Avec les nouvelles technologies des réseaux optiques, une quantité de données de plus en plus grande peut être transportée par une seule longueur d'onde. Cette quantité peut atteindre jusqu’à 40 gigabits par seconde (Gbps). Les flots de données individuels quant à eux demandent beaucoup moins de bande passante. Le groupage de trafic est une technique qui permet l'utilisation efficace de la bande passante offerte par une longueur d'onde. Elle consiste à assembler plusieurs flots de données de bas débit en une seule entité de données qui peut être transporté sur une longueur d'onde. La technique demultiplexage en longueurs d'onde (Wavelength Division Multiplexing WDM) permet de transporter plusieurs longueurs d'onde sur une même fibre. L'utilisation des deux techniques : WDM et groupage de trafic, permet de transporter une quantité de données de l'ordre de terabits par seconde (Tbps) sur une même fibre optique. La protection du trafic dans les réseaux optiques devient alors une opération très vitale pour ces réseaux, puisqu'une seule panne peut perturber des milliers d'utilisateurs et engendre des pertes importantes jusqu'à plusieurs millions de dollars à l'opérateur et aux utilisateurs du réseau. La technique de protection consiste à réserver une capacité supplémentaire pour acheminer le trafic en cas de panne dans le réseau. Cette thèse porte sur l'étude des techniques de groupage et de protection du trafic en utilisant les p-cycles dans les réseaux optiques dans un contexte de trafic dynamique. La majorité des travaux existants considère un trafic statique où l'état du réseau ainsi que le trafic sont donnés au début et ne changent pas. En plus, la majorité de ces travaux utilise des heuristiques ou des méthodes ayant de la difficulté à résoudre des instances de grande taille. Dans le contexte de trafic dynamique, deux difficultés majeures s'ajoutent aux problèmes étudiés, à cause du changement continuel du trafic dans le réseau. La première est due au fait que la solution proposée à la période précédente, même si elle est optimisée, n'est plus nécessairement optimisée ou optimale pour la période courante, une nouvelle optimisation de la solution au problème est alors nécessaire. La deuxième difficulté est due au fait que la résolution du problème pour une période donnée est différente de sa résolution pour la période initiale à cause des connexions en cours dans le réseau qui ne doivent pas être trop dérangées à chaque période de temps. L'étude faite sur la technique de groupage de trafic dans un contexte de trafic dynamique consiste à proposer différents scénarios pour composer avec ce type de trafic, avec comme objectif la maximisation de la bande passante des connexions acceptées à chaque période de temps. Des formulations mathématiques des différents scénarios considérés pour le problème de groupage sont proposées. Les travaux que nous avons réalisés sur le problème de la protection considèrent deux types de p-cycles, ceux protégeant les liens (p-cycles de base) et les FIPP p-cycles (p-cycles protégeant les chemins). Ces travaux ont consisté d’abord en la proposition de différents scénarios pour gérer les p-cycles de protection dans un contexte de trafic dynamique. Ensuite, une étude sur la stabilité des p-cycles dans un contexte de trafic dynamique a été faite. Des formulations de différents scénarios ont été proposées et les méthodes de résolution utilisées permettent d’aborder des problèmes de plus grande taille que ceux présentés dans la littérature. Nous nous appuyons sur la méthode de génération de colonnes pour énumérer implicitement les cycles les plus prometteurs. Dans l'étude des p-cycles protégeant les chemins ou FIPP p-cycles, nous avons proposé des formulations pour le problème maître et le problème auxiliaire. Nous avons utilisé une méthode de décomposition hiérarchique du problème qui nous permet d'obtenir de meilleurs résultats dans un temps raisonnable. Comme pour les p-cycles de base, nous avons étudié la stabilité des FIPP p-cycles dans un contexte de trafic dynamique. Les travaux montrent que dépendamment du critère d'optimisation, les p-cycles de base (protégeant les liens) et les FIPP p-cycles (protégeant les chemins) peuvent être très stables. / With new technologies in optical networking, an increasing quantity of data can be carried by a single wavelength. This amount of data can reach up to 40 gigabits per second (Gbps). Meanwhile, the individual data flows require much less bandwidth. The traffic grooming is a technique that allows the efficient use of the bandwidth offered by a wavelength. It consists of assembling several low-speed data streams into a single data entity that can be carried on a wavelength. The wavelength division multiplexing (WDM) technique allows carrying multiple wavelengths on a single fiber. The use of the two techniques,WDMand traffic grooming, allows carrying a quantity of data in the order of terabits per second (Tbps) over a single optical fiber. Thus, the traffic protection in optical networks becomes an operation very vital for these networks, since a single failure can disrupt thousands of users and may result in several millions of dollars of lost revenue to the operator and the network users. The survivability techniques involve reserving additional capacity to carry traffic in case of a failure in the network. This thesis concerns the study of the techniques of grooming and protection of traffic using p-cycles in optical networks in a context of dynamic traffic. Most existing work considers a static traffic where the network status and the traffic are given at the beginning and do not change. In addition, most