradi - page 47

Todays Network Segmentation

Long-Haul
Long-Haul transport
transport
200-600
km
200-600 km
Moving
Moving to
to 3000
3000 km
km
2.5
2.5 Gb/s
Gb/s ->
-> 10
10 Gb/s
Gb/s
Engineered
Engineered for
for
performance:
performance:
Broad
Broad band
band amplifiers
amplifiers (C
(C
++ LL bands)
bands)
Dispersion
Dispersion compensation
compensation
Optical
Optical nonlinearity
nonlinearity
management
management -LEAFfiber
LEAFfiber
Moderate
Moderate FEC
FEC

Regional Rings
200-300 km,
2.5 - 10 Gbps

Metro Rings
80-100 km,
622 Mbps - 10
Gbps

Metro
Metro Access
Access
Shorter
Shorter distances
distances -40-50
km
40-50 km
<< 622
622 Mbps
Mbps
Engineered
Engineered for
for
cost
cost::
No
No amplification
amplification
Direct
Direct mod.
mod.
transmitters
transmitters

Transparency will enlarge and merge

todays networks

Each of these network segments has significantly different

characteristics:
The long haul was characterized by distances up to 600 km but now
moving up to 3000 km, and bitrates of 10 Gb/s now under
implementation. The long haul network is the portion that requires
most of the new technologies. This part is engineered for high
performance as listed, I will come back to the details later.
Regional rings are much shorter, slightly easier to manage.
Whereas the Metro/Access portion is most cost sensitive, the costs of
terminal equipment needs a substantial cost reduction.
So each of these segments needs a different optical technology,
designed and implemented for most economic performance.

ICFC 2001 - Networks Telecom - Conradi - page 48

Enabling Technologies
Optical
Optical Amplification
Amplification
Erbium
Erbium Doped
Doped Fiber
Fiber Amplifier
Amplifier (EDFA)
(EDFA)
Distributed
Distributed Raman
Raman Amplifier
Amplifier

Wavelength
Wavelength Division
Division Multiplexing
Multiplexing
Forward
Forward Error
Error Correction
Correction
Optical
Optical Switching/Networking
Switching/Networking

These are the most enabling, recent technologies on which todays

and tomorrows networks are being built.
The Erbium doped fiber created a revolution in the late 80s. It is
almost like an ordinary fiber but in the core we put a small amount of
the element Erbium. The Erbium can be excited or pumped by
external light. If we would put mirrors at both ends, the erbium fiber
would be a laser. (Light Amplification by Stimulated Emission of
Radiation).
Recently the Raman amplifier emerged. In this amplifier very high
levels of optical power are actually sent down the transmission fiber
in the opposite direction of the signals. By a non linear effect the fiber
is converted from a lossy medium into an amplifying medium. This
adds considerable flexibility to system designers. Both types of
amplifiers (Erbium and Raman) work over a wide range of
wavelengths allowing economic application of WDM (Wavelength
Division Multiplexing).
Forward Error Correction is borrowed from radio. It enables to extend
the reach by a factor 2 to 4. This is a major advance, allowing very
long distance optical links without conversion of the light down to
electrical level.

ICFC 2001 - Networks Telecom - Conradi - page 49

Old Lightwave Systems

Transmitter

Receiver
Fiber

Transmitter

3R

3R

3R

Receiver

40-100 km

Limitations:
Speed of electronics
system capacity
u Transmitter power and fiber attenuation
transmission distance
u

Electronics speed and the speed at which lasers could be turned on

and off limits the capacity of a single channel.
A signal can travel only a limited distance, then it must be recovered
or regenerated. At this distance we need to convert the signal from
optical to electrical, re-amplify the electrical signal, reshape the digital
bits and retransmit the signal by using another laser. This is a 3R
regenerator: to Re-amplify, Re-shape and Re-transmit. The
equipment is expensive, is needed for every channel in the system
and the speed is limited by the electronics or the laser. The distances
between such stations are defined by the power of the transmitter
and the attenuation of the fiber.
Fortunately, the distance at which this needs to be done is maximized
because we operate at wavelengths where modern fibers have their
lowest loss as illustrated on the next slide.

