See discussions, stats, and author profiles for this publication at: [Link]

net/publication/224124873

Colorless, Directionless and Contentionless multi-degree ROADM architecture

for mesh optical networks

Conference Paper · February 2010

DOI: 10.1109/COMSNETS.2010.5431987 · Source: IEEE Xplore

CITATIONS READS

33 13,311

7 authors, including:

Gowrishankar Ramadurai Sai Kishore

Sri Sathya Sai Institute of Higher Learning Cisco Systems, Inc
117 PUBLICATIONS 317 CITATIONS 8 PUBLICATIONS 62 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Gowrishankar Ramadurai on 07 January 2017.

The user has requested enhancement of the downloaded file.

 Colorless, Directionless and Contentionless
multi-degree ROADM architecture for mesh
optical networks
Devarajan A, Sandesha K, Gowrishankar R, Sai Kishore B, Prasanna G, Ron Johnson and Pavan
Voruganti

Therefore, service providers and cable operators are looking

Abstract—A Colorless and Directionless (C&D) ROADM for next-gen optical networks that support remote
architecture has empowered today’s networks with the much reconfiguration of network resources by an optical bridge-and-
needed flexibility and scalability to handle the unpredictable high roll during the evolution of the network.
bandwidth demands, provide any-to-any connectivity and
At the heart of this next-generation agile optical network is
provision new services without disrupting the existing ones. In
this paper, we realize the C&D ROADM architectures using PXC the capability of reconfiguration which provides high
and WSS at a multi-degree node. We then evaluate both the bandwidth connectivity to any network site without
architectures based on the various advantages offered by them. interrupting the existing services or re-engineering the
We also discuss some of the drawbacks of the two architectures. network. Such functionality is offered by the remotely
Subsequently, we propose a Colorless, Directionless and Reconfigurable Optical Add/Drop Multiplexer (ROADM).
Contentionless (CD&C) architecture which combines the features
Integration of ROADMs into the present WDM networks is of
offered by both the C&D architectures. Finally, we discuss
practical implementation issues of the proposed architecture. critical value as they provide the pay-as-you-grow capability
and flexibility to provision wavelengths dynamically. In
Index Terms—Colorless Directionless and Contentionless addition, ROADMs enable transparent ring-to-ring
architecture, photonic cross connect (PXC), reconfigurable interconnection between metro ring networks that will
optical add/drop multiplexer (ROADM), wavelength selective eliminate the need for optoelectronic regeneration. As a result,
switch (WSS). the network operators can now move beyond simple ring
configurations to fully meshed optical networks with redefined
network resiliency and flexibility to deliver emerging services.
I. INTRODUCTION
Most of the ROADMs that are available today support a wide

T HE rapid growth of residential and business broadband

services is driving the service providers to look for means
to generate new revenue streams and increase their
variety of features such as colorless add/drop, dynamic
channel equalization (DCE) & power monitoring and multi-
degree interconnection. These valuable features make the
competitive advantage. Many of these services like video on ROADMs the ideal choice to realize agile “all-optical”
demand, voice over IP (VoIP), high-definition television networks.
(HDTV), etc. are unpredictable in terms of bandwidth demand Along with the deployment of ROADM, the choice of an
and traffic routing patterns, in addition to their stringent ideal architecture also plays a very crucial role in realizing a
requirement for a high level of quality of service. This poses a completely agile, flexible and dynamic optical network. Such
new operational challenge for service providers to effectively an agile optical mesh network calls for an architecture that
forecast bandwidth requirements at various sites spread across accelerates service deployment and enable intelligent
the network. Alternatively, provisioning high bandwidth wavelength applications like dynamic wavelength routing and
connectivity to every single site results in a significant elimination of O-E-O operations. These requirements are
increase in the CAPEX and the network complexity. satisfied by a Colorless and Directionless architecture. In this
paper, we describe two different implementations of the
Manuscript received September 23, 2009. Colorless and Directionless ROADM architecture – PXC
Devarajan A, Sandesha K and Gowrishankar R are with the Sri Satya Sai based and WSS based. The PXC based Colorless and
University, Prashanthi Nilayam, Puttaparthi, A.P., India 515134 (e-mail:
deva612@[Link]; sandeshak@[Link]; rgshankar76@[Link]). Directionless ROADM architecture is realized using 3-D
Sai Kishore B and Prasanna G are with Cisco Systems Inc., Bengalooru, MEMS based PXC and the Wavelength Selective Switch
India 560087(e-mail: sbhyri@[Link]; praganes@[Link]). based ROADM is realized using WSS devices in combination
Ron Johnson is with Cisco Systems Inc., San Jose, CA 95134 USA (e-
mail: ronjohn@[Link]).
with multiport splitter/combiners [1]. We discuss the specific
Pavan Voruganti is with Vello Systems Inc., Milpitas, CA 95035 USA (e- advantages offered by both the architectures and also their
mail: pavan@[Link]). drawbacks. Further, we propose a Colorless, Directionless and
Contentionless architecture which combines the features of
978-1-4244-5489-1/10/ $26.00 © 2010 IEEE
both the PXC and the WSS based ROADM. Subsequently, we
discuss the advantages of this architecture and also some of
the practical implementation issues such as performance
analysis in terms of OSNR & single channel power variation
along a path, choice of subsystem technologies, cost
considerations and the other challenges.

