Summary
1. Ciena has been trying to work on its own100G solution for a while. 2. Its current approach is a slight modification from PM-QPSK, which the OIF is trying to get everybody to follow. 3. There is no question that there will be cost concerns with Ciena’s implementation.
Analysis
With a PM-DQPSK, the transmit side cost appears to be close to the same as other techniques. The receiver is the big differential. Ciena is proposing modulating ten 10G signals onto a single ITU channel to transport across the network. So, it is really just an implementation from the client side coming in and multiplexing together to be able to modulate and to support the 100G single channel. This method is in contrast to using four or ten ITU channels.
The IEEE has agreed to both the 10x10 and the 4x25 as being modulation techniques to support 100G interfaces between two different elements in the network. Some people believe that 10x10 is going to be really cheap because it involves just 10G optics. Others are saying 4x25 is going to be potentially cheaper because one is not going to use as many rails – so it makes it more cost-effective for the long term as volumes ramp and efficiencies are gained.
In the case of Ciena, it is just talking about a demo involving only 80 kilometers of fiber. It is also hard to envision the supplier taking too active a role in developing a solution when there is a lot of investment being made by other vendors and when there are so many modulation schemes under consideration. In fact, it would not be surprising if Ciena goes back to Opnext, which is providing the 40G technology for its CoreStream product, which is using QPSK. The subsystem vendor’s strategy for 100G is to plug in a new transponder in one of the currently deployed shelves.