ICFC 2001 - Networks Telecom - Conradi - page 50

Fiber Types Have Evolved to Maximize System

Performance
(SMF)

0.5

20

Attenuation
(all fiber types)

EDFA
band

10

0.4

0.3

-10
-20

0.2

NZ-DSF

0.1
1100

1200

DSF

NZ-DSF

1300
1400
1500
Wavelength (nm)

1600

Dispersion (ps/nmkm)

Attenuation (dB/km)

0.6

1700

This chart illustrates that the blue loss curve is generally valid for all
fiber types. The minimum loss is at 1550 nanometers, very
fortunately corresponding with the range of the Erbium doped fiber.
Here, for a change, nature works with us.
The green lines give the so called dispersion. Light at different
wavelengths travels down the fiber at different speeds. A difference in
wavelength of 1 nanometer, causes at a distance of 1 km a difference
in arrival time of approximately 17 picoseconds for standard single
mode fiber (SMF). This causes pulse distortion, or pulse spreading,
and definition of the signal deteriorates. This can be compensated by
sections of fiber with an inverse dispersion character.
From this one might be tempted to conclude that zero dispersion
would be beneficial. This is the case for single wavelength systems,
but for multi-wavelength or wavelength division multiplexed systems
some residual dispersion is beneficial which is illustrated by the two
curves labeled NZ-DSF which corresponds to Cornings LEAF fiber
and MetroCor fiber.

ICFC 2001 - Networks Telecom - Conradi - page 51

Multiple Wavelength Networks

1990

1992

1994

1996

1998

2000

Point-to-Point (PTP)

WDM

NBWDM

WDM PTP

WDM Ring Network

Amp

NB
WDM

NB
WDM

WADM

Amp

On top: Single Channel Point To Point (PTP)

The optical amplifier, specifically the Erbium-doped-fiber amplifier,
replaced the 3R electronic regenerator and extended the distance
between regeneration in a single channel point-to-point configuration.
Center: Wave Division Multiplex PTP
This same device also can amplify many wavelengths simultaneously
and this allows the total carrying capacity to be increased also in a
point-to-point configuration.
Below: WDM Ring
More recently other optical elements are being placed inside these
amplifiers such that we can drop and add wavelengths, and this
enables optical networking as illustrated by this ring network. In fact
this add/drop technique is an analog device.
The so called Hero result, for an experimental set-up was 256
channels at 40 Gb/s, making a total of 10 Tb/s in one fiber.

ICFC 2001 - Networks Telecom - Conradi - page 52

Amplified Fiber Systems

Transmitter

Receiver
Fiber
EDFA
EDFA
in-line
post-amp.

Advantages
High speed capability
> 40 Gbit/s
Multichannel operation
> 100 channels
Long distances without
regeneration
> 9000 km

Fiber

EDFA
in-line

Fiber

EDFA
pre-amp.

Limitations
Amplifier Noise
Fiber dispersion
Fiber nonlinearity

Here we quantify the advantages of optically amplified systems.

Also the main physical limitations are listed such as the limitation due
to the noise of EDFAs and the non-linear behavior of optical fibers at
the high optical power levels needed for long distance transmission.
However, this can be significantly improved if we use another type of
optical amplifier in conjunction with the EDFA.

ICFC 2001 - Networks Telecom - Conradi - page 53

Raman-Assisted Transmission
Discrete Fiber Amplifiers
Pb

optical
power

Q amp =

Pa

Po , ch
Nh GFB

RX

TX
Distributed Raman-Enhanced Transmission

Distributed Raman
can be treated as
an virtual lumped
amplifier

Pb

optical
power

Pa

RX

TX
Raman pump

This is the Raman Amplifier. Here, a high power laser or Raman

pump is sent in a direction opposite to that of the signal and in fact
converts the transmission fiber from a lossy fiber to an amplifying
fiber.
In effect we have added another amplifier which allows us to reduce
the power per channel launched at the transmitter, which reduces
both the overall noise and the non-linear behavior, enabling more
than doubling of the distance.
I will spare you the details of the equation, but it should be noted that
the optical power of the Raman pumps is sufficiently high that eye
safety precautions are needed.

ICFC 2001 - Networks Telecom - Conradi - page 54

High Data Rate System Schematic

Amplifier
Amplifier Span
Span == 100
100 km
km

... 40
3
2

Tx 1

Advanced Line Amplifier

Single- Amplifier

System
System Reach
Reach == 100
100 to
to 1500
1500 km
km
...40
3

Rx 1
Dynamic Single-
Dispersion Compensator

Polarization Mode
Dispersion Compensator

This shows in schematic form all the components of a high data rate
system:
Please note that today the system reach is being extended from 1500
km up to 3000 km
Single optical wavelength amplifiers are sometimes needed to boost
the signals from the laser transmitter.
The multiplexer,
The De-multiplexer
Compensators at the receiving end
This is needed because with the longer distance come other
impairments that need to be corrected - these include:
Single wavelength dispersion compensation
Dispersion in a certain way also depends on temperature. In very
long distances mountains and deserts have to be passed causing
temperature differences that require such correction.
Polarization Mode Dispersion PMD
Perfectly circular fiber would not suffer polarization dispersion.
However, it cannot be avoided that fibers are very very slightly
elliptical and travel times will be different for light entering the long
axis or the short axis of the ellipse. That has to be compensated for.

