sonata project to play all-optical tune
By STEPHEN HARDY
Optical internetworking proponents say the use of high-speed switches optimized for data traffic will soon make several layers of Synchronous Optical Network equipment obsolete. The new Switchless Optical Network for Advanced Transport Architecture (sonata) effort, part of the European Advanced Communications Technologies and Services (acts) program, aims to relegate these switches to the network dustbin as well.
sonata researchers envision a single-layer optical network that would encompass both the concentration and distribution functions of the access layer and the switching and routing functions of the core network layer, as well as everything in between. Essential network elements will include network terminals, situated in the network much like the optical network units of fiber-to-the-home architectures, and a central node. The terminals will be both Otime- and wavelength-agile,O in the words of project coordinator Nunzio Paolo Caponio of cselt, one of the organizations involved in the project.
OThe use of time-and-wavelength-agile terminals, able to transmit and receive optical packets at different wavelengths, allows both the time and wavelength domains to be fully exploited with a simple network structure requiring only a single wdm [wavelength-division multiplexing] routing node in the whole network,O he says. Caponio further explains that only passive routing functions would be required if blocking statistics are acceptable. Actively controlled wavelength conversion would be available at the central node to overcome blocking as required (see figure).
The terminals would use a time-division multiple-access switching scheme. The total capacity of each terminal would be Oin the range of hundreds of megabits per second,O Caponio says, because of the relatively small number of user connections at each terminal. The architecture could be scaled to national levels, Caponio predicts. The single central node for such a network, which would perform all of the passive wavelength routing and wavelength conversion, would likely have 800 input and output ports and would be able to route up to 320,000 different optical channels. Each channel would be modulated at 300 to 600 Mbits/sec for a total throughput of between 100 and 200 Tbits/sec. The wavelength-conversion block would have 160,000 wavelength-converting elements and would process half of the total network throughput. This is the only point in the network where conversion between the optical and electrical domains might take place, Caponio says.
The systems and subsystems that would compose such a network are currently under development by the sonata team. Project members include cselt (network architecture and control studies, subsystem design and construction, test-bed experiments, and laboratory demonstrator implementation and evaluation), Politecnico di Torino (network control studies), BT Laboratories (coordination of the system studies), Coritel (network architecture and Physical layer modeling studies), eth Z?rich (multichannel burst-mode receiver and optical amplifier gate development), Instruments SA (wdm demultiplexer development), Thomson csf (all-optical wavelength converter development), imec (network management studies), and Universitat Politecnica de Catalunya (network control studies). Caponio expects that sonata components will become commercially available shortly after their development is completed. However, the subsystems will be developed only as lab prototypes.
The project began this spring and is expected to run through January 2000. Component development should conclude in August 1999, with subsystem development ending in October 1999. Test-bed experiments should run from January 1999 through September 1999, with laboratory experiments running from January 1999 through January 2000. q