NeoPhotonics claims optical transmission cost reductions

OvumECOC 2012, held in Amsterdam September 16–20, featured the optical communication industry delivering 100G today, but it also offered a glimpse of how researchers are developing next-generation solutions.
 
The industry anticipates continued bandwidth growth and hence the need for even lower-cost optical transmission. Consensus is forming that 400G is the next transmission rate, and the market is also coalescing around flexible grid and super-channels.
 
Two NeoPhotonics announcements struck me because they addressed immediate problems. First, the company introduced a mode-coupled receiver that lowers initial PON cost by increasing the number of connections per OLT from 32 to 128. Second, NeoPhotonics said it will offer its 10×10 transceiver in a CFP2 form factor.
 
While the 10×10 solution is seen as a stopgap measure, it supports distances not included in existing standards and is sorely needed today for connectivity in data centers and central offices.
 
Lowering initial PON installations costs
 
NeoPhotonics’…new GPON OLT transceiver…supports four OLTs where each can have up to 32 connections. The cost decreases because the splitting ratio increases: one OLT transceiver is used for 128 connections compared to the initial design of 32 connections per transceiver.
 
Bandwidth performance is not compromised in this implementation because initial PON turn-up typically has far fewer subscribers than 32. NeoPhotonics’ GPON OLT is equipped with a higher power downstream laser that is shared with all PONs. For upstream traffic it introduced a multi-port mode-coupling receiver so the four PONs are connected to a single receiver.
 
Ovum found surprisingly low ratios of deployed ONUs per OLT and published its findings in FTTx Deployment Review: Network Utilization Based on PON Equipment Ratios (June 2011). The low utilization rates suggest high initial costs. It is already understood that the highest PON deployment cost is associated with the infrastructure installation including laying fiber. The equipment cost ranges from 10% to 35%. Increasing the split ratio amortizes the equipment cost further, improving the attractiveness of PON.
 
 
British Telecom presented a joint paper with NeoPhotonics highlighting benefits of the new GPON OLT. The joint report signals the potential value to the carrier network. As the PON fills up with more subscribers, the service provider can decrease the split ratio to provide higher bandwidth.
 
Seizing opportunity with 10×10 transceiver to CFP2 form factor
 
Winston Way, chief systems architect, described NeoPhotonics’ development of a 10×10 transceiver in a CFP2 form factor in his “PIC for telecom and datacom applications” presentation. Previous ECOCs included colorful debates about the 10×10 transceiver because it directly competes with IEEE standards.
 
The former uses existing 10G technology, is cheaper, and is supported by Google and Facebook. At the time of the debates, 25G laser technology was just being introduced to the market, so there were few suppliers and prices were high; this is still the case today.
 
Last year after NeoPhotonics acquired Santur, the only 10×10 transceiver supplier, the product all but disappeared from the radar. Now we learn that 10×10 development continues but in the smaller form factor, suggesting it is shipping and that customers want more.
 
Note that, although we did not find any other 10×10 transceiver suppliers for this reach, we did find IC vendors developing the gear box for the line card to support both 4×28 and 10×10. This is a feather in the cap for 10×10 because it suggests widespread use since OEMs only need develop one line card and use either transceiver.
 
The 10×10 supports 2-km and 10-km distances, whereas the IEEE standards are for 4×28 supporting 100-metres and 10-km at 100G. There is an immediate need for low-cost, small-form-factor transceivers to cover these intermediate distances. The 10×10 in the CFP2 form factor is well suited for this purpose. We believe that the market in the long term will be served by the 4×28 (i.e. IEEE) solution, but until standards are available and technologies are ready, early adopters will use the 10×10.
 
Daryl Innis is practice leader for components telecoms at Ovum. For more information, visit www.ovum.com/