II. COLORLESS AND DIRECTIONLESS ROADM

ARCHITECTURES
ROADMs are crucial network elements that play a
significant role in building next generation reconfigurable
mesh optical networks. In addition to the tremendous benefits
offered by agile ROADM-driven optical systems, the
flexibility of the mesh network can be further extended by
providing Colorless and Directionless (C&D) feature to
add/drop ports at every node in the network. The colorless
feature at a local add/drop port refers to the ability of the
tunable transponders to have wavelength transparent access to Fig. 1. A sample mesh optical network
all the DWDM network ports. The directionless feature at a add/drop can directly access the core switch to guide any
local add/drop port refers to the ability of the transponders to individual wavelengths from/to the desired input/output fibers
have non-blocking access to all the DWDM network ports in due to colorless and non-blocking feature of the 3D-MEMS
the optical system. The C&D feature of the ROADM based core switch. The outgoing DWDM signals from the
architecture enables efficient use of all the tunable access local add sites and from the express paths (light paths that do
resources for on-demand and high bandwidth network not go to the drop ports at a node) are multiplexed using AWG
configuration and simplify the network architecture. In this based 40 channel multiplexers which are integrated into the
section we describe Photonic Cross Connect (PXC) based fiber block arrays at the output fiber ports. Thus, the PXC
C&D ROADM and Wavelength Selective Switch (WSS) based ROADM architecture at the 4-degree node provides a
based C&D ROADM architectures at a multi-degree node in Colorless and Directionless feature to the local add/drop site.
mesh optical networks. The main advantage of the PXC based C&D ROADM
A. PXC based Colorless and Directionless ROADM architecture at a multi-degree node is the efficient utilization
Consider a sample mesh optical network consisting of 7 of wavelengths. Due to non-blocking and colorless capability
nodes as shown in Fig. 1. The flexibility and the dynamic of the MEMS based large optical switch, same wavelengths
reconfigurability requirements in the network can be achieved carrying different information can be received/sent
by using a PXC based C&D ROADM architecture at every simultaneously with the help of multiple tunable transponders
node. from/to different input/output fiber ports. This Contentionless
Consider node A, a 4-degree node in the sample network. feature at the local add/drop site offered by the PXC based
The C&D ROADM architecture at node A consists of a three- C&D ROADM enables an efficient usage of wavelengths and
dimensional microelectromechanical system (3-D MEMS)-
based PXC in combination with a set of wavelength-selective
multiplexers and demultiplexers (Mux/DeMuxs) as shown in
Fig 2. With their low loss, optical transparency, and large-
scale capability, 3-D MEMS-based PXCs have become
imperative to next-generation transparent networks.
We have considered a 320x320 non-blocking 3-D MEMS
large optical switch at the core of the PXC that has 8
input/output fiber ports with each fiber port carrying a
maximum capacity of 40 channels [2]. An incoming DWDM
signal from any of the four directions is demultiplexed into
granular wavelengths using an array-waveguide grating
(AWG) based 40 channel demultiplexer which is integrated
directly into the fiber block arrays at the input fiber ports.
With the help of 3D-MEMS optical switch, a low-loss light
path of any individual wavelengths from each of the fiber
ports can be established to the desired fiber outputs or the
Fig. 2. PXC based Colorless and Directionless Architecture at a 4-degree
local drop ports [3]. The tunable transponders at the local node
access resources [4]. Also, the architecture can provide 100% switching time less than 10ms [8] are available. Another
add/drop with 320 tunable receivers and 320 tunable limitation of the PXC based C&D architecture is that it does
transmitters at the local add/drop site. However, depending on not support multicasting as it cannot drop a wavelength from
the number of wavelengths to be dropped/added, service given direction and send the same wavelength simultaneously
providers can afford a modular growth in the number of along the express path to other directions. A major drawback
transponders at the local add/drop site. Another important of the PXC based C&D ROADM architecture is that the core
advantage of the PXC based C&D ROADM architecture is switching fabric acts as a single point of failure (SPOF) that
that the loss suffered by the add/drop services at all the multi- can bring down the entire node and all the services which are
degree nodes is minimal. The 320x320 MEMS based PXC passing through it. A solution to such a situation is the usage
have a measured loss of less than 2.9dB with a median loss of of another identical core optical switch for providing node
1.4dB for all the possible 102,400 light paths [2]. Also the protection as shown in Fig. 3.
320x320 MEMS based large optical switch has a better The secondary core is kept active instead of redundant, to
wavelength dependent loss performance (less than 0.5dB avoid the challenges faced while switching the entire services
variation in the entire range of 1250nm–1650nm), low through core A to core B within the protection time in case of
polarization dependent loss PDL (less than 0.1dB at 1550nm), any failure in the core A. It is done with the help of 1x2 power
less chromatic dispersion CD (less than 0.5ps/nm) & splitter (S) that duplicates the incoming DWDM signal from
polarization-mode dispersion PMD (less than 10fs) and each directions and form as inputs to corresponding DeMuxs
minimal crosstalk (less than -60dB) [2]. at both of the switching cores. Each of the local add/drop ports
One of the major concerns with the PXC based C&D of the two switching cores are connected to a Protection
ROADM architecture is that as the number of ports N, in NxN Switching Card (PSC) which performs the functions of a
core switch grows the number of possible connections 2x1Protection Switch (PS) and also a passive splitter (S).The
increases as N2. However, with the MEMS technology splitter duplicates the signal from any transmitter to form
emerging as an important candidate for building high port- inputs at each of the local add ports of the two cores. The
count PXC switches suitable for deployment in core-transport multiplexed signal from the respective fiber output ports of the
networks [5], it is possible to realize PXCs of more than 1000 two core switches are given as inputs to a 2x1 PS that chooses
ports [6]. Also, MEMS based PXCs having an insertion loss of the best of the signal from the two core switching matrices. At
2.1dB on the average, 4dB maximum [7] and connection the local add/drop site, the protection is provided per

