Since polarization-mode dispersion (PMD) produces pulse-spreading, every increase in bit rate requires a tighter restriction on PMD. Measurement and deployment conditions may become the most difficult obstacles to tighter PMD requirements. PMD is a random variable whose value depends on the fiber, cable, and deployment.
Prof. Nicolas Gisin and co-workers at the University of Geneva showed there is an inherent uncertainty in a PMD measurement, which applies to any test set whose spectral range is finite (meaning all test sets). They showed for a typical test set measuring 1 km long cable with a PMD coefficient of 0.1 psec/km, the inherent uncertainty is ±28%!
An uncertainty this large can make the value of a single PMD measurement marginal. Fortunately, the percentage of uncertainty is less for the mean of a population (e.g., all of the fibers in a concatenation).
The International Telecommunication Union ITU-T and International Electrotechnical Commission take advantage of that by specifying a cable-link design value PMDQ rather than a PMD measured on a single fiber. In their usage, there is a 99.99% confidence level that the PMD coefficient of 20 concatenated cabled fibers will be less than or equal to the PMDQ value that characterizes the population from which the 20 were randomly chosen.
Before the advent of 40 Gbits/sec, ITU-T placed a maximum of 0.5 psec/km on the PMDQ of cabled fibers. Last year, they added three new categories within Recommendations G.652 and G.655, each having PMDQ ≤ 0.20 psec/km. This PMDQ should allow 40-Gbit/sec transmission up to a few hundred kilometers. Greater distances would require an even lower PMDQ and/or the introduction of PMD compensation.
Even if the manufacturing challenges continue to be met, the inherent measurement uncertainty becomes an increasingly serious problem as PMD requirements tighten.Dr. William B. Gardner represents OFS in ITU-T Working Party 4/15, where he serves as a Rapporteur. He can be reached at 770-798-2674 or [email protected].