The Impact of Polarization Mode Dispersion on Lightwave Systems
01 January 2001
Polarization-mode dispersion is emerging as one of the primary concerns when transmitting data at speeds of 40 Gb/s or higher in lightwave systems. Part of this concern in rooted in mystery: unlike most impairments, that due to PMD is inherently stochastic with a time scale for change that is much longer than a bit period. Thus, one cannot predict the impairment of a channel absolutely and must instead resort to descriptions that incorporate knowledge of the probability densities for the phenomenon. This also means that the compensation of PMD must be adaptive. Another problem, different from those experienced when handling most other impairments, arises when mitigating PMD penalties. For many sources of impairment, one can conceive of a compensator that handles the entire range of the phenomenon. This is because most sources of impairment occur within a limited range. PMD, in contrast, can provide infinite differential group delays (DGDs), at least in theory. In reality, one expects the probability density of the DGD to be truncated at some value short of infinity. Although the DGD values at which this truncation occurs is not well known, but one would expect them to be much larger than the mean DGD, perhaps 10 to 100 times the mean DGD. Most compensators are designed to cover only three to five times the mean DGD, so once again, even when compensating for PMD, one must use probability densities to understand the behavior of the system.