of these works concerns heuristic algorithms or methods suffering from critical lack of scalability. In the context of dynamic traffic, two major difficulties are added to the studied problems, because of the continuous change in network traffic. The first is due to the fact that the solution proposed in the previous period, even if optimal, does not necessarily remain optimal in the current period. Thus, a re-optimization of the solution to the problem is required. The second difficulty is due to the fact that the solution of the problem for a given period is different from its solution for the initial period because of the ongoing connections in the network that should not be too disturbed at each time period. The study done on the traffic grooming technique in the context of dynamic traffic consists of proposing different scenarios for dealing with this type of traffic, with the objective of maximizing the bandwidth of the new granted connections at each time period. Mathematical formulations of the different considered scenarios for the grooming problem are proposed. The work we have done on the problem of protection considers two types of p-cycles, those protecting links and FIPP p-cycles (p-cycle protecting paths). This work consisted primarily on the proposition of different scenarios for managing protection p-cycles in a context of dynamic traffic. Then, a study on the stability of cycles in the context of dynamic traffic was done. Formulations of different scenarios have been proposed and the proposed solution methods allow the approach of larger problem instances than those reported in the literature. We rely on the method of column generation to implicitly enumerate promising cycles. In the study of path protecting p-cycles or FIPP p-cycles, we proposed mathematical formulations for the master and the pricing problems. We used a hierarchical decomposition of the problem which allows us to obtain better results in a reasonable time. As for the basic p-cycles, we studied the stability of FIPP p-cycles in the context of dynamic traffic. The work shows that depending on the optimization criterion, the basic p-cycles (protecting the links) and FIPP p-cycles (protecting paths) can be very stable.

A interligação de várias redes de telecomunicação ampliou a cobertura, mas tornou a operabilidade entre elas complexa, principalmente por causa da arquitetura resultante, formada por várias camadas. Estas camadas lidam com protocolos e taxas de transmissão diferentes e com sinais elétricos e ópticos. Diante desse cenário, a alternativa usual de expandir os recursos proporcionalmente ao crescimento da demanda é inviável devido ao elevado custo. Assim, soluções eficientes que agregam os benefícios das tecnologias ópticas e eletrônicas na arquitetura de nós e no gerenciamento de tráfego tornaram-se uma necessidade importante no projeto, expansão e gerenciamento de redes de telecomunicação. Os nós que permitem a comutação de tráfego na camada óptica e eletrônica e a agregação de tráfego em várias granularidades têm sido empregados para fazer o melhor uso possível dos recursos disponíveis nas redes. Esses nós são conhecidos na literatura como MG-OXCs. Nessa pesquisa, foi proposta uma abordagem dos nós MG-OXCs de uma e três camadas com a inclusão de recursos como conversão de comprimento de onda e agregação de tráfego. Também foi proposto um método de cálculo do custo desses nós baseado no volume de utilização das portas. Posteriormente foram inseridos fatores de degradação do sinal óptico para a análise da camada física de redes com os nós MGOXCs. Adicionalmente, para o gerenciamento de tráfego, foi proposto um esquema de escolha de comprimentos de onda chamado de canal específico e um esquema de monitoramento baseado na intensidade de tráfego. A meta principal é a redução da probabilidade de bloqueio de solicitação de conexão. Em redes ópticas, os modelos de nós propostos conseguem diminuir o número de portas dos nós comutadores tradicionais além de realizar a comutação de tráfego de conexões de diferentes valores de largura de banda. O esquema do canal específico consegue melhorar a utilização da largura de banda dos comprimentos de onda e diminui a probabilidade de bloqueio e o esquema do monitoramento diminui o número de conexões bloqueadas devido ao aumento dos recursos. Resultados numéricos apresentados demonstram a potencialidade dos algoritmos propostos para gerenciar recursos e rotear o tráfego das redes de telecomunicação. / Integrating telecommunication networks has enlarged the coverage, but has made operations more complex, mainly because of the architecture, formed by various layers. These layers deal with different protocols and transmission rates, as well as electrical and optical signals. The usual alternative of expanding the resources proportionally to the demand is impractical due to the high cost. Therefore, efficient solutions which add to the benefits of optical and electronic technology in node architecture and traffic management are essential in the design, expansion and management of telecommunications networks. The nodes that enable traffic switching in the optical and electronic layer and traffic grooming have been used to make the best use of the available resources in the networks. These nodes are known as MG-OXCs. In this research, an approach based on MG-OXCs was proposed with one and three layers, which include the wavelength conversion and traffic grooming. Additionally, a method to calculate the node costs based on their use of ports was proposed. The work also considered optical signal impairments in order to analyze the network physical