You are on page 1of 4

ICTON 2009 Mo.B4.

The Market, Rationale and Technology Options


for Flexible Transparent Optical Networks
Richard Dorward, Fellow, IET
Ericsson Ltd, United Kingdom
Tel: + 44 1483 303 666, e-mail: richard.dorward@ericsson.com
ABSTRACT
This paper outlines the market for transparent optical networks and how these have developed, from their simple
origins of saving the costs of OEO regeneration in point-to-point WDM links, to today’s ring and mesh networks
with flexible/reconfigurable nodal traffic access, using multi-way ROADM’s. The paper then reviews and
compares the technology options for multi-way ROADM’s and the impact that full optical transparency has on
the overall network design. The paper also considers alternative approaches, using OEO conversion and
electrical switching, and suggests where these or transparent optical switching may be most applicable in an
increasingly cost-sensitive environment.
Keywords: WDM, ROADM, OXC, WSS.

1. INTRODUCTION
The recent rapid increase in service demand, particularly for high bandwidth video services such as YouTube,
Facebook, BBC iPlayer, video downloads and the emerging video-on demand and “broadcast” IP TV, as well as
for higher speed mobile broadband, is presenting operators with an unprecedented demand for network
bandwidth in all areas of the network infrastructure. The business challenge is to deliver these services at the
competitive prices demanded by the market. As well as the dramatic increase in capacity requirements, operators
are facing additional challenges due to the changing service mix that their infrastructure has to support: all areas
of the network are moving away from traditional TDM services to a more diverse mix of packet orientated
services.
Fixed DWDM systems have historically been very effective for delivering bulk capacity in core and metro
networks but the lack of service level flexibility has often resulted in less than optimum costs as network
capacity becomes stranded or where it prevents operators from reacting to rapid changes in customer demand.
Flexible transparent DWDM networks offer the potential to provide both the required bulk capacity and
service flexibility in an economic fashion. However, the technology for transparent optical switching is not yet
mature and there are still a range of competing options, some even electronic, that need to be evaluated as this
market evolves.

2. THE MARKET FOR FLEXIBLE TRANSPARENT OPTICAL NETWORKS


Since 2003, when the market stabilised Actual spending Most likely forecast
after the bursting of the telecoms bubble,
$30,000
the optical networks equipment market has
grown steadily, nearly doubling to a global $25,000
value of over US$ 16 billion in 2008. With
USD millions

$20,000
the current financial crisis, the rate of
growth was less in 2008 and the overall $15,000
market is now predicted to show a plateau $10,000
for a few years before starting to rise
sharply again [1]. $5,000
WDM equipment effectively constitutes $0
the transparent transport part of this market 2006 2007 2008 2009 2010 2011 2012 2013 2014
and makes up nearly 50% of the total value. Source: Ovum
Historically, WDM has enjoyed a higher
growth rate than the other optical segments Figure 1. Global optical networks forecast: 2006-2014
with even higher growth rates forecasts for
the future, particularly in the Metro WDM market segment, where there are the largest demands for flexibility.
Until recently this flexibility has been separately identified as a ROADM (Re-configurable Optical Add/Drop
Multiplexer) market segment, but some analysts have stopped using the ROADM terminology in their
segmentation as they expect all today’s Metro WDM networks to contain ROADM functionality. In addition,
a significant proportion of core or backbone WDM networks also contain transparent remote re-configurability.
From this, a reasonable estimate of the current market for transparent optical networks incorporating transparent
optical flexibility might be around US$ 5 billion per annum over the next few years.

