Measuring chromatic dispersion helps optimize network performance

0301feat10 1

Arthur J. Barlow

Chromatic dispersion has become the principal challenge in modern transmission-system design. Techniques including dispersion-compensating fiber, new signal formats, and soliton-like behavior can be used to measure and mitigate the effects of dispersion.

With modern gain-flattened optical amplifiers overcoming virtually all the effects of fiber attenuation, chromatic dispersion is now one of the most significant limitations on the performance of dense wavelength-division multiplexing (DWDM) systems. Even without nonlinear effects, the power penalty caused by dispersion increases as the square of the bit rate. A variety of techniques, including dispersion-compensating fiber and components, new signal formats, and soliton-like behavior are now used to mitigate the effects of dispersion.

Many of these techniques were pioneered initially for submarine cable systems, where high bit rates and wavelength counts combine with transoceanic lengths between electronic regenerators (terminals). Now terrestrial systems are using similar techniques to remove expensive and restrictive electronics along the route. Solving the dispersion problem requires a two-dimensional conjuring trick: reducing total average dispersion to a small percentage of the total uncompensated value—not just at a single wavelength, but over the full width of the operating wavelength—usually the entire C or L band.

To achieve this feat requires ever-more precise dispersion management techniques, including schemes using different types of transmission fibers and lumped dispersion-compensation elements, leading to complex distributed dispersion maps. For terrestrial systems, in particular, cost is a principal factor inhibiting transition to higher-performance, more-complicated networks. A main element driving increases in cost is the additional time and effort required to manufacture, test, and commission end-to-end systems with such complex, precise dispersion maps. The key is finding fast and convenient means of measuring dispersion, not just of the transmission fiber but of complete sections of transmission systems with all the in-line optical components such as optical amplifiers, filters, and dispersion-compensation components in place.

Dispersion is a measure of how the system group-delay changes with wavelength. The most direct and effective way to measure dispersion is the differential phase-shift method. Two different wavelengths are transmitted along the system to be measured, each one amplitude-modulated with the same frequency sine wave. The dispersion is defined as the change in group-delay between the two test wavelengths, which gives rise to a difference in the phase angle between the two sine waves after demodulation. Dividing this by the difference in the two test wavelengths provides dispersion directly. No matter how dispersion is measured, a finite optical bandwidth is required. This limitation can pose problems for DWDM systems, depending upon how the measurement is implemented

Among the techniques and types of test equipment available to measure dispersion, few are able to measure end-to-end dispersion. Most, if not all, portable equipment is suitable only for measuring fiber sections of the system.

One reason is that individual links do not support bidirectional transmission. In a unidirectional system (see Fig. 1), a single fiber supports a single direction of transmission due to the use of isolators in erbium-fiber optical amplifiers (EDFAs). This is the principal difference between measuring fiber and measuring a complete system. The test signals must travel in the same direction as the traffic signals. This requirement means that optical time-domain reflectometry (OTDR) techniques for measuring dispersion cannot be used. Bidirectional systems that enable both directions of transmission on a single fiber are not bidirectional within their operating wavelength, as each direction uses its own band of wavelengths. However, in most situations there is a separate identical link running in parallel and operating in the other direction.

Another major factor that differentiates between fiber-only and complete-system measurement is added noise from optical amplifiers. Amplified spontaneous emission (ASE) from optical amplifiers results in noise at the terminal receiver, which can impair or obliterate the dispersion test signals.

Lastly, narrow-channel bandwidths limit system measurement. There are likely to be narrow-band components within the transmission bands or channels of the system. The test signals must be confined within these bands if the correct dispersion is to be measured. Test equipment using OTDR techniques or LEDs would not be suitable for such a measurement.

Most commercially available dispersion-measurement equipment is suited for measuring relatively short lengths of fiber up to 150 km—comparable to an amplifier span. This restriction causes a problem for system design because it forces a piecemeal approach to dispersion setup. Measurement of distributed and end-to-end dispersion of links with several types of fiber and dispersion-compensation devices relies on data from the constituent parts. Inevitable errors in the actual dispersion map occur, resulting in time-consuming readjustment during system setup. Any residual error eats into the system operating margins. Test equipment suitable for system measurements must be designed to address these factors. It must also be suitable for the development, testing, and commissioning of such systems.

