ROB SHORE INFINERA Q&A
And all of this came out of Infinera’s own R&D centre?
So, what’s the next big thing in optical networking?
the effective bit rate of each individual subcarrier independently, which resulted in really tremendous performance improvements. When you use any kind of signal, particularly when you use sub carriers, the outer portions of a signal degrade faster than the inner portions of a signal, especially when you’re going through filters and amplifiers. If you just have one giant carrier, you have to adjust entire carrier up and down based on that performance of the worst part of the signal, so with subcarriers, you can actually adjust the performance of each individual sub pair. This means you can reduce the effective bitrate of the outer sub carriers, while increasing the effective bitrate of the inner sub carriers. So even when the outer ones start to degrade, you can maintain 800G over longer distances because while those outer ones have to be reduced, the inner ones can be increased. Also, and I know this is a particular differentiator from some of our competitors, we really recognise the value of probabilistic constellation shaping. The problem is in the 16 nanometre generation of DSPs. You just don’t have enough processing power on the DSP to take full advantage of probabilistic constellation shaping. So we learned not only from ourselves, but also from our competitors sometimes, and we made sure we added enough processing power right at seven nanometre technology. We used around five billion transistors which means DSPs have the extra processing power dedicated to probabilistic constellation shaping. This means we can use long code words and much bigger bit rates which take more processing but, give you that extra performance. Then add that to the ability to apply it across every individual sub carrier independently, and you end up with a massive improvement in performance. As we discussed, we’ve just completed our first customer trial of ICE6 and there are more scheduled for the first half of this year. These trials will include ultra-long reaches at 600G and 400G. We’ve already trialled 600G and achieved well over 1600km. We’re confident we’ll be able to achieve distances beyond 2500km, but we’re waiting until we can truly test its limits before announcing anything. We are still on track for entering production network deployments in the second half of the year.
OC
OC
Yes, we call it the Optical Innovation Centre, and I would contend that we have the most
I think the greatest inventions are not evolutionary but revolutionary, because they’re the ones that
RS
RS
completely vertically integrated capabilities in the industry. There are three fundamental components that go into building a coherent optical engine. There’s the DSP itself which we’ve designed in-house from scratch. Then there’s the actual electro-optical module, which converts the electrical signals into the optical signals that we not only designed in-house but which we also manufacture in-house. I think we have one of the very few large-scale Indium Phosphide fabrication centres in the world. Silicon photonics has its place, but it’s usually used for lower-end optical engines because silicon photonics-based lasers are usually low-power. For high-performance transmissions over long distances, it has to be Indium Phosphide. The DSP and the electro- optical module are two independent parts of the complete optical engine. The final piece to the puzzle is packaging. This is a lot more than just wrapping it in a box. There is RF interconnect, thermo- electric coolers, wire bonding, component alignment, and so on. Those three functions – DSP, optics, and packaging - are all combined to create what we call the Optical Innovation Centre, which is the organisation responsible for all the optical innovations we’ve achieved, whether it’s the PICs, the DSPs and the use of Nyquist subcarriers. It is, of course, also the place where XR optics was born. If you think about ICE6 as being second-generation Nyquist subcarriers, then XR optics is third generation, because we can now not only control each subcarrier individually, which is what we’re doing with ICE6, but we can actually steer data streams into each sub-carrier independently and hence send each subcarrier to a different destination. We’ve learned so much about Nyquist subcarriers since our first use of them, and we just feel that it’s going to give us a huge advantage in this generation of optics, because we believe we will be able to do a lot more with those subcarriers than anyone else will be capable of doing, and certainly more than anyone who is using them for the first time.
change the way networking works. Coherent receivers were a revolutionary invention, as was photonic integrated circuits and Nyquist subcarriers. I believe however that the next big thing in optical networking will be XR optics. The idea that one laser can generate multiple signals and communicate with multiple destinations at different rates, thereby eliminating having to match lasers on both ends of a circuit, as well as eliminating the need to nail up bandwidth between two locations constantly – this is significant. XR optics has the near-term benefit of eliminating a lot of transceivers and intermediate electrical aggregation devices. However, once you start building a network from the ground up with this concept in mind, you will build it totally differently. This can enable things like dynamic pools of bandwidth that can be turned up and down as needed across the network. The biggest challenge from a marketing perspective with XR optics, however, is getting people to understand the full implications of the technology.
How long will 800G be adequate, given the rising demand for bandwidth? There’s definitely going to be at least one generation after 800G, which will probably be a 5nm
OC
RS
technology and we’ll probably call it ICE7, although we haven’t announced anything yet, but it will probably be in the 130GBaud range. There will be at least two years of ramp up, so probably around 2023 we could be where we are now with 800G, although that is speculative. But that’s about how long it takes for this kind of technology to iterate. We’re currently in the evolving stage of coherent transmission and we’re taking that technology to its limits, but it will have a limit. There will need to be something new eventually, but there will be at least one more generation of coherent. We have a number of optical experts at our Optical Innovation Centre working on new ideas though, so it’s not like we’ll have to wait for someone else to invent something for us.
We’re currently in the evolving stage of coherent transmission and we’re taking that technology to its limits, but it will have a limit.
www.opticalconnectionsnews.com
19
ISSUE 21 | Q3 2020
Made with FlippingBook - professional solution for displaying marketing and sales documents online