Addressing PMD at 40 Gbit/s
Francis Audet, Senior Product Manager, Optical Business Unit
As the popularity for 40 Gbit/s transmission technology continues to grow,
it is important to consider the issues that emerge with the increase in speed.
Previously, in this series, we examined the effects of chromatic dispersion
(CD) at 40 Gbit/s and presented modulation formats and techniques to compensate
for CD, including per-wavelength compensations. Another form of dispersion
that affects 40 Gbit/s transmission is polarization mode dispersion (PMD).
Many long-haul network infrastructures contain older fibers that are more
subject to PMD; i.e., pulse spreading caused by the impurities and local
stresses on a fiber. While the modulation schemes discussed in the previous
articles increase PMD tolerance to some extent, they do not eliminate the
effects of this type of dispersion, so PMD remains a very real and growing
concern.
For 10 Gbit/s transmission, the average DGD (i.e., PMD) value must remain
lower than 10 ps in order to keep the power penalty below 1 dB, 99.9954%
of the time. As modulation formats increase the CD tolerance, they do the
same for PMD tolerance, but in different proportions. So even though the
efficient data rate at 40 Gbit/s is four times faster, the PMD tolerance
may not necessarily be four times less. Nonetheless, this threshold must
be verified with the system vendor to make sure the link can support the
given level of PMD.
As networks become more agile, with wavelength-selectable switches (WSSs)
and reconfigurable optical add/drop multiplexers (ROADMs), the situation
becomes more complex because the total optical link is longer, which in turn
proportionally increases PMD and its effects. This also holds true for CD
but, as previously mentioned, CD can easily be compensated for. The biggest
challenge in PMD lies mainly in the fact that there aren’t any commercially
available methods to compensate for PMD, which increases the risks associated
with PMD and, consequently, the related testing requirements.
As mentioned previously, CD compensation must be extremely precise because
of the tight tolerance. Another reason compensation needs to be perfect
is due to what is commonly referred to as second-order PMD.
PMD is defined as the average value (or the root-mean-square (RMS)) of the
DGD, which varies as a function of wavelength (see below):
Figure 1 – Graphical representation of PMD; i.e., average DGD as a function of wavelength
Although very complex, second-order PMD can be rudimentarily described as
the DGD difference within a single dense wavelength-division multiplexing
(DWDM) channel, as illustrated in the figure below:.
Figure 2. Difference in DGD (second-order PMD) in a single DWDM channel. DWDM channel represented by the orange dotted line, whereas second-order PMD represented by the red arrow.
Seeing that second-order PMD is expressed in the same units as CD, they do
in fact add up. Second-order PMD adds a random value added onto the CD (yielding
an average value over a certain period of time, but with local pikes in wavelength
or in time. This is yet another reason for applying perfect CD compensation,
as it,must also keep some margin for this unpredictable contributor.
In the next edition, we will discuss the impact of DWDM limitations and
testing for 40 Gbit/s networks. |
