Fast Channels Make Light Work

In real estate, it’s location, location, location. In data and communications networks, it’s bandwidth. No matter how often the word is repeated, bandwidth remains smaller than the amount of data people want to transport. VoIP, video on demand, and Internet music file downloads are only three of many applications adding to network traffic. As more applications like these become popular and the cost of data communications continues to decrease, the volume will grow to occupy the available bandwidth however large it may be.

Often, fiber-optic networks are underutilized because of modulation speed and channel switching bottlenecks, not because of the basic fiber technology. Several initiatives are being pursued to deliver greater bandwidth at lower cost:
•?Upgrade from 10-Gb/s to 40-Gb/s and 100-Gb/s Ethernet.
•?Improve flexibility with reconfigurable optical add/drop multiplexers (ROADMs).
•?Use phase information to code more bits per symbol with differential phase shift keying (DPSK).

As one white paper explained, “Service providers want to use existing fibers and network elements to maximize revenues from infrastructure already in place; they want to improve operation flexibility, which translates into more efficient provisioning and increased revenues from new services to their customers; and they want to increase the density of the network elements.”¹

While these goals may make good business sense, they are technically challenging. According to the white paper, “Upgrading from 10 Gb/s to 40 Gb/s or 100 Gb/s has a significant effect on a network’s tolerance to optical impairments such as noise, chromatic dispersion (CD), differential group delay (DGD), and polarization mode dispersion (PMD) as well as the bandwidth of the filters present in the network. For example, upgrading from 10 Gb/s to 100 Gb/s means that to achieve the same level of service, the optical signal-to-noise ratio (OSNR) needs to be 10 dB higher and CD and DGD 10 times smaller.”¹

Another major trend is the growth of passive optical networks (PONs) that split as many as 64 fiber to the x (FTTx) drop cables from a trunk cable to extend high bandwidth fiber-based communications into residences. Construction and maintenance of these optical networks present resolution challenges to traditional optical time-domain reflectometers (OTDRs).

Figure 1. Anritsu MT9083 ACCESS Master

According to Hiroshi Goto, optical product specialist at Anritsu, “The company recently enhanced the MT9083B ACCESS Master OTDR Series to provide improved fault location for all PONs (Figure 1). The OTDR is optimized at the most commonly used pulse widths, such as 100 ns, to test single splitter applications up to 1×64 or closely spaced distributed splitters.”

Core network upgrade testing and PON installation/maintenance have different requirements. Even if a fiber’s characteristics support upgrading, network performance must be verified under a wide range of conditions including various traffic patterns as well as different combinations and numbers of ROADMs.

These kinds of tests need programmable traffic generation and sophisticated result monitoring that involve higher protocol stack layers. Because of the speeds and distances, sensitive PMD, DGD, and CD measurements also may be needed. In contrast, most FTTx troubleshooting is based on OTDRs and power meters.

High-Speed Upgrade Tools

The OSNR is measured according to IEC 61280-2-9 when you can assume that the noise level between and under adjacent spectral peaks is generally flat. When this assumption cannot be made, you have to find another approach. EXFO Electro-Optical Engineering’s Model FTB-5240S-P Optical Spectrum Analyzer uses a noise measurement method based on the differences in polarization and spectral characteristics exhibited by signals and noise.

The overall input signal is split into its parallel and orthogonal polarization components, which then are measured by the two channels of the optical spectrum analyzer. The insight that noise is not polarized and that signals are highly polarized isn’t new. However, a straightforward implementation of a measurement system based on this principle is subject to errors caused by PMD. Instead, the FTB-5240S-P uses what EXFO terms a differential spectral response measurement method that provides accurate noise estimation even when the signal bandwidth is almost as large as the channel bandwidth.

This kind of capability is important in the world of 40-Gb/s and 100-Gb/s dense wavelength division multiplexing (DWDM) channels comprising separate wavelengths that may have passed through different combinations of network nodes. As ROADMs are more widely deployed, this situation will be commonplace. Also, in systems with channels spaced very closely relative to their signal bandwidths, the underlying amplified spontaneous emission (ASE) noise is obscured. In both cases, the IEC 61280-2-9 noise measurement method no longer is applicable.

