Master of Science in Electrical Engineering (MSEE)
Electrical and Computer Engineering
All-Optical Networks employing Dense Wavelength Division Multiplexing (DWDM) are believed to be the next generation networks that can meet the ever-increasing demand for bandwidth of the end users. This thesis presents some new heuristics for wavelength assignment and converter placement in mesh topologies. Our heuristics try to assign the wavelengths in an efficient manner that results in very low blocking probability. We propose novel static and dynamic assignment schemes that outperform the assignments reported in the literature even when converters are used. The proposed on-line scheme called "Round-Robin" assignment outperforms previously proposed strategies such as first-fit and random assignment schemes. The performance improvement obtained with the proposed static assignments is very significant when compared with the dynamic schemes. We designed and developed a simulator in the C language that supports the 2D mesh topology with DWDM. We ran extensive simulations and compared our heuristics with those reported in the literature. We have examined converter placement in mesh topologies and proposed that placing converters at the center yields better results than uniform placement when dimension order routing is employed. We introduced a new concept called "wavelength assignment with second trial" that results in extremely low blocking probabilities when compared to schemes based on a single trial. Our proposed schemes are simple to implement and do not add to the cost. Thus we conclude that wavelength assignment plays more significant role in affecting the blocking probability than wavelength converters. We further conclude that static schemes without converters could easily outperform dynamic schemes thus resulting in great savings.
Document Availability at the Time of Submission
Release the entire work immediately for access worldwide.
Bijja, Prabhulaiah, "Wavelength assignment in all-optical networks for mesh topologies" (2002). LSU Master's Theses. 3236.
Ahmed El. Amawy