ICFC 2001 - Networks Telecom - Conradi - page 55

Advanced Line Amplifier

Broadband
Dispersion
Compensating
Module

Erbium
Pre-Amplifier
Raman Amplifier

Dynamic Gain
Flattening
Filter/Spectral
Equalizer
Erbium
Post-Amplifier

Line
Line Amplifier
Amplifier Integrated
Integrated
Electronic
Electronic Controller
Controller
Optical Performance
Monitor

So that the modern optical line amplifier now is a fairly complex piece
of equipment consisting of each of the parts shown.
Located between the Pre- and Post- Erbium amplifier are the
compensators for dispersion and wavelength dependent
amplification. Also channels can be added or dropped. So all in all
this has become a fairly complex piece of equipment.

ICFC 2001 - Networks Telecom - Conradi - page 56

Dispersion Managed Fiber/Cable (DMF/C)

1 DMF block
Splice
X

Splice
X

- D fiber

Cumulative
Dispersion

+D fiber

Distance

A very recent improvement has been to take out the special

dispersion compensating fiber that was incorporated into the line
amplifier module and make it part of the transmission fiber in the
cable.
This has two advantages 1) The loss inside the amplifier is reduced making the amplifier
simpler.
2) This fiber is actually quite effective as a Raman amplifying fiber.
From a cable standpoint it does complicate matters as signals travel
in both directions, in separate fibers in the cable, and since two fiber
types exist in a cable section, we need to be certain that the right
fibers get spliced together. Some education of craftsmen will be
required as well as proper identification of the particular fibers.

ICFC 2001 - Networks Telecom - Conradi - page 57

Forward Error Correction (FEC)

Encoder

Decoder

Tx

Rx
Signal Format

BER vs Q for R-S 255 Code (t = 8)

1.0E-03

Data
FEC Overhead

BER in
1.0E-06

No Overhead
1.0E-09
BER

Redundancy
Redundancy is
is added
added to
to the
the data
data
Allows
Allows error
error detection
detection and
and correction
correction
Typical
Typical redundancy
redundancy is
is 7%
7% or
or 23%
23%

BER out

1.0E-12

Q noise rate penalty

1.0E-15

Q FEC benefit
reduction
1.0E-18

1.0E-21
10

11

12

13

14

15

16

17

18

19

20

Q (dB)

Source:T1X1.5/99-261, Nortel
Figure 1: Example of FEC Benefit and Noise Rate Penalties

The final technology, which has been borrowed from our radio
colleagues, is a purely electronic technology called Forward Error
Connection or FEC.
System performance is measured by the so called bit/error ratio or
BER: the bits that were erroneously detected versus the total number
of bits received. The FEC system adds extra bits to the data stream
which enables errors to be detected, the concept is highly
mathematical.
One bit wrong out of a billion gives a BER of 10exp(-9).
Now if BER is plotted versus received power, we see that FEC can
give a 6 dB improvement. That is to say that if you squint hard, the
same bit error ratio can be achieved with about -6dB or a factor of 4
less power. This is huge in when traded off for additional distance
and thus extend reach. A factor four in power can achieve two times
longer distance. In principle this now enables transcontinental
distances.

ICFC 2001 - Networks Telecom - Conradi - page 58

BER vs Q for R-S 255 Code (t = 8)

1.0E-03
BER in
BER out

1.0E-06

No Overhead

BER

1.0E-09

1.0E-12

Q noise rate penalty

1.0E-15

Q FEC benefit
reduction
1.0E-18

1.0E-21
10

11

12

13

14

15

16

17

18

19

Q (dB)

Figure 1: Example of FEC Benefit and Noise Rate Penalties

Enlarged graphics of page before.

ICFC 2001 - Networks Telecom - Conradi - page 59

20

So as we eliminate electronic regeneration, which must be done on a

channel by channel basis, and replace these electronic regenerators
with next generation multichannel optical amplifiers, significant cost
savings can accrue.

ICFC 2001 - Networks Telecom - Conradi - page 60

Value of
Long Reach
2003 Time-frame
1.00
0.90

Relative cost

0.80
0.70
0.60
0.50
0.30

Electronic fabric
and Tranceivers
Optical Fabric

0.20

EDFA and Raman

0.10

Mux-Demux

0.00

10 Gb/s Equipment
and FEC

0.40

Opaque

600 km

1200 km

2500 km

Unregenerated reach

The value of this longer reach is illustrated for a US network where

we expect in the 2003 time frame to be able to transmit out to
~2500km. Significant cost savings occur, because of the elimination
of electronic regenerators, and this occurs with a slight increase in
the cost of the hybrid EDFA - Raman amplifier, as well as FEC.