layer with MG-OXCs nodes. For the purpose of traffic management, a scheme which sets specific wavelengths for different bandwidths and a scheme which monitors the flow of traffic were proposed. The main aim is to reduce the blocking probability of connection requests. In optical networks, the node models proposed are able to reduce the number of ports used in usual optical cross connects and switch the traffic connections using different bandwidths. The scheme of setting specific wavelengths for different bandwidths improves the bandwidth use and the blocking probability. The scheme which monitors the flow of traffic achieves blocking probability reduction due to the increase in resources. The numerical results presented show the feasibility of the proposed algorithms to manage resources and switch traffic in telecommunication networks.

A tecnologia WDM (Wavelength Division Multiplexing) e a introdução de OXCs (Optical Cross Connect) e OADMs (Optical Add/Drop Multiplexer) puramente ópticos podem dotar as redes ópticas da função de networking, ou seja, da capacidade de manipular comprimentos de onda de forma a implementar o roteamento destes. Esta possibilidade implica em uma nova forma de relacionamento das aplicações com a camada física, sendo a arquitetura GMPLS candidata a estabelecer tal relacionamento. Soluções eficientes para o problema de alocação de recursos e roteamento de tráfego tornam-se uma necessidade imperiosa em projeto, expansão e gerenciamento de redes ópticas. A contribuição desta tese consiste em relacionar funcionalidades tais como: agregação (grooming) de tráfego, mecanismo de controle de admissão de chamadas (CAC), mecanismos de restauração e alocação de conversores em redes ópticas heterogêneas, avaliando-se as métricas de probabilidade de bloqueio, probabilidade do tráfego bloqueado e imparcialidade (fairness). Tais funcionalidades são tratadas separadamente na literatura. Com este objetivo em mente modela-se a rede com duas camadas: a camada física e a camada virtual. Estabelecem-se duas políticas de agregação de tráfego MrTV (minimização da rota na topologia virtual) e MrTF (minimização da rota na topologia física) e analisa-se o desempenho destas em relação à porcentagem de tráfego bloqueado. Em seguida um mecanismo de controle de admissão de chamadas (CAC) é implementado e sua influência em termos de imparcialidade (fairness) e probabilidade de bloqueio é analisada. A simulação e análise de redes ópticas, como a Rede NSFnet e a Rede Nacional Italiana são executadas por meio da implementação de um grafo baseado em Zhu e Mukherjee [28]. Como resultado, a política MrTF apresenta menor porcentagem de tráfego bloqueado do que a política MrTV para as redes simuladas e a implementação de um mecanismo de janela deslizante (rolling window) tornou o mecanismo de CAC mais otimizado. A utilidade de se implantar conversores de comprimento de onda apenas em alguns nós da rede (conversão esparsa) é estudada e uma análise sobre a probabilidade de bloqueio e a imparcialidade da rede desta distribuição de conversores é apresentada. Finalmente, técnicas de restauração na camada física e virtual são implementadas e uma análise da influência destes sobre a probabilidade de bloqueio e a imparcialidade da rede é executada. / The Wavelength Division Multiplexing (WDM) technology as well as both the introduction of all optical OXCs (Optical Cross Connect) and OADMs (Optical Add/Drop Multiplexer) may provide the optical network with the networking function, i.e, the capacity to manipulate wavelengths in order to implement their routing. This possibility implies a new type of relationship between applications and the physical layer. The likely candidate to establish such relationships is GMPLS architecture. Efficient solutions to both the problems of allocating resources and traffic routing become an enhanced requirement in the design, expansion and management of optical networking. The present study focus at the relationship between functionalities such as traffic grooming, network fairness improvement, protection/restoring mechanisms and wavelength conversion in heterogeneous optical networks, evaluating the metrics of blocking probability, probability of traffic blocked and fairness. These functionalities are separately treated in literature. With this goal in mind, a two-layer representation is used in order to model the network: the physical and virtual layers, respectively. Two policies on traffic grooming are set up, as follows: MrTV (route minimizing on virtual topology) and MrTF (route minimizing on physical topology). The performance of such policies is analyzed regarding the percentage of blocked traffic. Next, a mechanism for call admission control (CAC) is implemented and its influence in terms of fairness and blocking probability is discussed. The simulations of optical networks such as NSFnet and the Italian National Network are carried out through a graph based in Zhu and Mukherjee [28]. As a result, MrTF policy presents a smaller percentage of blocked traffic than the MrTV for the simulated networks and the rolling window mechanism has allowed the optimization of the call admission control (CAC) mechanism. The usefulness of placing wavelength converters in a few networks nodes (spare conversion) is studied and an analysis on the blocking probability is presented. Next, the network fairness for this distribution of wavelength converters is presented. Finally, techniques for restoring both physical and virtual layers are also implemented and an analysis regarding their influence on the blocking probability and the network fairness is carried out.

