While today’s network equipment technology can support higher data rates, including 40 Gbps, existing optical links are not necessarily ready for the upgrade. Polarization mode dispersion (PMD) was not a major concern until recently, as networks began to operate at these faster bit rates, and legacy optical fibers often exhibit excessively high PMD at these speeds.

Of course, PMD is typically less frequent in newer fibers than in older ones, and solving the problem can temporarily be delayed by avoiding the high-PMD fibers. However, when bandwidth is required, all available fibers will need to contribute. And today, this is more meaningful than ever before, for the following reason: ROADMs make the network reconfigurable, so all fibers are solicited.

Eventually, the problem must therefore be fixed, either by replacing the whole fiber – a very costly proposition – or pinpointing the worst link sections and replacing only those.

Although financially attractive, the latter approach is hardly feasible using traditional PMD analyzers, which provide a total link PMD value rather than a section-by-section breakdown of the PMD picture.

However, newly introduced distributed PMD analyzers make it very easy to achieve, as they are specifically designed to measure PMD following an OTDR-like method, span by span.

Below is a case study that presents a situation in which a network operator tried to pinpoint the high-PMD sections of a link, without using a purpose-built distributed PMD analyzer. We then examine how opting for such an instrument would have made the assessment much more accurate and incredibly faster.

The manual solution
This article refers to an 88-kilometer fiber link that had a total PMD (end to end) of 6.2 ps. This link is part of a core network and transports voice, video and data to maximum fiber capacity. It has therefore become necessary to mitigate PMD because of the desire to transport 40 Gbps data rates.

Network technicians sectionalized the PMD by breaking the fiber into various sections and testing for PMD section by section. Knowing that PMD adds quadratically, they were able to roughly determine which sections contributed the most to the overall value of 6.2 ps.

This proved to be a very tedious and time-consuming task, as it took about a week to perform the work and had to be done at night (due to the fact that it is a working fiber).

The results gave the network technicians only a rough overview of where the network had issues. It would have been necessary to break and measure each splice-to-splice section in order to improve the results, requiring countless hours of testing and a significant increase in risk to active services.

An automated solution
The distributed PMD analyzer from Exfo (FTB-5600) completed the entire test in an hour (versus a week). There was no second-guessing of where to start and stop the routing of new fiber, as the test results are very precise.

Exfo staff first executed a quick scan with the FTB-5600 to verify the link for distance, loss and approximate PMD range. The operator then set the FTB-5600 to scan just the first 55 km and measured a total PMD value of 5.5 ps.

The results showed two main splice-to-splice sections that significantly impacted the overall PMD (very high coefficient). These sections contributed over 60 percent of the PMD seen on the entire 88 km route.

EXFO - Table 1

The results table above clearly shows which sections contribute the most to the link’s overall PMD. Section 7 (highlighted) contributes 14 percent, and section 10 contributes 63 percent, to a total PMD of 5.5 ps.

Using the post-processing capabilities of the FTB-5600 software, it is possible to perform a “what if” analysis of the data. The first step was to divide the offending sections using the “Section Edition” function, which allows the operator to adjust the section lengths down to the actual range of high-PMD fiber. The algorithm actually flags sections based on the locations of splice points, occasionally including acceptable fiber. 

After adjusting section lengths based on PMD alone, rather than PMD-plus-splice points, the operator then subtracts the highest sections using the “Estimation” function. Results can then be observed as if offending sections had been replaced with pristine optical fiber. The table below shows the results of that exercise.

EXFO - Table 2

We can note that after setting the offending sections’ PMD to 1 ps, the cumulative PMD is reduced to 2.9 ps. This exercise only required a few minutes to reveal the outcome of replacing two sections.

Additionally, since section 7 only contributed approximately 14 percent to the overall PMD, it is worthwhile to conduct the “what if” exercise by only removing the contribution of the worst section (section 10). The table below shows the outcome of this exercise.

EXFO - Table 3

It quickly becomes clear that replacing section 10 (6.8781 km of fiber) will reduce total PMD to 3.5 ps. Replacing both offending sections (sections 7 and 10, with a combined length of just over 8 km) will yield a PMD value of approximately 2.9 ps for the measured portion of the link.

The graphic below shows the offending sections.

offending sections

Using the most expedient method available at the time, the network operator manually sectionalized the problem down to an 88 km portion of the fiber link and allocated money to replace the entire section. Using Exfo’s FTB-5600 Distributed PMD Analyzer, the results have been refined and the operator may now elect to only replace 8 km of optical fiber – leading to substantial cost savings.

Assuming a cost of $2.50 per meter to replace installed optical fiber, the first method (i.e., replacing the entire 88 km link) would cost $220,000. With the second method (i.e., replacing only 8 km of fiber), the cost would be $20,000 – a savings of $200,000.

Moreover, this does not take into account the savings in labor associated with a crew working for one week at night to manually sectionalize the problem rather than using Exfo’s distributed PMD analyzer (which reduces that effort to one hour), a figure that could easily reach $20,000 for a crew. Therefore, total savings can be as high as $220,000 for this particular case.

It is worth noting that approximately half of this 88 km link was aerial fiber, which reduces replacement costs to a certain extent. Should the entire link have been buried or installed in underground conduits, costs would have been substantially higher.

Also worth noting: Actual replacement costs vary considerably and are dependent upon the environment. For example, replacing aerial fiber in a metropolitan environment may cost as much as $20 per meter, while replacing a similar aerial fiber in a rural environment is considerably less, as low as $2.50 per meter.