While standards bodies continue to grapple with the thorny problem of how to specify polarization mode dispersion (PMD), a smaller, more-distant cloud has appeared on the horizon: second-order PMD or PMD2.
Conventional (first-order) PMD, or PMD1, is defined as either the mean or root mean square (RMS) value of the differential group delay (DGD) between the two principal states of polarization (PSP) of a singlemode fiber. The DGD and PSP in discrete components (such as single waveplates) are nearly independent of wavelength. But DGD and PSP both have a measurable wavelength dependence in long fibers. The wavelength dependence in these two parameters (with that of PSP being the larger contributor) gives rise PMD2. Although the second-order effect is subtler than the first-order effect, there is no assurance that its impact will always be negligible.
The Telecommunications Industry Association (TIA) Working Group FO-6.6.5 on Singlemode Fiber, chaired by Felix Kapron of Telcordia Technologies, recently heard presentations on this issue from three of its members: Felix Kapron and Paul Hernday of Hewlett-Packard and Doug Franzen of the National Institute of Standards and Technology. Here are some of the conclusions from their presentations.
Under the assumption that outside-plant cable lengths are long compared to the PMD coupling length, PMD2 is proportional to the square of PMD1. One consequence of this assumption is that PMD2 is proportional to length, by contrast with PMD1, which in this regime is proportional to the square root of length. Because of this, PMD2 adds or subtracts to the chromatic dispersion, which is also proportional to length. But PMD2's resemblance to chromatic dispersion ends there: The latter is a deterministic variable, while PMD2 is a statistical variable (like PMD1). The statistical nature of PMD2 creates a fluctuating component in the chromatic dispersion, which could create problems for dispersion compensation.
Intersymbol interference is the primary degradation that PMD2 introduces in digital systems. Therefore, the chances of PMD2 having an impact on the system increase as the bit rate increases. PMD2 could also impact broadband analog systems, the chances of which are enhanced if there is chirp in the transmitter.
Since PMD2 is defined through the fiber PSPs, the best test methods for measuring PMD2 would be those yielding information about the PSPs, namely the Jones Matrix Eigenanalysis (JME) Method and Poincare Sphere Method. The United States is proposing to consolidate these two, methods into a single Reference Test Method for PMD1.
Since these presentations were given, researchers at Lucent Technologies published a paper in the IEEE Photonics Technology Letters in September describing the Mueller Matrix Method for measuring PMD2. This method in many ways parallels the JME method but may be simpler to implement.
TIA FO-6.6.5 is pondering this information, but will probably not act on PMD2 until the standards work on PMD1 is completed and/or PMD2 can be shown to be a limitation for systems of practical interest.
William B. Gardner represents Lucent Technologies, Norcross, GA, on several fiber standards committees. He can be contacted at tel: (770) 798-2674; fax: (770) 798-4654; e-mail: email@example.com.