Fiber Type Matters When Simulating Optical Links and Latency

Posted by Kevin Miller on Thu, Apr 2, 2020 @ 11:04 AM

Simulating real-world fiber optic links and time delays in the lab environment is both a frequent and necessary task for engineers performing R&D and equipment certification testing processes.  With the evolution to more advanced network architecture, increasing speeds of 400G and beyond, and latency always being a key element, replicating the field network as closely as possible in the lab is critical to ensure systems will perform as expected post-deployment. 

While a number of considerations must be taken into account for network simulation, today we're going to discuss why choosing the correct type of optical fiber is one of the most important decisions related to the general performance and time delay aspects.

Should My Test Setup Match The Fiber Type Used In The Field?

Working with network engineering teams, a common question that often arises sounds something like this:  "We have Corning® SMF-28® Ultra deployed in our network - do we really need to use that exact same fiber in our test setup, or will a G.652D fiber from any manufacturer be acceptable?"

The answer to that question depends on how closely and accurately you want your test setup to replicate your field network.  When seeking to exactly simulate or replicate your field network, the best practice is to always utilize the same fiber type in your test setup.  The reasoning behind this is because like any other manufactured product, different fibers have different performance characteristics resulting from variances in the manufacturing processes. 

A few important factors engineers must consider is fiber loss/attenuation, chromatic dispersion, and possibly others depending on your application.  Let's take a look at four G.652D fibers, each from a leading and highly reputable global manufacturer:

Fiber Manufacturer Maximum Attenuation @ 1550nm
A ≤ 0.18 dB/km
B ≤ 0.21 db/km
C ≤ 0.21 db/km
D ≤ 0.20 db/km

If you have type A in your 100km Tx/Rx field network link, using a different fiber type when simulating the same 100km link in the lab could result in your attenuation calculation differing by as much as 3dB per direction between the test and field network.  However, if you have type B in your network, you might be able to get away with substituting Type C if only considering the attenuation value.  Since attenuation is not the only performance factor related to fiber, it's always best to utilize the same fiber type whenever possible as a best practice, which will ensure the most accurate replication of general performance characteristics.

Pro Tip:  When seeking to acquire optical fiber for your lab, always inquire as to the specific fiber the network simulator vendor is utilizing to ensure it matches your needs.  A reputable vendor will be able to both share and verify the exact manufacturer brand(s) they are offering.

For Simulating Fiber Latency / Time Delay, Does The Same Concept Apply?

With latency being one of the most important performance factors considered by network engineers and architects, choosing and acquiring the correct fiber type is absolutely critical when exactly replicating links and their respective time delays in the lab environment.  This is especially true for any entities with latency-driven applications, from fiber providers offering low-latency routes to both enterprise and data center applications where optical time delay is critical (ex: financial entities, high frequency trading, web streaming services, etc)

Today, millions and millions of dollars are gained or lost by financial entities that are trading in fractions of a second - in some instances, not just microseconds, but now in even smaller increments like nanoseconds and picoseconds.  How does the type of fiber play a role in these types of time delays?  Let's look at an example using the fiber characteristics of two random manufacturers:

Fiber Manufacturer Fiber Index of Refraction (IOR) @ 1550nm
A 1.468
B 1.467

As shown in the table and in general, IOR often varies slightly between manufacturers simply due to differences in the way the fiber manufactured by each respective entity.  Since IOR is a key variable when calculating latency in an optical fiber, using a different fiber type from the one in the network will in turn yield a different time delay, even if it's a minimal change.  While an extremely slight difference is likely acceptable for most general communications applications, for those where latency is being measured in such minute fractions, this difference in values may not be acceptable for those seeking to exactly simulate the expected link performance.  Lastly, it's important to recognize that additional factors that many do not consider, such as temperature changes, can also have a further effect on the refractive index and resulting delays, so matching the fiber type is strongly recommended to achieve the most similar performance values.

Pro Tip:  When attempting to simulate or replicate fiber delay to a very specific value, don't forget to include the length of the patch cables you plan to use when connecting to your delay simulator. 

Best Practice - Match Fiber Types, When Possible

While it's always ideal to match the fiber type in your test setup to the field network, it is also recognized that factors such as budget constraints and fiber availability may require an individual to utilize an alternative fiber that meets the project needs.  If that is the case, choosing the alternative fiber is still an important task to ensure your results will be as accurate as possible. 

Whether you are an equipment manufacturer generating a data sheet with performance certifications or a data center supporting financial customers, choosing the correct fiber type for simulating link performance and time delays should be one of your priority concerns when specifying your test setup.


Best Practices for Simulating a Physical Fiber Network

Optical Fiber Latency Calculator

Topics: latency