PMD achieves global test status
WILLIAM B. GARDNER
Approximately six years ago, a new restriction on optical-fiber transmission--polarization-mode dispersion (PMD)--was uncovered. The well-known term "singlemode" disguises the fact that two states of polarization propagate through optical fiber. In addition, the two polarizations can travel at slightly different velocities. In digital systems, this difference in speed can cause the transmitted light pulses to spread out, or disperse, in width while traveling along considerable lengths--kilometers--of optical fiber. In analog systems, it can produce signal distortion.
Chromatic dispersion, the general form of dispersion found in singlemode fiber, has a distinct value for a given fiber at each wavelength. This value can be measured accurately (that is, to a few percent). By contrast, PMD is affected by the random shifting of energy from one polarization to the other. Because this shifting fluctuates randomly with time (due to fiber flexing, temperature changes, etc.), successive measurements of PMD on the same fiber can differ by 20%--or more, if the PMD value is low. This difference in measurement occurs regardless of which test method or test set is used.
The statistical nature of PMD, therefore, has resulted in a thorny issue for standards specifiers. During the past three years, various draft standards have been discussed, written and edited, but none has been approved.
Fortunately, this situation is changing. Two PMD fiber-optic test procedures--Fotp-113 and Fotp-122--are expected to be approved this month by the Telecommunications Industry Association (TIA). Fotp-113 is jointly proposed by Lucent Technologies and EG&G Fibre Optics in the United Kingdom, whereas Fotp-122 is authored by Hewlett-Packard Co.
At its meeting in Geneva this past May, Working Party 4/15 of the International Telecommunication Union (ITU) approved 35 pages of text on PMD testing for its five singlemode fiber recommendations. Four test methods, including the two from the TIA, were approved. One of the other test methods originated in Italy, whereas an interferometric method evolved from a collaboration among Swiss Telecom PTT, Alcatel, France Telecom and Kokusai Denshin Denwa of Japan.
Included in the ITU-approved pages are definitions of the terms PMD, principal states of polarization, differential group delay, PMD delay and PMD coefficient. Industry experts say PMD delay is the key performance parameter and the most controversial of the definitions. In practice, certain definitions prove better suited to certain test methods, and all PMD delay definitions are only approximately equivalent.
The ITU presently recognizes three legitimate definitions of PMD delay, and fortunately, the three differ by less than the aforementioned 20% statistical uncertainty of a measurement. Eventually, two of the three definitions will probably not be sanctioned.
Definitions and test methods might be viewed as groundwork leading to an eventual specification of PMD. Some vendors, such as Lucent Technologies, provide cables that deliver a maximum PMD coefficient of 0.5 picosecond per square root kilometer. The ITU stopped short of a specification, but did introduce a paragraph acknowledging the 0.5 PMD value as state of the art. The ITU further affirmed that this level of PMD would allow 10-Gbit/sec STM-64 digital systems to operate unregenerated over a distance of 400 km with less than a 1-dB penalty due to PMD.
The International Electrotechnical Commission is also moving in a direction similar to that of the ITU on PMD testing. After three years of debate, the measurement of PMD has finally been recognized in a worldwide approved test standard. q
For more on PMD, see page 43.--Ed.
William B. Gardner represents Lucent Technologies, Norcross, GA, on several fiber standards committees. He received a B.S. from the University of Alabama and a Ph.D. from Johns Hopkins University, both in physics.