Conference flags optical progress

At the recent European Conference on Optical Communication (ECOC) in Geneva, we noted significant progress in many areas of technology which we have either written about recently or will shortly: ROADM architectures, 100+ Gbps transport, and 10G/40G/100G transceivers.

 

While some attendees complained the show was lackluster, we believe the absence of drama is a sign the industry is moving rapidly and smoothly. The event had the paradoxically motionless feel of cars driving at full speed on a well-paved highway. Last year, Google shook up the establishment with its push for nonstandard 10×10 modules in an effort to speed up the market. This year, established vendors had a flurry of new product demonstrations that left researchers scrambling to stay ahead of commercial development.

 

Last year, we described the ROADM architecture desired by operators as “elusive”. The challenge to provide colorless, directionless, possibly contentionless, and finally, flexible grid functionality had…OEMs and component vendors tied in knots – as much as their complicated block diagrams connecting too many components together.

 

We now see the industry has broken free and is moving forward again. Major component suppliers were at the show advocating a solution, and the same solution to boot. The block diagrams we saw in discussions with CoAdna, Finisar, JDSU, and NeoPhotonics were the same. The base technologies and implementations continue to differ, though all vendors are supporting the flexible grid.

 

Consensus does not mean that all ROADMs will immediately jump on board with full functionality. Operator choice of how much flexibility to adopt is expected to differ based on traffic patterns as well as cost of labor. Contentionless appears the most likely attribute to be left behind in the cost-benefit trade-off.

 

 

One researcher commented that coherent technology for long-haul transmission had progressed “faster than anyone expected” out of research labs and into widespread commercial development. Rapid progress in 100G has been due to industry alignment around the same modulation format (DP-QPSK) and a coherent receiver.

 

Japanese OC suppliers have further demonstrated the power of coordination: a consortium of Japanese vendors has developed the first merchant 100G coherent transponders with soft forward error correction (FEC), a major advance in performance from the current generation’s hard FEC. The critical signal processing IC is made by NEL – that is to say, under the NTT umbrella but with key contributions from Mitsubishi for the algorithm, and Fujitsu Microelectronics for the analog-to-digital conversion (ADC). Fujitsu Optical Components and Opnext [both] announced modules at ECOC.

 

Meanwhile, in the research workshops, Ciena described a programmable transmitter that can flexibly change its modulation format from, say, QPSK to 16-QAM. There was general agreement in the room that ADC plus signal processing had done such wonders to improve the receiver capabilities that the obvious next step was to add signal processing plus digital-to-analog conversion (DAC) to the transmitter.

 

The goal is not only to enable adaptive transmitters, but also to extend compensation from linear impairments (chromatic dispersion, polarization mode dispersion) to nonlinear impairments. Nonlinearities are seen as a major limitation to the higher transmit power required to achieve a signal-to-noise ratio suitable for 400G and 1Tbps transmission.

 

 

[The capabilities of] 10G…continue to advance as well. While Opnext, Finisar, and Oclaro all announced new tunable XFP modules including extended-reach zero-chirp variants, JDSU moved to the even smaller SFP+ form factor. Sumitomo and Finisar showed SFP+ modules with improved jitter to meet SONET/SDH specifications.

 

Some attendees were disappointed to see no follow-up fireworks on the 10×10 client 100GbE format. We believe the true alternative to the 10×10 is actually the 40GbE-LR QSFP+, which is similarly based on 10GbE technology. We expect to see it make a big impact in the coming year. Opnext and JDSU were both showing this product.

 

Inphi announced a CMOS gearbox IC which reduces cost and power consumption of the 100GbE CFP. But chip vendors expect the next generation beyond CFP soon: both Inphi and Gennum showed 28Gbps chips for the smaller CFP2 100GbE module.

 

Long-haul 100G using coherent technology and client-side 100G relying on optical integration are both in place, but there has been no cost-optimized solution for metro distances. Several companies have recently made announcements in this space, but most notably Adva and MultiPhy have proposed distinct approaches that avoid the need for the expensive and scarce coherent signal processing chip.