978-1-4244-4826-5/09/$25.00 ©2009 IEEE 1


ICTON 2009 Mo.B4.4

3. THE RATIONALE AND EVOLUTION OF WDM TECHNOLOGY AND NETWORK FLEXIBILITY


The concept of using WDM to carry multiple different communications channels over a common fibre can be
traced back to the 1960’s, but it was not until the invention of the erbium doped fibre amplifier (EDFA) in 1987
[2] and the growth in channel capacity that drove up the cost of the electronics required for regeneration that the
economic benefits of WDM really began to appear. Even then, the main driver for WDM systems came
primarily from long haul systems, particularly with leased fibre in the USA and in undersea applications, where
the cost of fibre was high. Nevertheless, even where the cost of fibre is not a consideration, the main rationale for
WDM transport is the lower cost of transparent multi-channel amplification compared to multiple regenerators.
In the late 1990’s, when 10 Gbit/s systems began to be deployed widely, the economics of WDM amplification
started to pay in over regeneration for anything over two channels, and this balance is still largely true today.
The other key rationale for non-regenerative transport is the transparency, or format independence, that this
affords, permitting service flexibility and channel capacity upgrades merely by changing individual terminal
equipments. Only very recently, with the drive to provide 100 Gbit/s channels on a WDM grid with only 50 GHz
channel spacing, has this flexibility hit some severe limits and required more complex modulation schemes to
achieve the channel capacity. However, WDM amplification is still favoured because of the even higher cost of
the more complex regeneration required.
The idea of using WDM to provide a range of network configurations based on wavelength routing and the
inclusion of flexible switching elements was first proposed as early as 1988 by British Telecom [3]. This concept
was developed and demonstrated in the RACE II Multi-Wavelength Transport Network (MWTN) project in the
mid 1990’s, in which Ericsson was one of the project partners, although, again, economic viability had to wait
another five or ten years.
In the meantime, WDM systems technology had developed to provide a degree of fixed (or manually
configurable) traffic routing using fixed add/drop filters and this is still the lowest cost solution today, as
epitomised by the short haul or campus, (typically) CWDM, passive systems. Various add/drop configurations
exist, ranging from the lowest cost single channel serial add/drop filter, to the full demultiplex/multiplex filter
combination, with its capability to manually configure the adding and dropping of individual channels without
affecting any of the remaining ones.
With flexible remote re-configurability, the requirement to avoid disturbing any of the other channels is
a given and so the architecture is essentially the same full demultiplex/multiplex one, but with the addition of
some suitable intervening switching element. The key to the economic realisation of this was some level of
optical integration to minimise the otherwise huge number of discrete components and fibre interconnections
required to handle a realistic number of WDM channels. The first commercial deployment was with the Corning
Wavelength Blocker used in the Ericsson Marconi PMA-32 ROADM in early 2001 [4]. Even then, most
operators were content with the limited manual configurability of fixed add/drop filters, but the same free space
optical integration and equipment architecture pioneered in that product have since been widely employed and
have evolved directly to today’s Multi-Wavelength Selective Switch (WSS) technology and node architectures.
In parallel with this ROADM/WSS evolution, a number of alternative wavelength switching technologies
and node architectures have also been developed, primarily focussed on the need for multi-way transparent
nodes, whereas the original target of the ROADM was solely ring network based. Together with the
development of different switching element technologies within the ROADM/WSS architecture itself [5], these
provide a wide range of levels of optical integration and optical switching capabilities. The following sections
explore the pros and cons of these from both a technology and cost perspective.

4. TECHNOLOGY OPTIONS AND DEGREES OF TRANSPARENCY


For full transparent network flexibility, multi-way nodes that can operate in a mesh-connected network are
needed. In such nodes, there are two underlying node architectures, the first based on the classical cross-connect
architecture with its protected central switch and WDM extensions, and the second based on the 2-port
wavelength blocker ROADM architecture and its extensions. To/from other ports

4.1 Photonic cross-connect architectures


Tx/Rx Tx/Rx
The photonic cross-connect architecture is the simplest in WDM WDM
concept, with an N × N optical matrix switch providing the port 1 Optical
port 2
most obvious equivalent of the electrical switch matrix, matrix
although, in the optical case, the requirement to protect the switch
central switch matrix adds much more significant cost and
Mux /
complexity (primarily in the additional level of switching for Demux
Matrix protection
protection and in the ensuing additional loss and multiplicity switches
of fibre connections to the switch matrix) than in the
electrical case. Transponders
Figure 2. Matrix switch architecture