Avec les nouvelles technologies des réseaux optiques, une quantité de données de plus en plus grande peut être transportée par une seule longueur d'onde. Cette quantité peut atteindre jusqu’à 40 gigabits par seconde (Gbps). Les flots de données individuels quant à eux demandent beaucoup moins de bande passante. Le groupage de trafic est une technique qui permet l'utilisation efficace de la bande passante offerte par une longueur d'onde. Elle consiste à assembler plusieurs flots de données de bas débit en une seule entité de données qui peut être transporté sur une longueur d'onde. La technique demultiplexage en longueurs d'onde (Wavelength Division Multiplexing WDM) permet de transporter plusieurs longueurs d'onde sur une même fibre. L'utilisation des deux techniques : WDM et groupage de trafic, permet de transporter une quantité de données de l'ordre de terabits par seconde (Tbps) sur une même fibre optique. La protection du trafic dans les réseaux optiques devient alors une opération très vitale pour ces réseaux, puisqu'une seule panne peut perturber des milliers d'utilisateurs et engendre des pertes importantes jusqu'à plusieurs millions de dollars à l'opérateur et aux utilisateurs du réseau. La technique de protection consiste à réserver une capacité supplémentaire pour acheminer le trafic en cas de panne dans le réseau. Cette thèse porte sur l'étude des techniques de groupage et de protection du trafic en utilisant les p-cycles dans les réseaux optiques dans un contexte de trafic dynamique. La majorité des travaux existants considère un trafic statique où l'état du réseau ainsi que le trafic sont donnés au début et ne changent pas. En plus, la majorité de ces travaux utilise des heuristiques ou des méthodes ayant de la difficulté à résoudre des instances de grande taille. Dans le contexte de trafic dynamique, deux difficultés majeures s'ajoutent aux problèmes étudiés, à cause du changement continuel du trafic dans le réseau. La première est due au fait que la solution proposée à la période précédente, même si elle est optimisée, n'est plus nécessairement optimisée ou optimale pour la période courante, une nouvelle optimisation de la solution au problème est alors nécessaire. La deuxième difficulté est due au fait que la résolution du problème pour une période donnée est différente de sa résolution pour la période initiale à cause des connexions en cours dans le réseau qui ne doivent pas être trop dérangées à chaque période de temps. L'étude faite sur la technique de groupage de trafic dans un contexte de trafic dynamique consiste à proposer différents scénarios pour composer avec ce type de trafic, avec comme objectif la maximisation de la bande passante des connexions acceptées à chaque période de temps. Des formulations mathématiques des différents scénarios considérés pour le problème de groupage sont proposées. Les travaux que nous avons réalisés sur le problème de la protection considèrent deux types de p-cycles, ceux protégeant les liens (p-cycles de base) et les FIPP p-cycles (p-cycles protégeant les chemins). Ces travaux ont consisté d’abord en la proposition de différents scénarios pour gérer les p-cycles de protection dans un contexte de trafic dynamique. Ensuite, une étude sur la stabilité des p-cycles dans un contexte de trafic dynamique a été faite. Des formulations de différents scénarios ont été proposées et les méthodes de résolution utilisées permettent d’aborder des problèmes de plus grande taille que ceux présentés dans la littérature. Nous nous appuyons sur la méthode de génération de colonnes pour énumérer implicitement les cycles les plus prometteurs. Dans l'étude des p-cycles protégeant les chemins ou FIPP p-cycles, nous avons proposé des formulations pour le problème maître et le problème auxiliaire. Nous avons utilisé une méthode de décomposition hiérarchique du problème qui nous permet d'obtenir de meilleurs résultats dans un temps raisonnable. Comme pour les p-cycles de base, nous avons étudié la stabilité des FIPP p-cycles dans un contexte de trafic dynamique. Les travaux montrent que dépendamment du critère d'optimisation, les p-cycles de base (protégeant les liens) et les FIPP p-cycles (protégeant les chemins) peuvent être très stables. / With new technologies in optical networking, an increasing quantity of data can be carried by a single wavelength. This amount of data can reach up to 40 gigabits per second (Gbps). Meanwhile, the individual data flows require much less bandwidth. The traffic grooming is a technique that allows the efficient use of the bandwidth offered by a wavelength. It consists of assembling several low-speed data streams into a single data entity that can be carried on a wavelength. The wavelength division multiplexing (WDM) technique allows carrying multiple wavelengths on a single fiber. The use of the two techniques,WDMand traffic grooming, allows carrying a quantity of data in the order of terabits per second (Tbps) over a single optical fiber. Thus, the traffic protection in optical networks becomes an operation very vital for these networks, since a single failure can disrupt thousands of users and may result in several millions of dollars of lost revenue to the operator and the network users. The survivability techniques involve reserving additional capacity to carry traffic in case of a failure in the network. This thesis concerns the study of the techniques of grooming and protection of traffic using p-cycles in optical networks in a context of dynamic traffic. Most existing work considers a static traffic where the network status and the traffic are given at the beginning and do not change. In addition, most of these works concerns heuristic algorithms or methods suffering from critical lack of scalability. In the context of dynamic traffic, two major difficulties are added to the studied problems, because of the continuous change in network traffic. The first is due to the fact that the solution proposed in the previous period, even if optimal, does not necessarily remain optimal in the current period. Thus, a re-optimization of the solution to the problem is required. The second difficulty is due to the fact that the solution of the problem for a given period is different from its solution for the initial period because of the ongoing connections in the network that should not be too disturbed at each time period. The study done on the traffic grooming technique in the context of dynamic traffic consists of proposing different scenarios for dealing with this type of traffic, with the objective of maximizing the bandwidth of the new granted connections at each time period. Mathematical formulations of the different considered scenarios for the grooming problem are proposed. The work we have done on the problem of protection considers two types of p-cycles, those protecting links and FIPP p-cycles (p-cycle protecting paths). This work consisted primarily on the proposition of different scenarios for managing protection p-cycles in a context of dynamic traffic. Then, a study on the stability of cycles in the context of dynamic traffic was done. Formulations of different scenarios have been proposed and the proposed solution methods allow the approach of larger problem instances than those reported in the literature. We rely on the method of column generation to implicitly enumerate promising cycles. In the study of path protecting p-cycles or FIPP p-cycles, we proposed mathematical formulations for the master and the pricing problems. We used a hierarchical decomposition of the problem which allows us to obtain better results in a reasonable time. As for the basic p-cycles, we studied the stability of FIPP p-cycles in the context of dynamic traffic. The work shows that depending on the optimization criterion, the basic p-cycles (protecting the links) and FIPP p-cycles (protecting paths) can be very stable.

Les interactions interindividuelles permettent la transmission de l’information, la dispersion des pathogènes et la mise en place des comportements dans une population. Cette thèse a permis d’évaluer l’influence des interactions sociales sur le succès de fondation colonial des différents reproducteurs de deux termites européens, l’invasif Reticulitermes flavipes et le natif R. grassei. Les résultats révèlent (i) un meilleur succès de fondation des reproducteurs primaires de R. flavipes, (ii) une organisation biparentale des soins aux jeunes toutes espèces confondues et (iii) une communication et des soins aux oeufs propres aux caractères invasif et natif des espèces d’étude. Pour finir, (iv) une meilleure survie et communication a été observée dans les colonies fondées avec reproducteurs secondaires tandis (v) qu’une communication supérieure et une survie moindre sont observées pour R. flavipes. Les origines évolutives de l’organisation biparentale et des variations de succès de fondations sont discutées. / Individual interactions permit information transmission, pathogen dispersion and shape behavioral strategies in a population. This thesis has permit to explore the influence of social interactions on the colonial foundation success of two European termites, the invasive Reticulitermes flavipes and the native R. grassei. The overall results revealed (i) a better foundation success of primary reproductives of R. flavipes, (ii) a biparental organisation of parental care in both species (iii) a level of communication and egg care reflecting native and invasive status of the two species studied. To finish, (iv) better survival and communication rates were observed in colonies founded with secondary reproductives than in colonies without any and (v) a better communication rate and a weaker survival rate for R. flavipes foundations with or without secondary reproductives. Evolutive origins of biparental care and of the variations of foundation success observed are discussed.

