contact button

Inside the Verizon 800G Field Trial

August 4, 2020
By Mark Snyder
Director, Systems Engineering

As Infinera’s lead System Engineer on the Verizon account, I have been called on over the years to model coherent DWDM performance on a wide range of optical fiber types and networking situations.  The fact is that fiber is an asset that remains in the ground for years and even decades, while the coherent optical transmission technology that rides on this fiber evolves much more rapidly, with two- to three-year technology transition cycles.  While optical fiber like G.655 large effective area fiber (LEAF) was not designed with coherent transmission in mind, it exists in some portions of service provider networks, including Verizon’s.  My team and I sought to take on this LEAF fiber challenge and put our 800G next-generation coherent optical engine, ICE6, to the test as part of our ongoing collaboration with Verizon.

Changing the Game

For the past three generations of coherent technology, at 400, 600, and 800 Gb/s data rates, the conventional wisdom has been that optical reach at the maximum data rate was limited to 100 to 200 kilometers.  In fact, the vast majority of first-generation 400G-capable transponders are running at 300G, 200G, or 100G rates to extend their reach.

So, when it came time to envision our fifth-generation ICE6 optical engine, Infinera’s engineers really wanted to break this convention to help our customers radically improve optical reach to maximize fiber capacity and reduce transmission costs.  In addition, the team wanted to be able to do so in all kinds of optical environments and on all types of fiber because the world is heterogeneous.  This was the backdrop when we began discussions with Verizon several months ago about an 800G field trial.  Verizon engineers have been great partners over the years and identified challenging links and fibers in their network for our ICE6 engine.

After some discussion, we settled on three LEAF paths to test ICE6’s optical performance – running at 800G, 600G, and 400G rates.  Our modeling tools indicated that we should be able to close these links, so we felt confident shipping equipment and beginning installations, but I’ve never done a field evaluation where I didn’t get both excited and a little nauseous all at the same time, hoping that I didn’t miss something.

Performance Matters

So, how did we do?  In a matter of two days, we executed three separate tests over three different routes in the Verizon network.

  • Nashville to Atlanta, 800 Gb/s over a fiber length of 667 km
  • Atlanta to Memphis with loopback, 600 Gb/s over a fiber length of 2,283 km
  • Atlanta to Dallas with loopback, 400 Gb/s over a fiber length of 4,091 km

Due to the current pandemic, we limited Infinera on-site personnel to two of my colleagues, but I participated in live video calls each day leading up to the trial and during the trial itself.  It was so gratifying to see how quickly the GX G40 node came up, and we were able to close the first link on the first day, between Atlanta and Nashville at 800G and over 667 km.  We proceeded to close the Atlanta to Memphis and Atlanta to Dallas paths at 600G and 400G, respectively, in the following hours and on the second day to enable some overnight soak testing.

Verizon 800G Field Trial - Map

Why is ICE6 So Much Better?

What is it about ICE6 that makes its performance so much better than prior coherent generations or competitive alternatives?

As you can imagine, ICE6 has an array of advanced coherent features, but from my perspective working with the technology these past months, I believe ICE6’s performance advantages come from three key areas:

HIGH BAUD RATES PLUS NYQUIST SUBCARRIERS

We know we need to operate at ultra-high baud rates – close to 100 Gbaud.  But high baud rates normally result in problems such as a higher residual chromatic dispersion compensation penalty.  But ICE6’s Nyquist subcarriers, which frequency-divide the wavelength, mean we can mitigate those penalties.

LC-PCS

Second is our long-codeword probabilistic constellation shaping (LC-PCS) implementation.  Not only is LC-PCS incredibly flexible in terms of a granular transmission rate, but it is key to getting close to the theoretical capacity limit on the fiber.

DEEP VERTICAL INTEGRATION

Maybe the most important aspect of ICE6 is that Infinera designs and builds all three of its key technology components in-house in our Optical Innovation Center: the photonic integrated circuit, the analog ASIC, and the coherent DSP.  Designing and packaging them into the final module in-house means that we do not lose any of the individual components’ performance – especially at ultra-high baud rates.

At Infinera, we know that 5G and helping Verizon to maximize their existing fiber assets – no matter what type – are critical to their success.  That is why this trial is so important to them and us.  We are so excited to be a part of their network and this journey.