Fig. 3. Protection scheme employed with dual core switching matrix in PXC based C&D ROADM architecture
transponder to ensure an efficient protection scheme and also two WSSs to provide local drop at the node A. The rest of the
to have a modular growth in the number of PSCs used ports are reserved for scaling in the future. The WSS at the
depending on the number of add/drops. drop site consolidates the wavelengths from the different
directions. The wavelengths are then split using a 1x8 splitter
B. WSS based Colorless and Directionless ROADM
that provides an expansion facility since the number of
Wavelength Selective Switch (WSS) is the latest step in the wavelengths to be received may exceed the number of
evolution of ROADM technology. It is an advanced fiber optic available ports on the WSS. Subsequently, a second WSS at
module that can dynamically select individual wavelengths the add/drop site provides a colorless add/drop across its ports.
from multiple DWDM input fibers and switch these to a In order to realize the colorless add feature, again WSS is
common output fiber. The innovation of the WSS is that it can used to consolidate all the wavelengths that originate at node
select and switch individual wavelengths on a A. Since any wavelength can be added at any port of the WSS,
multiwavelength fiber in the optical domain without expensive it becomes a colorless add. An 8x1 combiner then provides
electronic conversion. An important feature of a WSS is that it expansion similar to the splitter on the drop side. A splitter
provides Dynamic Channel Equalization (DCE) by attenuating then splits it 8 ways providing connections to all the
individual wavelengths exiting the output fiber, thereby directions. A WSS is present along each direction, which
equalizing optical power among the wavelengths to maximize either blocks or allows a certain wavelength along that
overall performance. direction. Thus a wavelength that is added at A can be sent
Fig. 4 shows the Colorless and Directionless architecture along any direction, hence realizing a directionless feature.
using 1x9 WSS modules and multi-port splitter combiners at a The WSS is a better choice at the add site as it offers many
4-degree node A in the sample mesh optical network (Fig. 1) intelligent functionalities. For example, a transient wavelength
[1]. The wavelengths coming from all the four directions or power drift in an add channel may cause adverse effects on
encounter a splitter that provides a 1x8 power-split. While the other channels like severe cross talk or even complete
three of the eight ports in the splitter are connected to the WSS attenuation. Since the WSS provides channel specific blocking
along the other three directions, two ports are connected to the feature, the drifting channel alone can be blocked. If a