A tecnologia WDM (Wavelength Division Multiplexing) e a introdução de OXCs (Optical Cross Connect) e OADMs (Optical Add/Drop Multiplexer) puramente ópticos podem dotar as redes ópticas da função de networking, ou seja, da capacidade de manipular comprimentos de onda de forma a implementar o roteamento destes. Esta possibilidade implica em uma nova forma de relacionamento das aplicações com a camada física, sendo a arquitetura GMPLS candidata a estabelecer tal relacionamento. Soluções eficientes para o problema de alocação de recursos e roteamento de tráfego tornam-se uma necessidade imperiosa em projeto, expansão e gerenciamento de redes ópticas. A contribuição desta tese consiste em relacionar funcionalidades tais como: agregação (grooming) de tráfego, mecanismo de controle de admissão de chamadas (CAC), mecanismos de restauração e alocação de conversores em redes ópticas heterogêneas, avaliando-se as métricas de probabilidade de bloqueio, probabilidade do tráfego bloqueado e imparcialidade (fairness). Tais funcionalidades são tratadas separadamente na literatura. Com este objetivo em mente modela-se a rede com duas camadas: a camada física e a camada virtual. Estabelecem-se duas políticas de agregação de tráfego MrTV (minimização da rota na topologia virtual) e MrTF (minimização da rota na topologia física) e analisa-se o desempenho destas em relação à porcentagem de tráfego bloqueado. Em seguida um mecanismo de controle de admissão de chamadas (CAC) é implementado e sua influência em termos de imparcialidade (fairness) e probabilidade de bloqueio é analisada. A simulação e análise de redes ópticas, como a Rede NSFnet e a Rede Nacional Italiana são executadas por meio da implementação de um grafo baseado em Zhu e Mukherjee [28]. Como resultado, a política MrTF apresenta menor porcentagem de tráfego bloqueado do que a política MrTV para as redes simuladas e a implementação de um mecanismo de janela deslizante (rolling window) tornou o mecanismo de CAC mais otimizado. A utilidade de se implantar conversores de comprimento de onda apenas em alguns nós da rede (conversão esparsa) é estudada e uma análise sobre a probabilidade de bloqueio e a imparcialidade da rede desta distribuição de conversores é apresentada. Finalmente, técnicas de restauração na camada física e virtual são implementadas e uma análise da influência destes sobre a probabilidade de bloqueio e a imparcialidade da rede é executada. / The Wavelength Division Multiplexing (WDM) technology as well as both the introduction of all optical OXCs (Optical Cross Connect) and OADMs (Optical Add/Drop Multiplexer) may provide the optical network with the networking function, i.e, the capacity to manipulate wavelengths in order to implement their routing. This possibility implies a new type of relationship between applications and the physical layer. The likely candidate to establish such relationships is GMPLS architecture. Efficient solutions to both the problems of allocating resources and traffic routing become an enhanced requirement in the design, expansion and management of optical networking. The present study focus at the relationship between functionalities such as traffic grooming, network fairness improvement, protection/restoring mechanisms and wavelength conversion in heterogeneous optical networks, evaluating the metrics of blocking probability, probability of traffic blocked and fairness. These functionalities are separately treated in literature. With this goal in mind, a two-layer representation is used in order to model the network: the physical and virtual layers, respectively. Two policies on traffic grooming are set up, as follows: MrTV (route minimizing on virtual topology) and MrTF (route minimizing on physical topology). The performance of such policies is analyzed regarding the percentage of blocked traffic. Next, a mechanism for call admission control (CAC) is implemented and its influence in terms of fairness and blocking probability is discussed. The simulations of optical networks such as NSFnet and the Italian National Network are carried out through a graph based in Zhu and Mukherjee [28]. As a result, MrTF policy presents a smaller percentage of blocked traffic than the MrTV for the simulated networks and the rolling window mechanism has allowed the optimization of the call admission control (CAC) mechanism. The usefulness of placing wavelength converters in a few networks nodes (spare conversion) is studied and an analysis on the blocking probability is presented. Next, the network fairness for this distribution of wavelength converters is presented. Finally, techniques for restoring both physical and virtual layers are also implemented and an analysis regarding their influence on the blocking probability and the network fairness is carried out.

