Abstract
M.Ing.
In this dissertation the emphasis falls on the use of the 1550 run WDM systems that make use
of standard single-mode fibre. The reason for this is that more than 50 million kilometres of
standard telecommunication fibre is already installed in the worldwide telephone network.
Therefore, from an economical point of view, the fourth generation of lightwave systems has
to make use of this existing base.
Using single-mode fibre at high bit-rate (- 10 Gb/s) dispersion becomes a problem.
Furthermore, if amplifiers are used, as in the case of long-haul WDM systems, non-linear
effects also become important although the input power levels are low (-1 mW). Thus, the
system performance is influenced by both the dispersion problem and fibre non-linearities. In
this dissertation the influence of these effects was investigated and the use of different
dispersion management schemes was considered in order to minimise the dispersion and the
effects of these non-linearities on the optical signals.
It was seen that by using pre-compensation, the optical pulses would become narrower
whereas by using post-compensation, the pulses would be broader than the original input
pulses. Thus, the combined effect of fibre non-linearity and fibre dispersion led to the fact
that the order in which dispersion compensation is realised is important. By using
symmetrical dispersion compensation that is defined as a dispersion compensation scheme
that makes use of an equal amount of pre- and post-compensation sections, the tendency of
pulse narrowing and pulse broadening can be balanced.
It was found from simulations, that pre-compensation led to a smaller SPM penalty than postcompensation,
whereas in the case of cross-phase ,modulation, the opposite is true. Using
lumped dispersion compensation at the,en(f:o£ the system, it was found that both SPM and
XPM penalties could be minimised. It was found that negative and positive lumped
dispersion compensation could be used to improve the system perfortriance of pre- and post compensation
schemes respectively. The performance of symmetrical compensation also
improved by using a positive lumped dispersion compensation regarding SPM penalty, but in
the case of XPM penalty, it was found that lumped dispersion compensation could not be
used to improve this penalty.
Furthermore, the effect of not using 100% dispersion compensation per span on the system
performance was determined. From simulations, it followed that under-compensation will
lead to an improvement in the system performances of the post- and symmetrical dispersion
compensation, whereas an over-compensation will lead to an improvement in the precompensation
case. It was also shown that XPM index penalty decreases in all of these
dispersion compensation schemes if 100% compensation per span is not used. The reason for
this is the inter -span walk-off that reduces the XPM distortion by reducing the localised buildup
of phase distortions.
Finally, it was shown that by independently controlling the input power levels in the SMF and
DCF the best system performance is possible. Furthermore, by making use of lumped
dispersion compensation or by not using 100% dispersion compensation per span, the system
performance of the pre-, post- and symmetrical dispersion conipensation schemes can be
improved quite a lot. Comparing the results of the pre-, post- and symmetrical dispersion
compensation schemes that made use of independent control of power levels with the pre-,
post- and symmetrical dispersion compensation schemes that did not use independent control
ofpower levels, the Q factors were improved from 6.2 to 8, from 5.4 to 9 and from 7.5 to 11
in each case respectively.