By Geoff Bennett
Director, Solutions & Technology
Do you remember that old cliché where you walk up to a local in a strange town and ask for directions to the station? The classic reply is, “If I wanted to get to the station, I wouldn’t start from here.”
While we may not always have a choice of starting place, we still must solve the problem of getting to our destination, and that means we need viable options. Today the “destination” for subsea operators is to deliver massive increases in subsea cable capacity in order to meet the staggering 41 percent global subsea traffic growth rate that Telegeography is forecasting over the next six years.
So…where are cable operators starting from? There are several options, and I’ve summarized them in Figure 1.
On the left side we see a cable operator with a legacy dispersion-managed fiber system. This would have been designed in the pre-coherent age with alternative positive and negative dispersion fiber lengths to limit the accumulated chromatic dispersion (CD) in the cable. This was a smart decision at a time when CD was considered an undesirable fiber impairment. But modern coherent technology can fully compensate for CD. In fact, fibers with low CD have a higher non-linear penalty for coherent transmission systems, and this limits the ability to achieve higher spectral efficiency using high-order modulation.
Moving to the center of Figure 1, perhaps the cable operator is lucky enough to operate one of the new, large area fiber cables such as MAREA, Seabras-1 or BRUSA, which have a much lower non-linear penalty and will allow higher-order modulations, such as 16 quadrature amplitude modulation (16QAM), to be transmitted across Atlantic distances. Recent deployments have shown that production-level capacity of 24 terabits per second (Tb/s) is possible over trans-Atlantic distances.
In both cases, the key requirement is to use a high-performance optical engine that can operate as close as possible to the theoretical capacity limit of the fiber, which can be calculated using the Shannon equation. The difference between a modern cable and a legacy cable is that the former will allow a given optical engine to approach the Shannon limit much more closely. In fact, implementations based on Infinera’s fourth-generation Infinite Capacity Engine (ICE4) technology are getting so close to Shannon’s limit that the only option is for cable designers to create more bandwidth in the cable.
Creating more bandwidth means deploying a new cable system – so the operator gets to choose the technology they are starting from. But what are the options for next-generation cables?
Classically, the industry uses the conventional or C-band, which is typically described as the wavelengths between 1530 nanometers (nm) and 1565 nm. The C-band is an extremely efficient place to transmit long-distance optical signals, and it’s also very familiar to the entire industry – two factors that conspire to make it very attractive to simply keep using C-band amplification and push transponder technology to deliver more capacity. But as this is getting increasingly difficult, it makes sense to look at the next band of wavelengths in the fiber – the long or L-band waves, from 1565 nm to 1625 nm.
As Figure 3 shows, we can achieve double the capacity per fiber pair using C+L-band – in this case, the example uses a goal of 80 Tb/s for the entire cable. If each fiber pair can deliver 20 Tb/s of capacity using C-band only, then we could either deploy four fiber pairs with C-band-only amplification, or two fiber pairs with C+L-band amplification.
So…could we install four fiber pairs, light up the C- and L-bands, and double the total cable capacity to 160 Tb/s? The answer may be no for a given cable, because the amp chain has to be powered by large electrical voltages at each end of the cable. The practical properties of the cable conductive layer limit just how much voltage and amp power can be supplied along the cable.
An alternative approach is to ignore the L-band, dial back the C-band amplifier power and tune the transponder modulation to something that is much more efficient when it comes to optimizing capacity and reach. Doing this will actually reduce the capacity per fiber, but it will also reduce the electrical power requirement of the amp chain for that fiber.
This will allow more fiber pairs to be deployed in a single subsea cable while remaining within the electrical power budget. Figure 4 gives an example in which it may be possible to run each fiber pair with a less demanding modulation and achieve 15 Tb/s per pair. But because it’s possible to power the amps for 20 fiber pairs, the total cable capacity is now 300 Tb/s, compared to a nominal 80 Tb/s if we had tried to extract the very maximum capacity from each fiber pair.
This technique is called spatial-division multiplexing (SDM), and it could open up a pathway to “petabit”-capacity subsea cables in the future. It’s always hard to tell, but I would say SDM was probably the hot topic at SubOptic this year.
SDM can use shared erbium-doped fiber amplifier (EDFA) pumps to help control power requirements, and less expensive optical fiber to offset the need for more fiber pairs. Ironically, the lower capacity per fiber pair may even be a commercial advantage, as it helps reduce the need for spectrum sharing as a mechanism for spreading the cost of a high-capacity fiber pair between multiple cable investors, and provides for more flexible fiber routing architectures.
Regardless of which of these starting places most applies to your subsea cable, it is essential to have the highest performance and most flexible transponder technology you can buy. To find out more about Infinera’s Infinite Capacity Engine technology, check out our website, and don’t forget to look in on the other topics discussed at SubOptic 2019.