Fig. 4. WSS based Colorless and Directionless Architecture at a 4-degree node

combiner were to be used instead, it cannot provide such WSS modules at the add/drop sites, but at the expense of
functionality as it is a passive device. increased Capital Expenditure.
The WSS based Colorless and Directionless ROADM
architecture offers other valuable advantages too. All the
wavelengths entering node A from any directions can be III. COLORLESS, DIRECTIONLESS AND CONTENTIONLESS
routed to any other directions using the splitter, while ROADM ARCHITECTURE
simultaneously making all of them available at the local drop An ideal Colorless and Directionless ROADM architecture
of node A. Also, the wavelengths that are added at the local should provide certain add/drop flexibility features like
node A can be routed in any direction as again a splitter is Colorless, Directionless and Contentionless add/drop, modular
provided at the add site. Thus any wavelength that is either growth, etc that enable efficient utilization of resources.
coming in from any direction or originating at that node can be Though the PXC based Colorless and Directionless ROADM
multicast to all the directions. Another important advantage of architecture provides Contentionless add/drop, it does not
the WSS based ROADM is that it offers the pay as you grow support multicast of wavelengths. The WSS based Colorless
approach as it is truly scalable in terms of both the node- and Directionless ROADM architecture addresses scalability
degree and the number of add/drop wavelengths. This enables more efficiently and also supports multicast of wavelengths.
the service providers to reduce the initial CAPEX by deferring However, it does not provide Contentionless add/drop.
the roll-out of as much equipment as possible, while allowing A Colorless, Directionless and Contentionless (CD&C)
modular growth as the network grows over various phases. ROADM architecture can be realized by combining the WSS
The number of 1x5, 1x9 or 1x20 WSS modules required at a based ROADMs with PXC as shown in Fig.7. This
4-degree node with increasing number of add/drop architecture is similar to the WSS based C&D ROADM
wavelengths per direction is shown in Fig. 5. architecture, except the design at the local add/drop site. In
The WSS based Colorless and Directionless ROADM this CD&C ROADM architecture, the WSS and the multi-port
architecture however does not provide Contentionless splitter combiners at the add/drop location are replaced by the
add/drop. Consider the case where two different services, one PXC. All the wavelengths coming along every direction
originating at G and the other originating at E, have a common encounter a 1x8 splitter. Three ports of the splitter are
end-point, node B. In case of a fixed architecture, this would connected to the WSS along the other three directions, while
lead to a contention and would require wavelength conversion. the fourth port is connected to one of the input ports of the
However, since two sets of drop modules are present, the two PXC placed at the local add/drop site. The output ports of
services on the same wavelength coming from two different PXC are connected to the WSS along the corresponding
directions can be received using the two different WSS directions.
modules at the drop site of node B as shown in Fig. 6. The Colorless, Directionless and Contentionless ROADM
Consider a third service on the same wavelength from a architecture not only offers all the features that are offered by
third direction, say from node A, which needs to be dropped at both the PXC and the WSS based ROADM solutions, it also
node B. This is not possible as the design at the local add/drop addresses their inherent deficiencies. There are multiple
allows a maximum of two directions from which the same advantages of this CD&C ROADM architecture as the
wavelength can be received. Hence, this would lead to a architecture supports multicast of wavelengths coming from
contention at the drop site of node B, resulting in loss of any direction and also those that are added at the local node A.
information. This problem can be sorted out by utilizing more

Fig. 5. WSS counts for a 4-degree node Fig. 6. Drop module at node B in the sample network.
The PXC which is used for the local add/drop provides the placement of amplifiers of suitable gain values, each of the
Contentionless add/drop access in addition to Colorless and systems work well within the necessary power and OSNR
Directionless add/drop with a provision for the modular limitations and also it shows that the proposed CD&C
growth of the tunable transponders depending on the number ROADM architecture has a better OSNR performance while
of channels to be dropped/added. This results in a better being economical in the usage of amplifiers. The 3D MEMS
wavelength utilization as the same wavelengths carrying technology is not the only choice for the sub-system of PXCs
different information can now be sent to different directions. at the local add/drop site of the proposed architecture. There
Also, the optical power loss for the add/drop wavelengths is are other competing technologies with associated pros and
significantly lesser compared to the WSS based ROADM cons, which are then discussed briefly. Finally, the major
architecture. As a result of this, lesser gain of amplification is factor that determines the feasibility of implementing any
required at various amplifiers placed at each node in the architecture is cost which is also addressed.
network. Another advantage is that this CD&C architecture
A. Performance analysis
completely eliminates the need for a number of WSS modules
and splitter combiners at the local add/drop site of every node. The power budgeting for the sample mesh network under
This results in a drastic reduction in the service providers’ consideration was performed and the OSNR was calculated
CAPEX. along the different links, wherein the parameters for the
components were based on industry standards [9]. The
distances between the nodes in the sample mesh network were
IV. PRACTICAL IMPLEMENTATION ISSUES assumed to be 80km. The launch power of a 10Gbps tunable
transponder was taken to be 1.5dBm for the case of PXC
In this section, we consider a few issues in practically
based C&D ROADM architecture; 4.5dBm for the case of
implementing the proposed Colorless, Directionless and
WSS based C&D ROADM architecture and 4dBm for the case
Contentionless ROADM architecture. We first present the
of CD&C ROADM architecture. We have varied the
performance analysis for PXC based and WSS based C&D
launching powers of 10Gbps transponders for different
ROADM architectures and then for the proposed CD&C
architectures to keep the OSNR levels within the specified
ROADM architecture. It shows that, with the proper

