Download Wavelength-division multiplexing(WDM) is currently being deployed

EE228a Project Proposal
Issues in All-Optical IP Router Design
Bo Wu, Fan Mo and Xuesong Jiang
Over the past decade, the volume of data traffic transported across telecommunication
networks has grown rapidly and the increase in traffic volume is forcing both incumbent
and emerging service providers to design, build and add capacity to their networks. In
order to satisfy the ever increasing demand for bandwidth brought about by the explosion
of Internet, optical networks using wavelength-division multiplexing(WDM) are
currently being deployed worldwide. It has been shown that the capacity of WDM is
growing at a rate beyond Moore’s law and several Tb/s data transmission rate has been
demonstrated in labs.
The benefits of optical networking include:
 Large bandwidth. The bandwidth available in the low-attenuation passband within a
standard single-mode optical fiber is 25THz.
 Multiple channels can be carried by a single fiber.
 More bandwidth can be requested and allocated dynamically by optical links.
 Easy Quality of Service(QoS) provisioning.
In this project, we will address issues of all-optical IP router design in order to bridge the
mismatch between the transmission capacities provided by the WDM layer and the
processing power of IP router. We will focus on all-optical networking in which the
transmission of data will not go through any optical-electrical conversions(Although the
connections or light paths may be controlled by electronics).
IP routers perform four main functions: routing, forwarding, switching and buffering. The
major elements in an all-optical IP router include:
 A large optical switching fabric that provide the bandwidth management features and
can be duplicated to provide high transmission availability.
 Transmission interfaces for connection to optical transmission facilities.
 A control structure that provide high control and maintenance availability.
 Primary non-volatile memory(NVM) and additional secondary NVM for robust
software and database management.
Currently, transmission and switching are executed in the optical domain, while routing
and forwarding are carried out electrically, where the relatively complex packet header
processing occurs independent of optical payload. This decoupling effectively permits the
optical packet layer to support a range of networking protocols while attaining the power
of WDM transmission. Nevertheless, in order to implement all-optical networking,
optical packets should be routed in optical domain as well.
Given that WDM allows cheap and easy incremental increases of the transmission
bandwidth, frequent updates of the transport layer transmission capacity can be envisaged
to match increased demand, which in turn placing heavy demands on the switching
process. WDM optical packet switching shifts the bulk of the switching burden into the
optical domain, permitting compatible scaling of the switching capability with WDM
transmission capacity.
The feasibilities and challenges in optical packet switching have been widely investigated
in the last few years. Optical IP routers can be used as core routers in internet backbones.
The inlets and outlets of a core router are optical fibers which operate in WDM mode
with a number of wavelengths per fiber depending on the technology. More wavelengths
per fiber could yield a higher degree of statistical multiplexing and lower blocking
probability in the router. Nevertheless, problems such as all-optical synchronization,
buffering, and node cascadability still need to be solved completely. The solutions
available at the moment are far from field implementations.
In order to effectively develop an IP router capable of providing a full optical data path at
speeds approaching the limits of current electronic devices, namely 10 Gb/s and higher, a
transfer mode has to be adopted robustly and efficiently enough to overcome
technological limitations. Burst switching has been proposed as a possible solution, and it
is well adopted in optical IP routing.
However, processing of predominantly short IP packets becomes a problem at such high
speed; therefore, it is necessary to introduce some lower limit on the granularity of
switched data units in the backbone. A burst switching transfer mode paradigm has been
proposed. Incoming packets are collected into bursts in the edge routers and sent through
the backbone network where core routers will perform forwarding according to the
destination in the burst switching. The core routers are supposed to exploit optical
technology to reach a throughput on the order of multiple terabits per second and beyond.
The various stages may be implemented by means of different technologies and with
different architectures, but the basic functions to be performed are those described above.
No optical-electrical conversion of data bursts is performed. This task is achieved by
means of an all-optical switching matrix, for instance, based on the use of tunable
wavelength converters and semiconductor optical amplifier space switches.
Reference:
[1]. S. Keshav and R. Sharma, “Issues and Trends in Router Design,” IEEE
Communications Magazine, May 1998, pp. 144-51
[2]. D. Hunter and I Andonovic, “Approaches to Optical Internet Packet Switching,”
IEEE Communications Magazine, September 2000, pp. 116-122