A interligação de várias redes de telecomunicação ampliou a cobertura, mas tornou a operabilidade entre elas complexa, principalmente por causa da arquitetura resultante, formada por várias camadas. Estas camadas lidam com protocolos e taxas de transmissão diferentes e com sinais elétricos e ópticos. Diante desse cenário, a alternativa usual de expandir os recursos proporcionalmente ao crescimento da demanda é inviável devido ao elevado custo. Assim, soluções eficientes que agregam os benefícios das tecnologias ópticas e eletrônicas na arquitetura de nós e no gerenciamento de tráfego tornaram-se uma necessidade importante no projeto, expansão e gerenciamento de redes de telecomunicação. Os nós que permitem a comutação de tráfego na camada óptica e eletrônica e a agregação de tráfego em várias granularidades têm sido empregados para fazer o melhor uso possível dos recursos disponíveis nas redes. Esses nós são conhecidos na literatura como MG-OXCs. Nessa pesquisa, foi proposta uma abordagem dos nós MG-OXCs de uma e três camadas com a inclusão de recursos como conversão de comprimento de onda e agregação de tráfego. Também foi proposto um método de cálculo do custo desses nós baseado no volume de utilização das portas. Posteriormente foram inseridos fatores de degradação do sinal óptico para a análise da camada física de redes com os nós MGOXCs. Adicionalmente, para o gerenciamento de tráfego, foi proposto um esquema de escolha de comprimentos de onda chamado de canal específico e um esquema de monitoramento baseado na intensidade de tráfego. A meta principal é a redução da probabilidade de bloqueio de solicitação de conexão. Em redes ópticas, os modelos de nós propostos conseguem diminuir o número de portas dos nós comutadores tradicionais além de realizar a comutação de tráfego de conexões de diferentes valores de largura de banda. O esquema do canal específico consegue melhorar a utilização da largura de banda dos comprimentos de onda e diminui a probabilidade de bloqueio e o esquema do monitoramento diminui o número de conexões bloqueadas devido ao aumento dos recursos. Resultados numéricos apresentados demonstram a potencialidade dos algoritmos propostos para gerenciar recursos e rotear o tráfego das redes de telecomunicação. / Integrating telecommunication networks has enlarged the coverage, but has made operations more complex, mainly because of the architecture, formed by various layers. These layers deal with different protocols and transmission rates, as well as electrical and optical signals. The usual alternative of expanding the resources proportionally to the demand is impractical due to the high cost. Therefore, efficient solutions which add to the benefits of optical and electronic technology in node architecture and traffic management are essential in the design, expansion and management of telecommunications networks. The nodes that enable traffic switching in the optical and electronic layer and traffic grooming have been used to make the best use of the available resources in the networks. These nodes are known as MG-OXCs. In this research, an approach based on MG-OXCs was proposed with one and three layers, which include the wavelength conversion and traffic grooming. Additionally, a method to calculate the node costs based on their use of ports was proposed. The work also considered optical signal impairments in order to analyze the network physical layer with MG-OXCs nodes. For the purpose of traffic management, a scheme which sets specific wavelengths for different bandwidths and a scheme which monitors the flow of traffic were proposed. The main aim is to reduce the blocking probability of connection requests. In optical networks, the node models proposed are able to reduce the number of ports used in usual optical cross connects and switch the traffic connections using different bandwidths. The scheme of setting specific wavelengths for different bandwidths improves the bandwidth use and the blocking probability. The scheme which monitors the flow of traffic achieves blocking probability reduction due to the increase in resources. The numerical results presented show the feasibility of the proposed algorithms to manage resources and switch traffic in telecommunication networks.