Fig. 7. Colorless, Directionless and Contentionless ROADM architecture at a 4-degree node

limits. The minimum permissible value of the power level, at gains 20dB (with 7.6dB noise figure) and 16dB (with 12.0dB
the receiver end is -26dBm (lower threshold of APD receiver noise figure) at suitable positions in each node, the various
sensitivity). losses were compensated.
For the case of PXC based C&D ROADM architecture (Fig. The single channel power variations along all the available
2 & Fig. 3), every wavelength added from a local add site of paths in the sample mesh optical network (Fig. 1) were
each node has to pass through a protection switch card PSC determined for the three ROADM architectures and for one of
(3.5dB loss), 320x320 MEMS switching core (2.5dB loss), a the paths A-E-F-B, the power variations are as shown in Figs.
40 channel multiplexer (4dB loss) and a protection switch PS 8, 9 and 10. The figures also show the EDFA sites at various
(3.5dB loss). The pass-through channels are sent through a locations for the service along the path A-E-F-B. It can be
1x2 power splitter (3dB loss), a 40 channel demultiplexer, observed that the minimum power requirement at the APD
320x320 MEMS switching core, a 40 channel multiplexer and receivers (-26dBm, lower threshold for APD receiver
a protection switch which. The wavelengths that are to be sensitivity) is well satisfied by all the three ROADM
dropped must pass through a 1x2 power splitter, a 40 channel architectures considered in this work.
demultiplexer, 320x320 MEMS switching core and a The optical signal-to-noise ratio, OSNR values at amplifiers
protection switch. Thus, each channel suffers a total loss of along the same path (A-E-F-B) for the three ROADM
13.5dB at the local add site of each node, a lump loss of 17dB architectures are shown in Fig. 11. It can be noted from the
along the express path at each node, a fiber loss of 20dB for Fig. 11 that the OSNR performance of all the three ROADM
each fiber link of 80km in the sample network (fiber loss = architectures are well within the minimum OSNR requirement
0.25dB/km) and a total loss of 13dB at the local drop site of of 9dB. Also, it can be observed that the proposed CD&C
each node in the sample mesh optical network. In our analysis, ROADM architecture has a better OSNR performance
24dB amplifiers (with 6.5dB noise figure) and 13dB
amplifiers (with 16.9dB noise figure) were placed at suitable
positions at each node to compensate for the various power
losses. Also, a dispersion compensation module at each
amplifier takes care of the dispersion loss accumulated over
each fiber link length of 80km in the sample network.
For the case of the WSS based C&D ROADM architecture
(Fig. 4), every wavelength added at the local add site of each
node has to pass through a protection switching card PSC
(3.5dB loss), a wavelength selective switch module WSS (6dB
loss), two 1x8 power splitters (9dB loss each) and again a
WSS module, thus suffering a total loss of 33.5dB at each
local add site of every node. The pass-through channels have
to pass through a 1x8 power splitter and a WSS which
constitutes a lump loss of 15dB along every express path at
each node. The wavelength to be dropped at a node must pass Fig. 8. Power variation along the path A-E-F-B for the case of PXC based
through two 1x8 power splitters, two WSS modules followed C&D ROADM architecture
by a PSC and thus suffering a total loss of 33.5dB at local
drop site of every node. These power losses at various stages
were compensated by placing amplifiers of gain values 18 dB
(with a noise figure of 9.6dB) and 12 dB (with a noise figure
of 18.7dB) at suitable positions of every node and at local
add/drop sites.
Similarly, for the case of the CD&C ROADM architecture
(Fig. 7), every wavelength that is added from a local add site
of a node must pass through a PSC (3.5dB loss), 320x320
MEMS switching core (2.5dB loss), a 40 channel Mux (4dB
loss), a protection switch (3.5dB loss) and a WSS module
(6dB loss) bearing a total loss of 19dB at every local add site
of each node. The pass-through channels have to go through a
1x8 power splitter & a WSS module and hence suffering a
lump loss of 15dB along each express path at every node. The
drop wavelengths encounter a 1x8 power splitter (9dB loss), a
1x2 power splitter (3dB loss), a 40 channel DeMux (4dB loss),
Fig. 9. Power variation along the path A-E-F-B for the case of WSS based
320x320 MEMS switching core (2.5dB loss) and a protection C&D ROADM architecture.
switch (3.5dB loss) which constitutes a total loss of 22dB for
every local drop site of each node. By placing amplifiers of
compared to the PXC based and WSS based C& D ROADM
architectures for the considered path along A-E-F-B in the
sample network.
For the path along nodes A-E-F-B in the sample network,
the proposed CD&C ROADM architecture has a maximum
OSNR value of 24dB at the amplifier placed just after the
node A (post-amp at node A) and a minimum OSNR value of
16.4dB at the amplifier placed just before the node B (pre-amp
at node B) which are much higher compared to the OSNR
values at post-amps of node A and pre-amps of node B for the
cases of PXC based and WSS based ROADM architectures.
Due to the high losses associated with the three ROADM
architectures, repeated amplification is required to sustain the
strength of the signal. This in turn results in the OSNR of the
signal dropping below the required level if the amplifier
placements are not done appropriately. In the performance Fig. 10. Power variation along the path A-E-F-B for the case of CD&C
ROADM architecture.
analysis, we have placed the amplifiers of suitable gain values
at appropriate positions in each of the nodes in the sample
network for all the three ROADM architectures. This ensures
that the power level could be steadily maintained well above
the mark of -26dBm which is the lower threshold of the APD
receiver sensitivity and also the OSNR values at much higher
level than the minimum requirement of 9dB. It is to be noted
here that the proposed CD&C ROADM architecture could
also be 40Gbps ready as it readily provides the 6dB higher
OSNR margin (Fig. 11) required for 40Gbps signals [10].
Power balancing of all the channels is another important
issue for proper amplification of the signal in the proposed
CD&C ROADM architecture. The power balancing can be
achieved at the WSS site in each direction by software
controls at the optical layer. Thus a proper power balancing
among all the channels is ensured before the signal enters an
EDFA placed just after the WSS in each direction.
B. Choice of suitable sub-system technology
The choice of suitable switching sub-system for the PXC
Fig. 11. OSNR values at amplifiers along the path A-E-F-B.
plays an important role in performance of the CD&C
architecture. Over the last couple of years, different of low and uniform insertion loss with low wavelength
technologies have emerged of which some were fast, while dependency under a wide variety of operating conditions.
