Methods for in-service measurement of polarization mode dispersion

Abstract

Polarization mode dispersion (PMD) is one of the major limitations for optical transmission systems at 10 Gb/s and beyond. While first- or second-order PMD compensators (PMDC) can be driven with a feedback signal, more complex broadband PMDCs have to be set feed forward. An exact knowledge of the fiber's PMD characteristics - e.g. the PMD vector - is needed for the feed forward setting. Since PMD changes with time, real-time PMD measurement without data traffic interruption is necessary. Some recently published frequency- and time-domain methods meet these conditions. In this publication we are going to examine and compare different on-line measurement methods. Using numerical simulations, the performance of the measurement methods is assessed in terms of the accuracy of the PMD vector measurement and the qualification as feed forward control signal for setting a PMDC. The measurements exhibit an inherent inaccuracy if the signal is launched close to one of the principal states of polarization (PSP). Although these combinations of PSP and signal polarization result in inaccurate PMD vector measurements, the transmitted signal is not degraded by first order PMD. Consequently, the accuracy of the PMD vector measurement is a bad figure of merit for the performance of a system including a feed-forward set PMDC. Furthermore, due to the averaging over the signal bandwidth, the measured PMD vector is a better control variable for a PMDC than the analytically calculated PMD vector if second order PMD is considered.