Wavelengths switching and allocation algorithms in multicast technology using m-arity tree networks topology

Abstract

In this thesis, the m-arity tree networks have been investigated to derive equations for their nodes, links and required wavelengths. The relationship among all parameters such as leaves nodes, destinations, paths and wavelengths has been found. Three situations have been explored, firstly when just one server and the leaves nodes are destinations, secondly when just one server and all other nodes are destinations, thirdly when all nodes are sources and destinations in the same time. The investigation has included binary, ternary, quaternary and finalized by general equations for all m-arity tree networks. Moreover, a multicast technology is analysed in this thesis to transmit data carried by specific wavelengths to several clients. Wavelengths multicast switching is well examined to propose split-convert-split-convert (S-C-S-C) multicast switch which consists of light splitters and wavelengths converters. It has reduced group delay by 13% and 29% compared with split-convert (S-C) and split-convert-split (S-C-S) multicast switches respectively. The proposed switch has also increased the received signal power by a significant value which reaches 28% and 26.92% compared with S-C-S and S-C respectively. In addition, wavelengths allocation algorithms in multicast technology are proposed in this thesis using tree networks topology. Distributed scheme is adopted by placing wavelength assignment controller in all parents’ nodes. Two distributed algorithms proposed shortest wavelength assignment (SWA) and highest number of destinations with shortest wavelength assignment (HND-SWA) algorithms to increase the received signal power, decrease group delay and reduce dispersion. The performance of the SWA algorithm was almost better or same as HND-SWA related to the power, dispersion and group delay but they are always better than other two algorithms. The required numbers of wavelengths and their utilised converters have been examined and calculated for the researched algorithms. The HND-SWA has recorded the superior performance compared with other algorithms. It has reduced number of utilised wavelengths up to about 19% and minimized number of the used wavelengths converters up to about 29%. Finally, the centralised scheme is discussed and researched and proposed a centralised highest number of destinations (CHND) algorithm with static and dynamic scenarios to reduce network capacity decreasing (Cd) after each wavelengths allocation. The CDHND has reduced (Cd) by about 16.7% compared with the other algorithms.