others were slow. Some were ideal for small scale solutions, Also, this switch exhibits minimal degradation of the optical
while some provided a large number of inputs and outputs signal-to-noise ratio, which is mainly caused by crosstalk,
[11]. Liquid crystal based switches did not have any moving polarization dependent loss (PDL), and chromatic and
parts, but required precise design as any slight defects in the polarization mode dispersions.
fabrication would introduce considerable losses. Though the
‘bubble’ based switch provided fast switching time, the C. Cost considerations
maximum switch size realized was 32x32. The thermo-optical In the WSS based C&D ROADM architecture, it is evident
waveguides provided still faster switching times (in the order that the overall cost of the architecture at a node depends
of few ms), but constant heating and cooling had adverse critically on the number of WSS modules and power splitters
effects on the device. MEMS devices had a large number of used. The high loss due to cascading WSS and power splitters
moving parts which made these devices not so reliable. calls for more number of amplifiers with higher gains, which
Moreover, these devices were vulnerable to vibrations and further increases the costs. Also, the number of individual
shocks and hence required vibration-free environment. In spite modules increases with more number of amplifiers. As a
of these drawbacks, the MEMS based switches have become result, power consumed in wattages and the space occupied in
the favorite of many due to its ability to achieve input- and the racks by these modules would further increase the overall
output-port counts of over one thousand, in which spatial cost of implementation of the architecture. Similarly, in the
parallelism is utilized. In particular, the MEMS switch PXC based ROADM architecture, the overall cost of
provides high application flexibility in network design because implementation of the architecture is very expensive as we
need 320x320 dual core switches to avoid single point of node. The same costs were considered for the cost analysis of
failure. On the other hand, the proposed Colorless, the proposed CD&C ROADM architecture.
Directionless and Contentionless architecture involves far The cost analysis depicted in Fig. 12 shows that at lower
lesser number of WSS modules, power splitters and add/drop ratios the WSS based ROADM and the CD&C
amplifiers, and hence the power consumed and the space ROADM involves much lower costs than the PXC based
occupied is much lesser compared to the WSS based and PXC ROADM. In the case of the WSS based ROADM, 1x9 WSS
based C&D ROADM architectures. Therefore, the CD&C are used which can handle up to 9 wavelengths and hence,
architecture calls for relatively lower CAPEX compared to the provide modular growth with increase in number of add/drop
WSS based and PXC based C&D ROADM architectures, wavelengths. This is not possible with the PXC based
while providing similar OPEX savings. ROADM as it is a 320x320 device. However, as the add/drop
Fig. 12 shows the cost comparison of PXC based ROADM, ratio increases, the increase in the cost of the WSS based
WSS based ROADM and the proposed CD&C ROADM ROADM is more than the PXC based ROADM, as more 1x9
architectures. The overall cost of the architecture at a node is WSS devices are required to handle the increased number of
calculated by considering the cost of the devices and the cost add/drop wavelengths. On the other hand, the CD&C
based on number of ports used for a particular add/drop ratio ROADM involves much lesser costs compared to the other
at the local ports. The number of ports used can be calculated architectures at all add/drop ratios, as the PXC is used at the
as, add/drop site instead of multiple WSS modules.
No. of ports used = (No. of degrees) x (No. of channels per
D. Other challenges
degree) x 2 + (Add/drop ratio) x (No. of degrees) x (No. of
channels per degree) x 2. Another challenge faced by the service providers is the
In the cost analysis, we have considered the architecture at a connection between the core switches and the
4 degree node having a total of 40 channels per degree with multiplexers/demultiplexers, as these components often
add/drop ratios of 0.25, 0.50, 0.75 and 1. For the case of PXC involve complex interconnections that need to be made
based ROADM, we have a 320x320 non-blocking switch with manually. This problem can be alleviated by integrating the
a cost of $200 per port. The device cost can be calculated core switches and the multiplexers/demultiplexers on a single
based on total no. of ports in the device and in our case, we substrate. Doing so provides significant benefits as it
have a total of 640 ports (320 input ports & 320 output ports) eliminates the need for fiber interconnections and allows tight
per PXC. Also, we use two such PXCs to avoid single point of coupling between the two to support efficient protection.
failure at the node. The unidirectional nature of the PXC ports However, this integration also has the drawback of making it a
has been taken into account in the calculation of the cost based single-vendor solution. This would then force the service
on number of ports. Also, a cost of $3,000 for a pair of AWG providers to buy all their WDM equipments from the same
based Mux/DeMux has been considered per degree at the 4 vendor in order to realize this simplification. Some service
degree node. For the case of WSS based ROADM, we have providers prefer to build their network by mixing and
considered 1x9 WSS with a cost of $9,000 per WSS module. matching "best-in-class" equipment from multiple vendors.
The overall cost of the architecture at a node is calculated by With the service providers opting for mesh optical networks
determining the number of 1x9 WSS required for handling the over ring-based networks and the increase in the wavelength
services from all the directions at the local add/drop site of the capacity of the fiber, scalability beyond eight degrees and
forty wavelengths is yet another challenge faced by this
architecture. However, with the advances in MEMS design,
optical cross connects with large port counts (>1000) have
been demonstrated that provide high switching capacity,
reduced footprint and lesser loss [12].
The CD&C ROADM architecture provides a lot more
advantages compared to the WSS and the PXC based C&D
ROADM architecture. However, there are still a few
challenges that need to be resolved in order to realize a truly
cost effective solution. The PXC that is used for add/drop at
the local node acts as single point of failure and requires a
similar PXC that acts in conjunction with the first one. As far
as the protection for the PXC is concerned, different
architectural implementations are possible. The final choice,
however, rests with the service providers as optical power loss
associated with the various implementations is not uniform.