ICFC 2001 - Networks Telecom - Conradi - page 61

USA Long-Distance Network

46 nodes
62 links
2.7 node degree
17 rings, 5.2 nodes
5000 km scale

Just to illustrate that this concept doesnt just apply to the USA where
the network has 46 nodes or Points of Presence connected by 62
links and a 5000 km East/West, the next slide shows Europe.

ICFC 2001 - Networks Telecom - Conradi - page 62

Pan-European Network
20 nodes
32 links
3.2 node degree
13 rings, 4.2 nodes
4000 km scale

20 nodes over 4000km, and this is not very different. An American

way of saying that Europe is not very different from the US.

ICFC 2001 - Networks Telecom - Conradi - page 63

Comparison of Network Scale

Scale
(km)

Nodes

Node
Degree

Rings

Nodes
per ring

USA

5000

46

2.7

17

5.2

Europe

4000

20

3.2

13

4.2

NE USA

1500

21

2.5

5.5

UK

700

26

2.9

13

4.2

Metro

150

24

2.5

6.2

These networks are similar except for scale.

So that the two major western regions of the world have very similar
long haul network sizes, while other network segments are similar
except for scale.

ICFC 2001 - Networks Telecom - Conradi - page 64

To increase network flexibility we can now also build optically

transparent networks by including a Wavelength Add/Drop
Multiplexer or WADM inside the Advanced Optical Amplifier to create
a WDM chain. This allows us to drop a given number of wavelengths
at their destination and allow the other wavelengths to pass through
optically. Previously this would have had to take place using an
electronic router or switch, which we have now eliminated. This
property of the all optical network is named transparency.

ICFC 2001 - Networks Telecom - Conradi - page 65

Optical Network Architecture

Two small interconnected
WDM Rings

Big WDM Ring

WDM Chain

WDM Crossconnect
Cluster

Islands of Transparency

Using WADMs and optical switches called wavelength selective

cross-connects or WSXCs, we can create a WDM chain and from
that a series of interconnected rings. At the intersections of the rings
the WSXC switches selected wavelengths that go from one ring to
another.

ICFC 2001 - Networks Telecom - Conradi - page 66

Value of Optical Routing & Protection

1.00
0.90
0.80
0.70

Relative cost

0.60
0.50
0.40
0.30
OEO Fabric

0.20

EDFA and Raman

0.10

Mux-Demux

0.00

10 Gb/s IP ports

IP/SDH/WDM

IP/WDM

10 Gb/s SDH gear

A very preliminary estimate says that the value of such optical routing
functions is more than 20% in terms of saving equipment costs.
Given time it could be much more on basis of first installed cost.

ICFC 2001 - Networks Telecom - Conradi - page 67

Value of Optical Configurability

10 Gb/s Connections
Fixed vs Configurable Network

48 days

Relative Cost

1.00
0.75

8 days

0.50
31 days

0.25
0.00

Long Path

8 days

Lost Revenue
Craft Labor
Optical Layer

Short Path

Value of fast provisioning, additional advantage.

Potentially, more important than the savings in first installed cost is
the value of being able to quickly configure optical networks with a
consequent avoidance of lost revenue due to long provisioning times
that are needed to install and configure electronic networks.

ICFC 2001 - Networks Telecom - Conradi - page 68

Summary Outcomes of Technology Evolution

Optical
Optical Amplification
Amplification
Longer
Longer distance
distance transmission
transmission with
with less
less electronics
electronics

WDM
WDM

Share
Share fiber
fiber and
and optical
optical amplifiers
amplifiers
Huge
Huge capacity
capacity increases
increases

FEC
FEC
Longer
Longer distance
distance transmission
transmission

Optical
Optical Switching
Switching
New
New lower
lower cost
cost networking
networking paradigms
paradigms
Faster
Faster more
more flexible
flexible service
service provisioning
provisioning

To summarize our technology story:

Optical amplification first allows longer distance transmission with
much less electronics,
But the amplifiers are also broadband devices so that they really are
the enabler of Wavelength Division Multiplexed transmission which
permits huge increases in system capacity and reduced cost through
equipment sharing.
Raman amplifiers and FEC then further extend the distances to
continental dimensions which when combined with
Optical Switching further lowers network costs and enables future
flexible networks with faster provisioning times.
Finally:

ICFC 2001 - Networks Telecom - Conradi - page 69

Lest We Become Complacent

Need for Optical & Electronic Integration

If we are not too complacent about technology reaching an eventual

limit, without exaggeration one could state we are at a stage
comparable to the first discrete transistors of the 1960s. It took a
little while to reach the integrated circuit with millions of transistors on
a single chip. At this stage we are nowhere near integrating optical
and electrical components on a chip, comparable to todays electronic
integrated circuit with its millions of devices.
Hopefully this technology evolution can be achieved.
Thank you for your attention.

ICFC 2001 - Networks Telecom - Conradi - page 70