The Impact of Dispersion on Today’s Networks

With a reliable fiber foundation being critical to support all types of communication including 5G, it’s important that networks evolve seamlessly to support the higher speeds and major bandwidth demands of today’s data-intensive world.

A significant consideration for today’s networks is dispersion, which can be the enemy of high-quality transmission. Systems operating at under 10 Gbps are not really affected. But at higher speeds, chromatic dispersion (CD) and polarization mode dispersion (PMD) can impede transmission. These speed-limiting occurrences threaten the integrity of the signal which in turn affects quality of service (QoS).

Let’s answer some of the more common questions about the impact of CD and PMD on today’s networks.

What are CD and PMD?

CD and PMD are two phenomena that have different impacts – but ultimately both can adversely affect performance.

CD is a natural characteristic of optical fiber and relates to the fact that wavelengths (colors) travel within the fiber at different speeds, which results in dispersion or pulse spreading. When purchasing fiber, you have access to “natural” dispersion fiber or fibers where the dispersion curve has been shifted to reduce the impact at certain wavelength ranges.

To simplify, PMD occurs when the core of the fiber is not perfectly round, which causes the two different axes of the light (say, vertical and horizontal) to travel down the fiber at different speeds. This disparity potentially induces a bit overlap and impedes the signal. A good fiber should have almost zero PMD and, if not, the faulty sections must be mitigated for optimum network performance. When you buy fiber-optic cable, you always expect the PMD to be as low as possible.

So would any high-speed fiber-optic network experience CD and PMD?

Yes, CD is a characteristic of the fiber that is unavoidable but can be managed by selecting the proper transceivers. However, retaining some level of CD is important to prevent other non-linear effects from happening. So even if it could be done, network operators would not want to have zero CD on their fiber.

Statistically, PMD will happen more often on older fiber-optic cables, but even recently deployed new cables have shown out-of-specification PMD values. These fiber sections with high levels of PMD must be detected, isolated, and replaced, as the goal is to reduce PMD to a minimum.

Are there types of fiber that can reduce these dispersion factors?

G.653 zero dispersion-shifted fibers are optimized to reduce CD – but they do not support DWDM transmission and are therefore not a popular choice. There are also G.655 non-zero dispersion-shifted fibers that have a lower CD but not null around 1550 nm -- so they are optimized for long distance but are more expensive. However, most fibers deployed in today’s networks are standard G.652 optical fiber with “natural dispersion” which is cost-effective and easy to manufacture.

PMD could be mitigated during fiber manufacturing by:

• providing a perfect homogeneity of the dopant
• controlling the cylindricity of the core
• adding a spinning step in the drawing process.

What other techniques can network operators use to reduce CD and PMD?

Back in the day, CD was the main limiting factor to achieve long distance, and compensators using negative chromatic dispersion were developed to avoid regenerating the signal using an optical/electrical/optical (OEO) regeneration site.

Now, rather than having to deploy fiber with dispersion-shifted CD, transceiver manufacturers have developed new modulation techniques and tricks like negative chirping and forward error correction (FEC) to deal with a small amount of dispersion. Short- to medium-reach transceivers can use these techniques, and it is very important to select the right transceiver flavor for the characteristic of your network fiber type.

For longer distances and faster rates like 400 Gbps, post compensation with digital signal processing developed for coherent transceivers can achieve more in managing CD and PMD to a certain extent.

When does it become critical for network operators to measure CD and PMD?

Today’s higher-speed networks carrying intensive amounts of data and mission-critical services have zero tolerance for service interruptions. It’s imperative that network operators pay attention to fiber characterization to ensure QoS.

Measuring CD and managing PMD should be a priority for network operators, and they shouldn’t wait until service-affecting problems arise. Also, the rate deployed today will be superseded soon with always faster rates, so if you want your network to be ready for the next wave of evolution, fiber characterization is critical.

What are best practices for managing CD and PMD?

Knowing the kind of fiber that’s deployed in the network is paramount. If a network operator has a typical 80% standard G.652 fiber, they need to know what type of fiber comprises that remaining 20% and where such fibers are. Otherwise, when they move to higher-speed transmission, there could be some issues during the “turn-up” or poor QoS to the end customer.

Today’s test technology enables network operators to determine the fiber PMD throughout the network. And it is entirely possible to isolate specific fiber spans having high PMD.

Is there anything unique regarding CD and PMD as they relate to 5G networks?

The earlier point about the importance of reliability to avoid service interruptions is even more relevant in 5G networks. Imagine mission-critical activities like using a self-driving car, remote surgeries, or world-class sports or entertainment broadcasts being interrupted mid-stream. That’s totally unacceptable and can be avoided with the diagnostics mentioned here to isolate fiber cable issues and ensure they are addressed, such as replacing faulty spans.

It’s the same as if you have a bad connector. You don’t buy a more powerful amplifier to compensate for the loss of power due to the bad connector, you go into the field and clean or repair the connector. The same applies to ensuring the integrity of the fiber from end-to-end within the network.

PMD is a tricky phenomenon because even if the average PMD delay is not changing over time, specific wavelengths will see specific delays at specific times. So you can have different delays on the same wavelength measured at different times, or a peak of delay moving from channel to channel. This is very challenging to troubleshoot. An operator might feel confident that their mobile network operating at 25 to 50 Gbps is operating reliably, but an hour, a day, or a month after the activation, the differential group delay (DGD) can change on a specific channel and interrupt the service. Therefore, it’s especially important to measure PMD before activating services. Don’t just plug and pray!

How will transceivers likely evolve because of CD and PMD?

While coherent transceivers have helped make strides in managing CD and PMD, developing next-gen transceivers to suit 25- and 50-Gbps mobility is a real challenge if you want the cost to remain acceptable. Digital signal processing can be implemented to reduce the impact of dispersion, but that can increase cost.

People sometimes assume that CD and PMD are no longer a problem, but that’s absolutely not the case. Ignoring that fact can be costly for network operators because of having to fix incidents that could have been avoided. There is also potential exposure for QoS and service-level agreements.

Won’t coherent transmission in high-speed networks make CD and PMD concerns obsolete?

Coherent transmission – which essentially means using coherent transceivers with digital signal processing (DSP) to support high-speed networking of 100G, 400G and 800G – does not eliminate completely the need to test for CD and PMD.

For CD, the gating factor is knowing the type of fiber deployed in your network. If you have a greenfield network within which you have deployed a known fiber type, then selecting the right type of transceivers to match the fiber type means there is no need to test for CD.

On the other hand, deploying any transceivers on unknown fiber type can lead to malfunctions and errors. Many network operators have acquired fiber networks over the years where documentation and fiber type are missing. As an example, deploying long reach coherent transceivers and amplifiers on dispersion-shifted fiber (DSF) could generate nonlinear effects that would degrade the QoS.

In any network using any type of fiber, PMD continues to present a risk. Remember I noted earlier that even new cables can have out-of-specification PMD values. To mitigate that risk, testing is a must, even if you have a high-speed network relying on coherent transceivers.

That’s because PMD can always be present, and the only way to know if PMD is at a level that won’t impede transmission is to test. Not all coherent transceivers are equally tolerant to PMD, and an outage even for one second at the 800G rate means losing 800 billion bits of data. That’s not a risk most network operators want to take by reverting to plug and pray.

The bottom line: When coherent transponders emerged 15 years ago, we never stopped testing CD and PMD because these effects could still impact a transmission, even with digital post compensation. CD and PMD continue to be real considerations that should always be taken into account as high-speed connectivity and 5G networks proliferate.

Gwennael Amice is subject matter expert and senior member of technical staff at EXFO.