The public's thirst for ever-increasing bandwidth, driven by streaming video and other bandwidth-intensive multimedia applications in both the wireline and wireless space, has resulted in service providers ramping up the capacity of their networks. While the network equipment technology is available to support the required higher data rates, the existing optical network is not necessarily ready to accommodate it. Optical network anomalies such as chromatic dispersion (CD) and polorization mode dispersion (PMD) can create havoc on high-speed networks.

While cost-effective commercial solutions to compensate for CD have been developed and are currently being deployed in networks, for optical transmission speeds of 10 Gbit/s, 40 Gbit/s or beyond, PMD remains a major problem for network carriers. In particular, legacy fibers installed at a time when PMD was not of concern for the prevalent bit rates often exhibit such high PMD values that they are not suitable for upgrade to the high bit rates of contemporary backbone optical networks.

There are two solutions to this PMD problem: either to use "PMD-tolerant" transmission systems, for example PMD compensators or advanced modulation schemes, or to renew the fiber infrastructure by replacing those links whose PMD values exceed the limit for high-speed transmission. While the former solution "goes around" the problem and delays its resolution, the latter fixes it, however raising a critical question: how much will it cost?

This article focuses on the possible approaches to dealing with excess PMD, and highlights the preferred approach from a cost-effective standpoint.

Finding a Solution - Sooner or Later
PMD is typically less frequent in newer fibers than in older ones. However, when bandwidth is required, all available fibers will need to contribute. And today this is more meaningful than ever before, for several reasons:

• ROADMs make the network reconfigurable, so all fibers are solicited.
• In most cases, good low-PMD fibers have already been upgraded to high-speed.
• Service providers' #1 priority is margin, generally ruling out costly upgrades.

Once it has been established that available fibers have PMD levels that exceed the thresholds, three approaches can be examined:

Approach #1: Locate Another Suitable Fiber
When the intended fiber displays excess PMD, the simplest, obvious workaround is to test and re-test other fibers until a suitable one is found. This strategy is extremely time-consuming and costly. Moreover, if this is not the first high-speed upgrade being performed, it is very likely that most low-PMD fibers are already used, which leaves very few options.

Another scenario that commonly occurs when low-speed services were deployed is the process of selecting fibers in sequence: choosing them without regard to specific characterization measurements other than basic optical measurements such as optical return loss and power levels. Since PMD and CD affect transmission only at 10 Gbit/s and beyond, testing for PMD and CD was a non-critical issue in low-speed deployments. If the few fibers still available exhibit high PMD, using other in-service fibers becomes even more problematic, since it forces network operators to perform time-consuming actions that imply loss of revenue and intense planning and that they would rather avoid, such as rerouting and decommissioning.

After all these efforts, the best outcome attainable with this approach is to "hopefully" identify a suitable fiber. And even then, the "bad" fiber stays unusable for high-speed transmission, and time will come when it will have to be used anyway. Conclusion: the problem has not been solved, only deferred.

Approach #2: Installing New Fiber
While digging trenches and laying out new fiber can be a good long-term investment, it remains extremely expensive. Metro areas tend to cost more than rural, however the costs can range between $50 and $125 per meter, which rounds up for a 60 km deployment to somewhere between $3 million and $7.5 million. In addition, such an endeavor takes a lot of time and planning, delaying services and leading to loss of revenue. Conclusion: more expenses with less revenue are usually not service providers' dream solution.

Approach #3: Mitigating the Problem
Pinpointing the cause of PMD on a fiber span is a difficult assignment. By the statistical nature of this optical phenomenon, PMD distribution along a fiber is very seldom uniform. Most of the time, one or two faulty sections will contribute to the bulk of the total PMD. Therefore, accurately pinpointing these sections allows service providers to perform local fiber upgrades or rerouting, optimizing their most valued asset: the network itself.

This is how it works: from a central office (CO), or any test point, one technician tests for PMD using a single-ended unit to assess the total PMD of the link. For the sake of argument, let's assume that the total PMD exceeds the acceptable limit for the service required to be transmitted.

At this stage, that sole technician can travel to the first connectorized portion of the link, typically a second CO through which the fiber is passing. Performing a second test from this location gives the PMD of that same fiber, less the PMD of the section removed from the test. A simple subtraction then reveals the PMD of the first section. This can be re-done any number of times, as long as there are connectorized sections, to hopefully find a section that contributes massively to the overall PMD.

The example shown in Figure 1 shows how the faulty section is identified between locations 1 and 2 with three tests.

This technique does require a bit of time and traveling (only from a single technician), but it allows the technician to find sections that contribute high PMD, and these sections can then be bypassed or replaced, significantly decreasing the total link PMD.

No Need to Break the Bank
While the mitigation technique allows for pinpointing a faulty section between two connectors, this section often consists of several spans spliced together. An even more powerful and less expensive way to resolve the PMD problem is to identify the one or two faulty spans within the faulty section. To accomplish this, the ideal solution would incorporate a cost-effective test instrument capable of identifying and quantifying those localized "bad sections" with high PMD, a polarization optical time-domain reflectometer (POTDR), also called a distributed PMD analyzer.

Now available on the market, field-optimized distributed PMD analyzers are designed to do exactly that, deliver a robust, reliable assessment of each span's PMD contribution and pinpointing the faulty sections. Compared to replacing the entire fiber link, upgrading or rerouting on just a few kilometers both fixes the PMD problem and provides a fiber suitable for transmitting at 10 Gbit/s, 40 Gbit/s and beyond.

One such instrument is the FTB-5600 Distributed PMD Analyzer by EXFO. It is a single-ended instrument that requires only one technician with minimal technical expertise to troubleshoot the fiber. It not only measures the PMD from the far-end connector, but uses distributed Rayleigh backscattering to actually measure cumulative PMD on the entire fiber link.

Figure 2 shows a sample reading where high-PMD spans are found in a few minutes. Replacing or rerouting around the fiber at the 4.4 km and 9 km splice points brings the PMD measurement within acceptable limits for high-speed transmission. The unit gives the contribution of each section relative to the entire PMD value, and also can simulate the replacement of a faulty section by a good one, facilitating network planning and upgrades.

Negligible at low speeds, PMD can quickly become a hindrance when it comes to upgrading networks to 10 Gbit/s and 40 Gbit/s transmission. And until recently, network operators could ponder only two options to address high-PMD issues: crossing their fingers that they can find other suitable fibers already installed, or emptying their wallet to replace the whole link.

About the Author
Francis Audet, Eng., is Senior Product Manager, EXFO. He has more than 9 years experience in network deployments and dispersion testing. Robert Conway is Senior Account Manager, EXFO. He has more than 25 years experience in the sale of network infrastructure elements and test and measurement systems to Tier 1 service providers. For more information visit www.exfo.com.

What is your experience with this? Tell your fellow readers now!