Nokia called out misleading 800 Gb/s fiber optic field trials in a blog post Tuesday, just days after the telecommunications vendor doubled down on 400 Gb/s optics with the launch of its own WaveFabric Elements portfolio.

"Carriers should expect 100-220 kilometer optical reach in most wave division multiplexing applications," wrote Nokia’s Randy Eisenach, discussing competitors’ 800 Gb/s offerings. "For carriers, it’s important to understand actual realistic network performance that can be expected when using high-capacity wavelengths, as opposed to demo experiments."

While Eisenach, who serves on Nokia’s high-speed optics and wave division multiplexing product management teams, didn’t name vendors, he’s referring to Ciena and Infinera, which have both announced 800 Gb/s optical portfolios.

Infinera has published field trials and demos of their latest optics, and Ciena has announced customer deployments. Both vendors have vehemently denied misrepresenting the capabilities of their coherent optics.

However, Nokia claims these trials don’t tell the full story, and it argues that 800 Gb/s speeds are next to impossible under real-world conditions using the current generation of digital signal processors (DSP).

According to Eisenach, this is due to a combination of the optical fiber type, fiber attenuation, signal degradation, and the need for higher-order modulations when working with 800 Gb/s wavelengths. Combined, these significantly reduce the distance over which an 800 Gb/s signal can travel.

"Vendors strive to differentiate their products and performance, but the reality is that most modern fourth- and fifth-generation coherent digital signal processors combined with leading optical component technologies (modulators, drivers, lasers) result in similar capacity vs reach curves across the industry," Eisenach wrote.

In other words, the race is much closer than the vendors’ field trials and marketing might suggest.

“It’s not about having the most powerful, awesome DSP, it’s about having DSPs that are tailored to the application at hand,” said Kyle Hollasch, director of product marketing for Nokia’s optical networking division. While Nokia certainly could have done 800 Gb/s optics, the market simply isn’t ready for it, he added.

Nokia says that while higher-order modulations are unavoidable at 800 Gb/s line rates, vendors can tweak the other variables by using low-latency fiber, eliminating spare margin, and removing guard band channels, to eke out higher performance.

"Using these field demo strategies can result in significantly longer 800 Gb/s optical reach than what is achievable in a real-world production network using standard SMF-28 fiber," Eisenach wrote.

He explains that spare margin helps account for degradation brought on by component aging, temperature variations, component variance, polarization-dependent loss, polarization mode dispersion, and non-linearity impairments. Eliminating spare margin can improve performance by 40% to 50%.

While Nokia accuses its competitors of rigging trials to achieve better performance, it remains unclear if the race is actually as close as Nokia contends.

In March, Infinera announced its sixth-generation optics had achieved 800 Gb/s line speeds across a 950-kilometer production network using standard optical fiber.

According to Infinera SVP of Marketing Rob Shore, that test was conducted on a live production network — no funny business intended. "We wouldn't have done a trial on anything that somebody wouldn't deploy," he said.

In the interest of full transparency, Shore admits that the trial in question was conducted on a network using a fairly aggressive margin, which attributed in part to the company’s performance. But he noted that this was also the standard operating margin for the network.

"From the most aggressive margin that this customer was using to the most conservative, you'd maybe lose 20% to 25% performance," he said. "The worst we could possibly do is 25% worse if the customer wants to be ultra conservative with how much margin they're reserving."

Accounting for a 25% loss on a more conservative configuration, Shore says the technology is still capable of pushing an 800 Gb/s signal in excess of 700 kilometers.

"These guys [Infinera’s competitors] really really want this not to be true, because it really significantly damages the benefits of their technology. They assume the worse case in all aspects of our trial and it just wasn't true," Shore said. "It wasn't the average deployment case, 100%, but it was a real deployment case at the far end of what a realistic deployment should be able to achieve."

Even using a narrow margin to bolster the capabilities of its optics, the performance achieved by Infinera is impressive, said Dell'Oro optical analyst Jimmy Yu.

Turning to Ciena, the company says it has never misrepresented the capabilities of its optics, calling Nokia's insinuations patently false.  The company maintains that its latest generation of optics, WaveLogic 5 Extreme, is capable of pushing an 800 Gb/s line rate over one to two span lengths — anywhere from 80 to 240 kilometers.

"Ciena has not made a claim significantly outside of the 100-220 kilometer reach that Eisenach himself mentions," a company spokesperson wrote in an email to SDxCentral.

However, when tuned to lower line rates, Ciena says its optics can achieve much longer distances. Without excluding spare margin or removing guard bands, the vendor claims carriers can expect to push a 600 Gb/s signal up to 1,000 kilometers, or a 400 Gb/s signal between 2,000 and 4,000 kilometers.

Editors note: This story has been updated to clarify the capabilities and claims made with respect to Ciena's 800 Gb/s-capable WaveLogic 5 Extreme optical portfolio.