For time-domain measurements, EXFO has developed the PSO-100 Series of optical sampling oscilloscopes. Because the signal is sampled optically, relatively slow optical-to-electrical conversion is avoided until after the sample has been taken.

The resulting 500-GHz bandwidth is completely modulation independent, which means that this is a future-proof solution. It can handle today’s 40- and 100-Gb/s network upgrades and any bit rate up to 640 Gb/s that may be encountered in the future. The two-channel PSO-102, together with a suitable demodulator, can address more complex modulation formats such as DPSK and differential quadrature phase shift keying (DQPSK).

Anritsu’s MP1595A 40G SDH/SONET Analyzer provides 40-Gb/s and 43-Gb/s support for all network quality measurements including stress testing and jitter. The company’s Mr. Goto elaborated, “The spread of rich-content broadband services supporting HDTV, video on demand, and other applications demanding high data throughput is driving the development of 40-Gb/s core networks. The MP1595A is an all-in-one solution for ensuring the quality of these networks.”

In contrast, the Ultra OC-3/STM-1 PCI Card from GL Communications specifically addresses medium-speed communications networks. Up to the OC-3 688-Mb/s rate, it allows a PC to perform SONET/SDH analysis, testing, simulation, and monitoring. You can add and drop signals to and from an OC-3/STM-1 signal as well as generate BERT patterns in all framing modes. Interfaces include optical OC-3/STM-1 and copper STS-1 and T1/E1. Data is transferred to the PC via a four-lane PCIe bus.

Agilent Technologies provides a range of equipment used in the development and testing of optical network components. One example is the Model N437x Lightwave Component Analyzer. The last digit in the model number changes to indicate single mode (SM) or multimode (MM) and modulation frequency range. These instruments measure the performance of electrical-to-optical components such as modulators, lasers, and LEDs. They also deal with photo diodes and receivers that are optical-to-electrical components. Optical-to-optical devices can be characterized as well.

Figure 2. Agilent Technologies
N4391A Optical Modulation Analyzer

The Model N4391A Optical Modulation Analyzer combines an optical polarization-sensitive coherent receiver with Agilent’s 90000 Series Infiniium Scope and a special version of the company’s 89600 Vector Signal Analysis Software (Figure 2). With this equipment, engineers can measure optical transmitter performance using the same types of metrics that have become well established in wireless communications.

Eye diagrams, constellation diagrams, error vector magnitude, and separate I and Q analyses are available. In addition, narrowband high-resolution spectrum as well as spectrogram displays are developed. And, based on the type of modulation being used, a symbol table with demodulated bits and error statistics can be presented. These capabilities extend to 40 Gb/s and, depending on the modulation scheme, to 100-Gb/s optical transmitters.

FTTx Tools

Power Meters
For basic or Tier 1 fiber certification, a power meter and source directly determine the overall loss for each fiber. This testing can characterize fiber loss at several wavelengths if multiple sources are available. In contrast, an OTDR provides extended or Tier 2 measurements at each splice and connector. Overall loss is measured indirectly by an OTDR.

Fluke Networks’ SimpliFiber™ Optical Loss Test Kits offer six combinations of tools for fiber installation and troubleshooting. The basic FTK150 kit has only the SimpliFiber power meter and a dual 850-nm/1,300-nm source. The other kits add 1,310-nm and 1,550-nm SM sources, a visual fault locator, and a 200x or 400x FiberViewer to inspect terminations.

Jason Tarn, product marketing manager at the company, said, “SimpliFiber Pro allows a single technician to quickly perform tasks that previously required a two-person team. In addition, the SimpliFiber Pro’s capability to perform loss testing simultaneously at dual wavelengths and save both measurements into one record increases efficiency by cutting test times in half.”

OTDRs
An OTDR launches a short pulse of light into an optical fiber and measures the backscatter from it. Detected event timing and amplitude are analyzed to locate fiber discontinuities. These could be breaks, splices, connectors, or just sharp bends, but if the associated loss is too great, data integrity can be affected. Tier 2 testing with an OTDR allows an installer to certify that the loss from each splice and connector falls within the allowed limits.