2
ICTON 2009 Mo.B4.4

Undoubtedly, this architecture provides the full level of port-to-port connectivity and transponder access
flexibility required, but it also has serious implementation issues. Firstly, the highest cost element, the duplicated
switch core, has to be provisioned even for a minimal initial level of equipping, so it is not attractive where the
traffic growth is uncertain. Then there is the major issue of the number of ports that can be economically
delivered. With today’s maximum optical matrix sizes of 160 – 320 ports, and the lack of integration of the
WDM multiplexing and demultiplexing function, this means that optical cross-connect implementations, even of
this size, intrinsically require over a thousand fibre connections.
The scalability, both for initial low level equipping and for the maximum size required, can be addressed by
providing a number of parallel switch planes, each associated with a fraction of the total WDM band. In the
extreme case, this becomes the ”waveplane” architecture, with one small, multi-way switch for each wavelength.
Because of the transparent continuity of each wavelength through the node, this imposes no restrictions on the
functionality. However, it does not reduce the major problem of the enormous number of fibre connections, and
the modularisation to provide efficient scalability also adds cost and complexity in itself. The end result of these
issues is that, to date, there have been several product designs based on this type of technology and architecture,
but little widescale commercial deployments.
4.2 ROADM/WSS architectures
To / from other ports
In contrast to the photonic cross-connect
architecture, the basic 2-way ROADM architecture,
using either wavelength blockers or wavelength WSS unit

x x x x x x
selective switches, has a relatively small number of
fibre connections, determined predominantly by the Tx/Rx Tx/Rx
number of transponders fitted, as the “through” paths WDM WDM

x x x x x x
consist only of WDM connections. While the port 1 port 2
architecture scales badly with the number (M) of
WDM ports in the wavelength blocker case (~ as WSS unit
M2), the use of multi-port WSS modules allows it to
scale linearly up to the fan-in/out of the modules
Mux /
(typically 8 – 10 today, moving towards 20 within Demux
the next year). Combined with modules offering up
to over 80 wavelengths, the WSS architecture offers
Transponders Transponders
incremental equipping with number of WDM ports
and simple scaling to the equivalent of over 1000 Figure 3. Multi-way WSS ROADM architecture
channels.
The limitations of this architecture lie in the flexibility, or lack of it, in the transponder access provided.
Firstly, in the simple case shown, the transponder access is always associated with each WDM line port. This
may be perfectly acceptable in a 2-port ROADM, but it imposes a severe limitation on the directional routing
that can be provided in a multi-way node. Also, the cheapest transponder access mux/demux elements are simply
fixed filters, which means that the transponders have to be provisioned in fixed physical positions associated
with the wavelengths (or “colours”), or additional manual patching needs to be provided.
Neither of these is acceptable for the full, remotely reconfigurable, flexibility necessary to support dynamic
traffic routing and restoration. The simple architecture needs to be extended with additional/cascaded WSS
modules, with their attendant penalties in increased cost and optical loss, to achieve the same “directionless” and
“colourless” flexibility as the optical cross-connect architecture. The result is to effectively increase the number
of WSS modules per WDM port from one to three [6], with the possibility of additional intermediate optical
amplification to maintain the necessary signal levels and optical signal to noise ratio (OSNR). Even under these
circumstances, today’s costs still heavily favour the ROADM/WSS architecture, which is why the vast majority
of “Tier 1” component and equipment vendors are following this approach, and several tens of thousands of
ROADM/WSS nodes of various sorts have been commercially deployed.
4.3 Alternative switching technologies and architectures
Today the prime challenge to the ROADM/WSS architecture comes, not from other photonic switching
technologies, but from a return to electronic switching. With the advent of photonic integration, the prime
rationale for transparent optical transport has been called into question: if the OEO conversion function can be
cost reduced sufficiently, then the complexities and limitations of optical switching and transparent transport can
be avoided. So far, only one vendor has achieved the necessary scale of photonic integration [7] to provide
a viable product, and that is in a vertically integrated company and product environment. Even here, the prime
rationale has not been violated as EDFA’s are still used extensively for intermediate-node amplification, and the
benefits are predominantly of a second order in the transport and exist mainly in the costs and functionality
provided by the OEO conversion and electronic switching. Essentially, the use of full OEO conversion at each
switching node removes much of the optical levelling and control needed to maintain optimum performance in