The objective of this dissertation is to investigate the factors that influence the design of wide-area wavelength division multiplexing (WDM) optical networks. Wide-area networks are presented as communication networks capable of transporting voice and data communication over large geographical areas. These networks typically span a whole country, region or even continent.The rapid development and maturation of WDM technology over the last decade have been well-received commercially and warrants the development of skills in the field of optical network design.The fundamental purpose of all communication networks and technologies is to satisfy the demand of end-users through the provisioning of capacity over shared and limited physical infrastructure. Consideration of the business aspects related to communications traffic and the grooming thereof are crucial to developing an understanding of customer requirements in terms of the selection and quality of services and applications. Extensive communication networks require complex management techniques that aim to ensure high levels of reliability and revenue generation.An integrated methodology is presented for the design of wide-area WDM optical networks. The methodology harnesses physical, logical, and virtual topologies together with routing and channel assignment (RCA) and clustering processes to enhance objectivity of the design process. A novel approach, based on statistical clustering using the Ward linkage as similarity metric, is introduced for solving the problem of determining the number and positions of the backbone nodes of a wide-area network, otherwise defined as the top level hub nodes of the multi-level network model. The influence of the geographic distribution of network traffic, and the intra/inter-cluster traffic ratios are taken into consideration through utilisation of modified gravity models and novel network node weighting. / Dissertation (MEng)--University of Pretoria, 2005. / Electrical, Electronic and Computer Engineering / unrestricted

Les réseaux optiques sont aujourd’hui l’élément de base des systèmes de communica- tions modernes, en particulier l’Internet. Grâce au multiplexage en longueurs d’onde et au groupage du tra?c, la bande passante disponible sur une ?bre optique est supérieure à plusieurs térabits par seconde. Cependant les équipements opto-électroniques qui permettent d’opérer ces réseaux sont très coûteux car il doivent fonctionner à un débit très important. Le problème de groupage et du routage d’un ensemble de requêtes couplé avec l’affectation des longueurs d’onde (GRWA) est donc un problème stratégique de première importance. L’objectif est de minimiser le coût du réseau, évalué comme le nombre de ports optiques installés aux nœuds. Il peut être modélisé sous la forme d’un problème d’allocation de ressources dans un réseau à capacité multi-niveaux avec multi-?ots non bifurqués. Cette catégorie de problème est connue pour être très dif?cile compte tenu de la faiblesse de la relaxation linéaire des formulations associées. Les travaux réalisés durant cette thèse ont consisté en le développement de méthodes de résolution pour ce problème à partir de multiples techniques de recherche opéra- tionnelle : méta-heuristique de type recherche avec tabous, décomposition de Dantzig- Wolfe, décomposition de Benders, reformulation en variables binaires, méthode de plans coupants, heuristique d’arrondi. Les méthodes résultantes, dont certaines sont hybrides, permettent d’avoir un aperçu des méthodes ef?caces pour ce type de problème. En partic- ulier, les méthodes basées sur la décomposition de Benders, qui donnent lieu à des procédures d’optimisation hiérarchique dans lesquelles l’affectation de longueurs d’onde est placée au dernier niveau, sont les méthodes les plus ef?caces car elles permettent de séparer le routage optique du routage physique. En?n, nous utilisons la meilleure méthode de résolution pour observer l’impact des contraintes de délais sur la qualité des solutions. / Optical networks are the core element of modern communication systems and in particu- lar Internet. With wavelength multiplexing and grooming capability, terabits per second bandwidth can be reached. However, opto-electronic equipment used to operate these networks are very expensive as their bit rate must be very large. The grooming, routing and wavelength assignment (GRWA) problem, which consists in minimizing the net- work cost, evaluated by the number of required optical ports, while guaranteeing that each request is granted, is of great interest. The GRWA problem can be modeled as a multi-layer capacitated network design problem with non-bifurcated multi-?ows. This type of problem is known to be hard to solve as their linear relaxation is weak. The objective of this work was to develop solution methods based on multiple oper- ations research techniques : Tabu search based meta-heuristic, Dantzig-Wolfe decompo- sition, Benders decomposition, 0 1 reformulation, cutting-planes, rounding heuristic. The resulting solution tools, some of them hybrid, give a perspective on the effective solution approaches for this type of problem. From the experiments, it turns out that the methods based on Benders’ decomposition, which lead to hierarchical optimization procedures, are the most ef?cient as they allow to separate the optical routing from the physical routing with the wavelength assignment decisions taken in the lower stage sub- problem. In addition to the approach comparison, we use the most effective method to evaluate the impact of the delay constraints on the solution quality.

No description available.