An OTDR’s dynamic range determines the smallest signal it can detect. Typically, this is equated to a maximum SM or MM distance. For example, if a SM fiber attenuates 0.2 dB/km at 1,550 nm and has splices every 2 km with 0.1-dB loss each, then a 35-dB dynamic range allows fibers up to 120-km long to be tested. Checking the math, 120 x 0.2 = 24 and 60 x 0.1 = 6, so the total loss is 30 dB. EXFO recommends allowing an extra 5-dB to 8-dB margin, hence the 35-dB dynamic range of this example OTDR.

Because the backscatter detector is very sensitive, it can become overloaded by strong reflections. If it does become overloaded, it cannot respond to events during the period of time it takes to recover. A detector cannot even recognize that another event has occurred during the event dead zone. It can recognize an event but not be capable of accurate measurement during the attenuation dead zone. These short time periods are expressed as dead-zone distances.

Light propagates at about 2E8 m/s in a fiber, so a typical 100-m launch cord is equivalent to 500 ns. Pulses as long as a few hundred nanoseconds can be launched while allowing sufficient time for the detector to recover from the launch and still respond to the actual fiber connection. Longer pulses increase the amount of backscatter, which corresponds to a greater fiber length that can be examined, but the attenuation dead zone also increases as does the distance uncertainty.

The shortest dead zone corresponds to the shortest pulse width. For example, an EXFO datasheet footnote explains that both the quoted event and attenuation dead zones are typical values for reflectance below -45 dB using a 5-ns pulse.

Several EXFO OTDRs are similar to each other but optimized differently. The FTB-7600E ultra-long-haul model with a 50-dB dynamic range characterizes SM links up to 250 km in length. It has an event dead zone of 1 m and an attenuation dead zone of 5 m. The 7200D, 7300E, 7400E, and 7500E models have specifications tailored for access and LAN/WAN, FTTx-PON/multidwelling unit (MDU), Metro, and Metro/long-haul applications, respectively.

Anritsu’s Mr. Goto described the company’s MT9090A Network Master OTDR. “It provides 5-cm resolution and dead zones less than 1 m and has a built-in 10-m launch fiber to ensure everything is evaluated. When equipped with a 780-nm fault locator, the MT9090A can troubleshoot in-service FTTx networks without costly filters and without disruption to other customers.”

The MT9090A has a 1-km to 10-km range and <1-m event and <5-m attenuation dead zones. You can set the index of refraction from 1.3000 to 1.7000 in 0.0001 steps for the particular fiber being tested. The pulse width is less than 10 ns so the integral 10-m launch cord is more than adequate.

Anritsu specifies the Network Master as a series of products that comprise the MT9090A mainframe plus another module. For FTTx drop cable fault location, the MU909011A module provides automated installation verification. The MU909020A module is optimized for CDMA, and the MU909060A1/A2/A3 modules adapt the Network Master for Ethernet testing.

Platforms

Basically smart chassis, platform-based instruments can be configured to address a wide variety of applications by selecting the appropriate mix of plug-in modules. Anritsu’s MD1230 family of products is one such platform that the company promotes as “the solution of choice for all your next-generation network measurement needs.” A broad statement, but it does appear to be reasonable for most core, Metro, FTTx, and Ethernet/LAN applications as fast as 10 Gb/s.

The portable MD1230B Data Quality Analyzer at 15 kg is large enough to have an integral 8.3″ SVGA display and slots for five interface modules. The rack-mount MP1591A has similar analysis capabilities but weighs 28 kg, supports 16 modules in A and B banks of eight slots each, and is controlled by and displays results on an external PC. Weights for both models are excluding modules and options.

The MD1230B accepts a range of copper and fiber Ethernet modules, some with options that support clock measurement, power over Ethernet (PoE) analysis, and link-fault signaling. The MP1591A can use any of these modules but also has dedicated slots for bank A and B control modules and an expansion interface module required when using bank B slots.

Both mainframes can be ordered with several options that support different control interfaces and test configurations. Mr. Goto said, “The MD1230B network data analyzer has been developed for both network troubleshooting and device characterization in the lab. A unique capability is the provision of up to 20 test ports, which means that a single analyzer can measure 10 Gb/s systems with 16 splits.