3
ICTON 2009 Mo.B4.4

an extended transparent transport link, and the availability of the signals in electrical form permits full
performance monitoring and wavelength translation, where needed, at every switching point.
Unfortunately, these benefits come at the cost of the other key rationale for non-regenerative transport,
namely the transparency that permits service flexibility and channel capacity upgrades merely by changing
individual terminal equipments. In the five or six years since the original announcements, this approach has
achieved significant commercial success and market component vendors are now promising similar capabilities,
albeit with a lower level of integration, which might raise questions about the available yields and costs of the
photonic integration technology. Furthermore, the associated electronic switch technology is a major issue for
scalability and upgradeability: to match the capacity of the photonic switching available today would require an
electrical switch approaching 100 Tbit/s in size [8]. Even estimates of the 10-year evolution of the technology to
achieve such a switch lead to a power consumption of over 50 kW [9], which is at least two orders of magnitude
higher than in the photonic case today, although the electronic switch does have the possibility of finer than
wavelength granularity.
Taking the best electronic “cross-bar” 45
switching technology and the photonic integration WSS low end
40
technology promised on the open market today, WSS high end
PIC/OEO low end
comparisons can be drawn with the dominant 35
PIC/OEO high end
ROADM/WSS technology. A cost model 30

Normalised cost
comparison at 10 Gbit/s per wavelength shows
25
a range of cross-over points with ROADM/WSS
technology, depending on the exact WSS 20

functionality and photonic integration costs 15


assumed, so this appears to be the best area for the
10
OEO and electronic switching paradigm. (Higher
bit rates increase the electronic switching cost 5
significantly with no impact on the ROADM case.) 0
This suggests that the optimum application area for 0 2 4 6 8 10
this technology may be in Metro and Edge 25% add/drop channel count
networks where 10 Gbit/s per wavelength is the
expected level of capacity demand in the more Figure 4. WSS and PIC/OEO cost model
immediate future.

5. SUMMARY AND CONCLUSIONS


The market for flexible transparent optical networks has grown to around one third of the total optical networks
market and seems likely to maintain that proportion, even in the present market downturn. In this market, the
predominant approach to flexible and multi-way node architectures, today, is based on multi-way ROADM and
WSS technology. Alternatives based on photonic integration for OEO conversion and electronic switching at
10 Gbit/s per wavelength show some operational benefits and are likely to be most applicable in Metro and Edge
DWDM networks, where this is the expected level of capacity demand. Whereas, transparent photonic
switching, as provided by ROADM and WSS technology, is capable today of supporting the 40 and 100 Gbit/s
per wavelength channels that Metro Core and Core network systems demand.

REFERENCES
[1] I. Redpath: Optical networks volumes and revenue history and forecast, 2006-2014, Ovum Market
Research Report, March 2009.
[2] R.J. Mears, L. Reekie, I.M. Jauncey and D.N. Payne: Low-noise Erbium-doped fibre amplifier at 1.54 μm,
Electronics Letters, 1987, 23, pp.1026-1028
[3] G.R. Hill: A wavelength routing approach to optical communications networks, IEEE INFOCOM, New
Orleans, USA, March 1988.
[4] Marconi Communications: Marconi announces further European success winning contract to provide
completely new optical network solution to Telfort, Marconi press release, Pittsburgh, USA, September
2000.
[5] P. Wall, et al: WSS switching engine technologies, OFC/NFOEC, San Diego, CA, USA, February 2008
[6] Ericsson AB: Switch Node, International Patent Application No. PCT/EP2008/060893, August 2008.
[7] D. Welch, et al.: Performance results of optical transport system using 100Gb/s photonics integrated
circuits, ECOC 2004, Stockholm, Sweden, Sept. 2004
[8] J.D. Ash and R.M. Dorward: Market trends in optical networks – The service drivers and technology
impact, ICTON, Athens, Greece, June 2008.
[9] S. Aleksić: Power consumption issues in future high-performance switches and routers, ICTON, Athens,
Greece, June 2008.

You might also like