Fig. 12. Cost comparisons of PXC & WSS based ROADMs with CD&C
ROADM.
V. CONCLUSION
The proposed Colorless Directionless and Contentionless
architecture enabled by the combination of the Wavelength transmission on a common all-Raman single-wide-band platform,”
ECOC 2002, Vol. 1, Page(s):1 – 2.
Selective Switch based ROADM and the Photonic Cross
[11] X. Ma, G. S. Kuo, “Optical Switching technology comparison: Optical
Connect provides Contentionless add/drop apart from other MEMS vs. Other technologies,” IEEE Optical Communications
valuable features like Colorless and Directionless add/drop, magazine, Novemeber 2003, pp. S16-S23.
dynamic optimization of lightpaths by optical bridge-and-roll, [12] J. I. Dadap, P. B. Chu, I. Brener, C. Pu, C. D. Lee, K. Bergman, N.
elimination of wavelength conversion and alternate protection Bonadeo, T. Chau, M. Chou, R. Doran, R. Gibson, R. Harel, J. J.
Johnson, S. S. Lee, S. Park, D. R. Peale, R. Rodriguez, D. Tong, M.
paths in a mesh optical network. This coupled with other Tsai, C. Wu, W. Zhong, E. L. Goldstein, L. Y. Lin, J. A. Walker,
features like flexibility, ease of provisioning, modular “Modular MEMS-based optical cross-connect with large port-count,” in
add/drop and minimal pre-planning will allow the service IEEE Photon. Technol. Lett., Vol. 15, no. 12, Dec. 2003, Page(s):1773 –
1775.
providers to experience reduced CAPEX and OPEX. This will
enable the service providers to offer customers next-
generation DWDM technology without the added cost and
management of separate equipment. Thus, service providers
will be able to benefit from higher margins on the services
they deliver. Moreover, the customers also will receive more
bandwidth for their dollar. Hence, the proposed CD&C
architecture can truly change the economics of service
delivery in the days to come.

