抄録
Parametric wavelength converters are capable of simultaneously converting multiple wavelength pairs in contrast to traditional tunable converters. We therefore attempt to reduce the total number of converters required in a network by employing parametric converters network-wide under a first-fit wavelength assignment. Unfortunately, contrary to general intuition, our analyses indicate that parametric converters achieve no significant reduction with a sequential wavelength indexing based first-fit assignment, since it leads to a misalignment of wavelength conversion pairs at individual nodes. This letter proposes a median flip-flop indexing scheme that remedies this issue and significantly improves converter utilization. Numerical results indicate up to 57% reduction in total converters required.
