419511 : MAJOR : TELECOMMUNICATIONS AND COMPUTER NETWORKS ;
M.S. (TELECOMMUNICATION AND COMPUTER NETWORKS)
KEY WORD : WAVELENGTH DIVISION MULTIPLEXING GENETIC ALGORITHMS
PEMIGA SANGCHANSOMPORN: A STUDY OF GENETIC ALGORITHMS
FOR WAVELENGTH ASSIGNMENT IN WAVELENGTH DIVISION MULTIPLEXING
NETWORK. THESIS ADVISOR: PISIT CHARNKEITKONG, Ph.D. 64 p.
The aim of
this thesis is to examine the possibility of using a Genetic
Algorithms to solve the problem of wavelength assignment in wavelength division
multiplexing (WDM) networks focusing on network, optical nodes, non
conversion switches and groups of w
avelengths available in each different link. The
research presented herein, indicates that a routing algorithm, based on minimum hops
routing, which when modified for the networks was not transparent for all wavelengths. In
general wavelength assignment pr
oblem could be classified as a Non Polynomial (NP)
complete problem, which needs specific heuristic methods to solve. This thesis
addresses a wavelength assignment problem performed as a graph
where each node of a graph represents a connec
tion and nodes in the graph are
connected to each other by an edge if and only if they share some common links.
Moreover, a greedy algorithm and a hybrid mutation technique both, which mimic the
idea of Tabu search, are also presented herein. The main obje
ctive of this thesis is to
minimize the number of wavelengths used.
The wavelength assignment algorithms mentioned above were applied to
three sample networks having four, six and ten nodes. The results indicate that these
proposed algorithms can provide
an acceptable number of wavelengths.