ACKNOWLEDGMENT
The authors thank Prof. Byrav Ramamurthy and all the
reviewers for their comments which led to a significant
improvement of the paper.

REFERENCES
[1] G. Prasanna, B. Sai Kishore, G. Omprasad, K. Sathyanarayana Raju, R.
Gowrishankar, K. Venkataramaniah, R. Johnson, P. Voruganti,
“Versatility of a colorless and directionless WSS based ROADM
architecture,” in Communication Systems and Networks and
Workshops, 2009. COMSNETS 2009, 5-10 Jan. 2009, Page(s):1 – 8.
[2] V. Kaman, R. J. Helkey, J. E. Bowers, “Compact and scalable three-
dimensional Microelectromechanical system optical switches,” Journal
of Optical Networking, Vol.6, No.1, 19-24 (2007).
[3] X. Zheng, V. Kaman, S. Yuan, Y. Xu, O. Jerphagnon, A. Keating, R. C.
Anderson, H. N. Poulsen, B. Liu, J. R. Sechrist, C. Pusarla, R. Helkey,
D. J. Blumenthal, and J. E. Bowers, “Three-dimensional MEMS
photonic cross-connect switch design and performance,” IEEE J. Sel.
Top. Quantum Electron. 9, 571–578 (2003).
[4] V. Kaman, R. J. Helkey, J. E. Bowers, "Multi-Degree ROADM's with
Agile Add-Drop Access", in International Conference on Photonics in
Switching, 2007, 19-22 Aug. 2007, Page(s): 31-32.
[5] L. Y. Lin, E. L. Goldstein, R. W. Tkach, “On the expandability of
freespace micromachined optical cross connects,” Journal of Lightwave
Technology, Vol. 18, pp. 482–489, Apr. 2000.
[6] P. B. Chu, S. S. Lee, and S. Park, “MEMS: the path to large optical
crossconnects,” IEEE Commun. Mag., vol. 40, Mar. 2002. Page(s): 80-
87.
[7] J. Kim, C. J. Nuzman, B. Kumar, D. F. Lieuwen, J. S. Kraus, A. Weiss,
C. P. Lichtenwalner, A. R. Papazian, R. E. Frahm, N. R. Basavanhally,
D. A. Ramsey, V. A. Aksyuk, F. Pardo, M. E. Simon, V. Lifton, H. B.
Chan; M. Haueis, A. Gasparyan, H. R. Shea, S. Arney, C. A. Bolle, P. R.
Kolodner, R. Ryf, D. T. Neilson, J. V. Gates, “1100 × 1100 port
MEMSbased optical crossconnect with 4-dB maximum loss,” in IEEE
Photon. Technol. Lett., Vol. 15, no. 11, Nov. 2003, Page(s):1537 – 1539.
[8] D. T. Neilson, V. A. Aksyuk, S. Arney, N. R. Basavanhally, K. S.
Bhalla, D. J. Bishop, B. A. Boie, C. A. Bolle, J. V. Gates, A. M.
Gottlieb, J. P. Hickey, N. A. Jackman, P. R. Kolodner, S. K. Korotky, B.
Mikkelsen, F. Pardo, G. Raybon, R. Ruel, R. E. Scotti, T. W. van
Blarcum, L. Zhang, C. R. Giles, “Fully provisioned 112×112 micro-
mechanical optical crossconnect with 35.8 Tb/s demonstrated capacity,”
in OFC 2000, Vol. 4, Page(s):202 – 204.
[9] A. Gumaste and T. Antony, DWDM Network Designs and Engineering
Solutions, Indianapolis, IN : Cisco Press, 2003.
[10] A. Kung, A. Agarwal, S. Banerjee, D. F. Gros, D. N. Maywar, M.
Movassaghi, T. H. Wood, “10Gb/s and 40Gb/s ultra-long haul

View publication stats