“Recently, we developed traffic impairment emulator (TIE) application software for the MD1230B. It captures video and voice data on a network and generates delay, packet jitter, packet loss, and errors so technicians and engineers can accurately reproduce complex faults and verify network impact,” he continued. “With the emulation software, the MD1230B performs highly accurate, realistic simulation of complex network faults and loads, allowing real-time verification of traffic-delay tolerance as well as evaluation of the impact of errors on video streams and other IP-based traffic.”

Fluke Network’s hand-held MetroScope™ Test Tool provides optical power measurement, traffic generation, and packet jitter testing for up to 1-Gb/s Ethernet. Further, the built-in FrameBERT function supports dark fiber qualification.

“MetroScope handles both copper and fiber testing in a single device,” said Mr. Tarn. “This allows testing of carrier Ethernet over copper or fiber and adds the flexibility to test the customer’s side of the network, which reduces time-consuming and costly finger-pointing between the end user and service provider.”

EXFO’s FTB-500 platform communicates with you via its 12.1″ touchscreen display and supports optical as well as transport and datacom testing (Figure 3). The range of modules includes OTDRs, optical spectrum analyzers, and a PMD analyzer in addition to SONET, Fibre Channel, and Ethernet test capabilities.

Several types of I/O ports and an internal 80-GB hard drive are standard. Optionally available storage media include flash USB drives, an ExpressCard memory card, and an external USB read/write DVD drive. The mainframe also can be ordered with the PM-500 power meter and a visual fault locator.

Digital Lightwave’s original Network Information Computer (NIC) was one of the first scalable platforms designed to support a wide range of testing applications through a number of interchangeable modules. The 40/43G Testing Module working with additional modules in the NIC Plus chassis provide a single-box solution for 10/100/1000/GigE/10GigE and NextGeneration testing. In addition, the 40/43G module has a tunable C-band or L-band laser option.

For physical layer analysis, an NIC-compatible Optical Spectrum Analyzer module handles wavelength, power level, and OSNR testing for DWDM and coarse wavelength division multiplexing (CWDM) networks. The spectrum analyzer measures C-band and L-band parameters relative to a -60-dBm noise floor with ±60-pm wavelength accuracy.

Yet additional ranges of platform-based instruments are available from JDS Uniphase (JDSU). The MTS–8000 Scalable Optical Test Platform features agile optical network (AON) test capabilities. Broadly, this term relates to the dynamic nature of networks with ROADM switching capabilities. Compatible OTDRs and CD and PMD modules are available.

JDSU also provides a fully updated version of the venerable T-BERD, now called the T-BERD 8000 Scalable Optical Test Platform (Figure 4). This instrument name has survived changes of ownership from TTC, a merger with Wavetek Wandel Goltermann resulting in the formation of Acterna, and now JDSU. The T-BERD 8000 addresses fiber characterization through OTDR and optical spectrum analyzer modules, CWDM/DWDM and transport layer test, and 10-Gb/s Ethernet test.

Summary

Suitable PON and core network test tools can be listed hierarchically, starting with a power meter. Several brands of hand-held instruments are available, many with adaptors allowing their use in both copper and fiber network installation. For fiber-link testing and installation verification, an OTDR is indispensable. It displays a graph of loss vs. distance, which highlights poor quality splices, loose connectors, and physical damage to a fiber.

For very high-speed networks, fiber imperfections such as CD, PMD, and DGD must be measured, and this requires specialized equipment. Rather than buy instrumentation that only can address these issues, a scalable platform might be more appropriate. In addition to measuring physical parameters, by changing to a different mix of modules, you can generate BERT patterns or perform stress testing with predetermined traffic profiles.

It’s not surprising that a fiber’s physical imperfections should affect signal integrity. When close channel spacing, wide signal bandwidth, and variable routing are considered as well, achieving a good QoS is far from certain. Actually testing each fiber’s performance at higher data rates and with a variety of traffic must be done to ensure that the required QoS can be provided.

Reference

1. Abou-Shaban, M., and Côté, M., “Testing Efficiency and Challenges When Upgrading to 40 Gb/s,” Application Note 214, EXFO Electro-Optical Engineering, 2009.

September 2009