Chapter 3.2.4 - Optical Switch Fabric Technology and Architecture

3.2.4 Optical Switch Fabric Technology and Architecture

The optical switch fabric is the heart of the optical packet switching node. The
choice of the optical switch fabric technology weighs heavily on the performance
of an OPS node. The most important parameter of an optical switch fabric is its
reconfiguration speed, that is, the time it takes to prepare for a switching function.
To match the current transmission speeds, optical packet switches require reconfiguration
times in the order of nanoseconds [10]. Unfortunately, commercially
available optical switching products such as the ones based on micro-electrical
mechanical switches (MEMS) feature speeds in the order of milliseconds, which
does not suffice. Other important characteristics of the switch fabric include the
ability to scale to large port counts, the ease of manufacture and low cost, the temperature
independence and the blocking characteristics. These refer to the ability of a
switch fabric to provide a path from any of its inputs to any of its outputs at all times
without the need for reconfiguration. Promising optical switch fabric technologies
for optical packet switching include semiconductor optical amplifiers (SOAs),
used as on–off gates, and electro-optic lithium niobate (LiNbO₃) switches.

Another issue surrounding the optical switch fabric is the domain in which
switching will be performed. Connections between switch inputs and outputs may
be achieved in the space or wavelength domains [11]. Wavelength-selective switches
are commonly implemented using arrayed waveguide gratings (AWGs) and tunable
wavelength converters [12]. An AWG component has the interesting feature that
when light is inserted via one of its input ports, the output port it will come out of
depends on the wavelength used [13]. Tunable wavelength converters at the AWG
inputs convert a packet to the wavelength that will allow it to access the desired
output port. The suitable wavelength is determined by a table lookup operation.

A common approach taken for space switching involves the broadcast-and-select
principle implemented, for example, with SOA gates. In such switches, input signals
are broadcast through a splitting stage to all output ports. Using two successive SOA
stages, a single wavelength signal is kept for each output port. The first SOA array is
used to select only one of the input fiber signals for each output port. The second
selection stage uses an SOA array and a wavelength-selective component to keep
only a single wavelength per output port [13].

The main advantage of the broadcast-and-select architecture is its inherent multicast
capability, which the AWG approach lacks. However, the asset of the AWG-based
architecture is that it relies on a passive component and does not suffer
from splitting losses as does the broadcast-and-select architecture [13]. The splitting
stage is a major impairment of the broadcast-and-select switch, which degrades the
optical signal and does not allow the architecture to scale to large port counts. On the
other hand, a wavelength-selective switch may not be able to realize a connection
between an input and an output port if the tenability of the wavelength converters
is limited. This also poses restrictions on the switch scalability.

Clearly, in both approaches, scalability is an important concern. A solution is
to employ multistage architectures such as the ones discussed in Chapter 2
(Section 2.4). As was pointed out in Section 2.4, a very important property of a
switch architecture constructed by cascading small switches is its blocking characteristics.
A switch that can realize any connection between an input and an output
port at any time without disturbing previously established connections is said
to be strictly nonblocking. If the switch must first reroute existing connections
in order to accommodate the new request, it is said to be rearrangeably nonblocking.
A switch that cannot always guarantee the ability to realize a connection between
an idle input and an idle output port is said to be internally blocking.
Switches employed in OPS networks are required to be at least rearrangeably
nonblocking.

A popular architecture for building multistage switches is the three-stage Clos
network (illustrated in Fig. 2.12). The number of switches that are used to construct
such a network depends on the degree of blocking that is required for proper
network operation. For synchronous optical packet switching, it suffices to have a
rearrangeably nonblocking switch because all connections are simultaneously set
up and torn down at the beginning and end of a timeslot. On the other hand, asynchronous
switching requires strictly nonblocking switches, because connections
have variable durations and are established independently in the presence of
other connections. Thus, a switch with a three-stage Clos network architecture
constructed for use in an asynchronous network will require a higher number of
components.

< Previous Excerpt Next Excerpt >

Purchase This Book

TABLE OF CONTENTS

Chapter 3.2.4 - Optical Switch Fabric Technology and Architecture

Contact Preferences

This is embarrasing...

Customize Your GlobalSpec Experience

Select Your Free Newsletters

Industry Newsletters

Select Your Free Product Alerts

This is embarrasing...