Next show in
 d  h  m
18 October 2011
The next generation of reconfigurable optical networks will require optical components vendors to develop a completely new set of building blocks, says JDSU. Speaking at the Market Focus forum at ECOC in Geneva last month, JDSU outlined its approach to what it terms "self aware networks".
"We're moving from a 'broadcast and select' to a 'route and select' architecture, and for ROADM networks that's a pretty big deal," said Brandon Collings, JDSU's CTO for communications and commercial optical products. "These new networks require a full new set of optical components.  We can't take what we have today, and that includes the WSS [wavelength-selective switch]."
Optical companies have been talking about next-generation ROADMs for a number of years, of course. Carriers aspire to ROADM nodes that are colourless (can use any wavelength without physical reconnection), directionless (connect any port to any port in any direction), and contentionless (non-blocking in the wavelength domain), and have recently added a new requirement that will enable efficient use of spectrum by high-speed optical channels, the flexible grid.  The "self aware network" is characterised by these four capabilities.
"All these functions can be delivered with existing technology but the solutions are expensive. Technical innovation is needed to provide a cost-effective solution," said Daryl Inniss, lead analyst, optical components for Ovum, in a recent research note. 
JDSU's announcement indicates that components vendors are getting closer to delivering such products.   JDSU had talked about and demonstrated a high-port-count 1x23 WSS in early 2010, but now says it plans to release two 1x20 WSSs integrated into a single package in 2012.  Higher port-count devices increase the number of degrees supported at the node as well as the number of add-drop channels.  Putting two devices into one package enables the "route and select" approach, where there is a WSS on the ingress of each direction rather than a power splitter (as well as on the egress.  
Using "route and select" with back-to-back WSS components relaxes the isolation requirements on individual devices, since there is a second chance to filter the signal, says Collings. The architecture also enables faster switching time and lower optical loss overall. The drawback is that every channel flowing through the device now experiences two filters.  "We need to make the pass bands squarer and wider so that we can still cascade the same number of nodes," he explained.
The WSS is just one part of the solution, however.  Ovum's Inniss notes that "widespread deployment [of next-generation ROADMs] is still a few years away because high-port-count WSSs are a partial solution to a broader problem."
Aside from the WSS, JDSU is also developing the multicast switch (MCS), which transports the add-drop channels in any direction in a colourless, non-blocking way.  There are still two ideas on the table to produce a subsystem with acceptable losses: tunable filter and amplifier arrays to offset the losses of the MCS, or coherent detection, which has lower loss because the coherent receiver is tuned to pick out a specific wavelength (no filtering required).  JDSU says it is working on both.
Other vendors are making progress towards similar goals.  In the run up to ECOC, both Capella and Finisar announced 1x20 WSS products.  Finisar can claim to be the only vendor to offer software-definable grid spacings thanks to the liquid crystal-on-silicon technology inside – although this is more desirable than essential as it will be some years before carriers deploy speeds beyond 100G in their networks.  Finisar is exploring the idea of a programmable mux/demux device as an alternative to the MCS.
By Pauline Rigby