To measure dispersion in a range of situations and types of systems, other considerations include the measurement/system interface, wavelength agility, and dynamic range. End-to-end measurement on individual channels is ideal for measuring the total dispersion and required compensation, some of which may be on a per-channel basis. Wavelength-scanned measurement of the "flat" system, omitting the multiplexers and demultiplexers, is also needed during the initial development or to check subsections of the system during installation. Terminal-to-amplifier node measurements are needed to check segments of the complete system.

Dispersion measurement is needed at any wavelength point in the passband when measuring either the "flat" system or a particular channel. With the L band now in use, an instrument capable of performing both these measurements in the C and L bands is required. In amplified systems, dynamic range is not a problem because the output power level of the test signals are comparable to that of the input signal, with a loss of less than one amplifier-span of fiber. However, the raw fiber of such a span may need to be examined, and require up to 50 dB of dynamic range.

The test equipment must be capable of high accuracy in the presence of the noise produced at the end of the link by the ASE from all the optical amplifiers. The ASE power has the potential to saturate the optical receiver in the measurement equipment. When measuring in a single channel in the presence of traffic signals, the interference from adjacent channels should not cause serious measurement error. Nor should the test signal interfere with traffic signal in adjacent channels.

The accuracy required depends upon the measurement situation. When dispersion compensation is set "blind" after a single measurement, the accuracy required may be better than 1%. In other situations where more iterative dispersion-compensation techniques are employed, lower accuracy is adequate. There is also a trade-off between repeatability and measurement speed; some means of adjusting this trade-off is desirable. While absolute accuracy depends on calibration, repeatability gives the user confidence in the measurements and allows precise relative measurements to be made.

The key factor in dispersion measurement is the use of a narrow-linewidth laser as the optical source. This laser provides a high signal-to-noise ratio, the ability to measure in confined wavelength regions, improved wavelength resolution, and wavelength accuracy. Because dispersion is measured at multiple wavelengths, a tunable laser source (TLS) must be used. Improvements in TLSs, such as ruggedness and lower cost, have enabled new portable test equipment to be developed. The LEDs used successfully in existing equipment generations have been replaced by a TLS, which maintains the well-established phase-shift method and advanced signal-processing methods used to measure phase differences (see Fig. 2).

Measurement equipment consists of two parts, a transmitter and a receiver, deployed at each end of the transmission link. A communications link is required between the transmitter and receiver, implemented over a separate dark fiber or channel of a WDM system. Such links are commonly available. The transmitter sends a sequence of modulated wavelengths over the DWDM system and the receiver measures their relative phase difference. The TLS in the remote transmitter provides both wavelength agility and compatibility with line-amplified systems, and is controlled remotely from the receiver. Specialist software runs on a PC connected to the receiver, and provides overall control, as well as a method of storing and displaying results.

The operator has control over measurement wavelengths by means of a preset program from the receiver. The equipment offers many possible chromatic-dispersion measurement modes including the measurement within a single free-WDM channel while the system is carrying live traffic. This option is useful if dispersion monitoring is required or when channels are being added or adjusted.

From dispersion measurements at various wavelengths, the delay/wavelength relationship, zero-dispersion wavelength, and dispersion slope can be derived (see table). An approximate fiber-length measurement is also provided from the delay measurement. An additional spectral power-level measurement is incorporated, which is used to determine relative power levels over the entire C or L bands.

This equipment allows transmission system suppliers and operators to measure the total end-to-end dispersion of both repeaterless and line-amplified systems efficiently and with the minimum of disruption. Dispersion at any wavelength in the C or L bands can be measured accurately, enabling the performance of 10- and 40-Gbit/s WDM systems to be predicted and optimized. Existing installed systems can be assessed for upgrades, and dispersion-compensation management schemes can measure and monitor over the entire wavelength range.

Arthur Barlow is technology and development manager at PerkinElmer Optoelectronics, Sorbus House, Mulberry Business Park, Wokingham, Berks, RG41 2GY, UK. He can be reached at

More in DWDM & ROADM