As service provides begin to deploy larger networks to increase bandwidth, new challenges are presented when measuring the optical signal-to-noise ratio (OSNR). When dealing with 40G and 100G networks, as well as with networks that contain reconfigurable optical add/drop multiplexers (ROADMs), the OSNR measurement reports often contain inaccurate results, which could result in potential network failures and higher OPEX.
TelecomEngine spoke with EXFO’s Jean-Sébastien Tassé, product line manager of the Optical Business Unit, about EXFO’s latest optical spectrum analyzers (OSAs)—the FTB-5240S and FTB-524BP—and what this means for the new challenges that next-generation networks will present.
TelecomEngine: The deployment of larger networks presents new challenges in testing. What are these new challenges, specifically when measuring the optical signal-to-noise ratio (OSNR)?
Tassé: To cope with consumers’ increased need for bandwidth, operators need to use novel technologies, such as 40 Gbits/s systems or reconfigurable optical add/drop multiplexer (ROADM). ROADMs are devices that can switch wavelengths in different directions. New networks present a unique challenge when someone wants to test the optical signal-to-noise ratio (OSNR).
OSNR has been around for a long time and has been recognized as one of the key parameters to assess the performance of a fiber span—since there is an indirect relationship between OSNR and bit error rate (BER): higher OSNR means lower BER. Therefore, ensuring that OSNR meets specific requirements enables operators to deliver top-notch services. Correct OSNR also reduces errors in transmission, downtime and truck rolls. In other words, properly assessing OSNR is a key ingredient to decreasing operational expenses.
Historically, people have used the IEC method to measure OSNR. IEC is a standards body (like IEEE or ITU) that publishes international standards in the field of electrotechnology. The issue is that the IEC method fails in 40G networks and ROADM-based networks. The technical details would be too long to describe, but in a nutshell, the IEC method fails due to the large spectral width of the 40G signals and due to the presence of filters inside ROADMs.
TelecomEngine: How do EXFO’s new optical spectrum analyzers, the FTB-5240S and FTB-5240BP, work to resolve these challenges?
Tassé: To tackle the problem of measuring OSNR in 40G and ROADM networks, the testing industry—including EXFO—has come up with new methods to measure the in-band OSNR, i.e., the OSNR inside each channel. Different companies have different approaches; EXFO’s is called "PROS OSNR", which stands for polarization-resolved optical spectrum OSNR. The PROS OSNR method is available in our FTB-5240S Optical Spectrum Analyzer  and FTB-5240BP Optical Spectrum Analyzer .
TelecomEngine: How do EXFO’s OSAs work with other EXFO equipment to test networks?
Tassé: All of our OSAs are modular, which means that they can be combined with other modules in a single testing platform. There are two companions that are typically used with EXFO’s OSAs: inspection probes and transport modules. Inspection probes are great tools to ensure that fiber connectors are clean and to avoid damaging the OSA. We should remember that about 80% of all network failures are related to dirty connectors or bad splices. This makes it a good point for testing networks with inspection probes. The other companion to OSA is the Transport modules, like EXFO’s FTB-8130NGE Power Blazer Next-Generation Multiservice Test Module  or FTB-8140 Transport Blazer 40/43 Gigabit SONET/SDH/OTN Test Module , which are used to test the protocol layer of different types of networks, but mostly optical transport networks (OTN) that support ODU0 to ODU4 and ODUflex. So, together with a transport module and an inspection probe, EXFO’s OSA becomes the ultimate testing tool for wavelength division multiplexed (WDM) networks testing.
TelecomEngine: How can operators improve their efficiency when troubleshooting WDM networks?
Tassé: That’s a very good question. At EXFO, we strongly believe that proper testing and rigorous procedures can significantly decrease the cost and duration of any troubleshooting operation. For example, let’s say that an alarm is received at the network operation center showing high BER for a given channel. What should be done? Typically, high BER is often related to low OSNR. What we recommend is going to the receiver because a measurement at this location takes into account everything that happened during the transmission. An OSA measurement of OSNR will reveal whether the alarm is due to low channel power or high noise, or both. In the first case, the faulty channel should be analyzed with an OSA at different places in the light path to find the location of the excessive loss. In the second case, each amplifier should be tested as they are the main contributors of noise.
We could go through several other examples, but the key to remember here is that spectral testing increases efficiency in the field when troubleshooting WDM networks and EXFO’s FTB-5240S/BP family is a great tool for reducing OPEX and increasing the likelihood of getting it right the first time.
TelecomEngine: Why are amplifiers so critical to a network? How has the testing of amplifiers changed with these optical spectrum analyzers?
Tassé: Signal power decreases as light propagates inside the fiber due to bad reflections and other phenomena. Amplifiers are installed at specific locations in fiber networks to boost signal power. The most common type of amplifiers are erbium-doped fiber amplifiers (EDFAs). Although EDFAs play a vital role in proper signal transmission, they also have a negative impact on OSNR since EDFAs generate additional noise on top of amplifying the input noise. Therefore, testing OSNR at the input and output of EDFAs becomes important, a task that can be performed with an OSA. Gain flatness (i.e., the gain uniformity in a given spectral region) is also a key amplifier parameter to measure with an OSA.
Even though recent amplifiers are certainly a lot more stable than older ones, they are still active components that are prone to aging and degradation. In fact, in recent interviews we’ve done with operators, many told us that amplifiers are one of the most common causes of WDM networks failures, along with dirty connectors and bad splices.
TelecomEngine: What is the importance of drift measurements and how are they tested?
Tassé: Drift refers to the change of wavelength, power or OSNR as a function of time. It is critical to measure drift at turn-up to ensure that the transmitters and the amplifiers are stable—that is to say that the transmitter wavelength and output power is stable, and that the amplifier gain is constant. A network without drift leads to significant reduction of errors, repeat jobs and truck rolls.
Measuring drift merely consists in connecting the fiber to an OSA, setting the OSA in Drift mode and let the OSA acquire for the desired period, often several hours. EXFO’s OSA provides a dashboard in Drift mode, providing an easy way to spot any signal irregularities over long periods of time.
TelecomEngine: With service providers building 40G and 100G networks, there are multiple bandwidths in any given network; why is this significant when it comes to testing, and what can be done to accurately test for this?
Tassé: The advent of 40G and 100G brings about new challenges for network planners. For instance, a fiber might transmit simultaneously 10G and 40G signals at different wavelengths. Network planners will often separate 10G signals from 40G signals with a guard band and will leave a part of the spectrum empty to avoid or at least reduce crosstalk. At the testing level, this coexistence of different data rates on the same fiber requires a few extra precautions. First, an OSA should be used to optimize the spectral width of the guard band in order to limit noise coming from crosstalk or non-linear effects. Second, the typical measurements of central channel wavelength, power flatness and OSNR still remain important to test.
Another aspect of 40G signals is that their spectral width is larger than that of 10G signals. For accurate OSNR measurements, the coexistence of 10G and 40G signals on the same fiber calls for a setup that is specific for each wavelength. EXFO will soon be announcing a product that will change the way OSNR is measured on a per channel basis.
Finally, we should not forget that a sizable part of 40G deployments are coherent, and that most of all 100G deployments will be coherent; these coherent rollouts bring new challenges for measuring OSNR that EXFO is looking to address in the future.