Zhang, Yin. / Thesis submitted in: December 2008. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2009. / Includes bibliographical references (leaves 56-60). / Abstracts in English and Chinese. / Chapter Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Telecommunications network hierarchy --- p.2 / Chapter 1.2 --- PON architectures for access networks --- p.4 / Chapter 1.2.1 --- TDM-PON --- p.5 / Chapter 1.2.2 --- WDM-PON --- p.7 / Chapter 1.3 --- Data delivery mode in WDM-PON --- p.8 / Chapter 1.3.1 --- Point-to-point --- p.8 / Chapter 1.3.2 --- Broadcast --- p.9 / Chapter 1.3.3 --- Multicast --- p.10 / Chapter 1.4 --- Motivation of this thesis --- p.10 / Chapter 1.5 --- Outline of this thesis --- p.13 / Chapter Chapter 2 --- Previous Multicast Architectures in WDM-PON --- p.14 / Chapter 2.1 --- Introduction --- p.15 / Chapter 2.2 --- Previous WDM-PON architectures with multicast capability --- p.15 / Chapter 2.2.1 --- Subcarrier multiplexing --- p.16 / Chapter 2.2.2 --- All-optical based multicast enabled architecture --- p.18 / Chapter 2.3 --- Summary --- p.21 / Chapter Chapter 3 --- A Multicast enabled WDM-PON Architecture Using ASK-DPSK Orthogonal Modulation --- p.23 / Chapter 3.1 --- Introduction --- p.24 / Chapter 3.2 --- System architecture --- p.25 / Chapter 3.3 --- Experimental Demonstration --- p.27 / Chapter 3.4 --- Discussion --- p.31 / Chapter 3.5 --- Summary --- p.36 / Chapter Chapter 4 --- A WG filtering and its suppression in quaternary ASK-DPSK based multicast enabled WDM-PON --- p.37 / Chapter 4.1 --- Introduction --- p.38 / Chapter 4.2 --- Principle of narrowband filtering --- p.38 / Chapter 4.3 --- Simulation model --- p.40 / Chapter 4.4 --- Simulation results and discussion --- p.42 / Chapter 4.4.1 --- Different extinction ratios --- p.43 / Chapter 4.4.2 --- Different AWG filter shape and bandwidth --- p.47 / Chapter 4.5 --- Summary --- p.50 / Chapter Chapter 5 --- Summary and Future Works --- p.51 / Chapter 5.1 --- Summary of the thesis --- p.52 / Chapter 5.2 --- Future works --- p.53 / List of Publications --- p.55 / BIBLIOGRAPHY --- p.56

Two research topics in next-generation optical networks with wavelength-division multiplexing (WDM) technologies were investigated: (1) scalability of network management and control, and (2) resilience/reliability of networks upon faults and attacks. In scalable network management, the scalability of management information for inter-domain light-path assessment was studied. The light-path assessment was formulated as a decision problem based on decision theory and probabilistic graphical models. It was found that partial information available can provide the desired performance, i.e., a small percentage of erroneous decisions can be traded off to achieve a large saving in the amount of management information. In network resilience under malicious attacks, the resilience of all-optical networks under in-band crosstalk attacks was investigated with probabilistic graphical models. Graphical models provide an explicit view of the spatial dependencies in attack propagation, as well as computationally efficient approaches, e.g., sum-product algorithm, for studying network resilience. With the proposed cross-layer model of attack propagation, key factors that affect the resilience of the network from the physical layer and the network layer were identified. In addition, analytical results on network resilience were obtained for typical topologies including ring, star, and mesh-torus networks. In network performance upon failures, traffic-based network reliability was systematically studied. First a uniform deterministic traffic at the network layer was adopted to analyze the impacts of network topology, failure dependency, and failure protection on network reliability. Then a random network layer traffic model with Poisson arrivals was applied to further investigate the effect of network layer traffic distributions on network reliability. Finally, asymptotic results of network reliability metrics with respect to arrival rate were obtained for typical network topologies under heavy load regime. The main contributions of the thesis include: (1) fundamental understandings of scalable management and resilience of next-generation optical networks with WDM technologies; and (2) the innovative application of probabilistic graphical models, an emerging approach in machine learning, to the research of communication networks.

Internet traffic has surpassed voice traffic and is dominating in transmission networks. The Internet Protocol (IP) is now being used to encapsulate various kinds of services. The new services have different requirements than the initial type of traffic that was carried by the Internet network and IP. Interactive services such as voice and video require paths than can guarantee some bandwidth level, minimum delay and jitter. In addition service providers need to be able to improve the performance of their networks by having an ability to steer the traffic along the less congested links or paths, thus balancing the load in a uniform way as a mechanism to provide differentiated service quality. This needs to be provided not only within their domains but also along paths that might traverse more than one domain. For this to be possible changes have been proposed and some are being applied to provide quality of service (QoS) and traffic engineering (TE) within and between domains. Because data networks now carry critical data and there are new technologies that enable providers to carry huge amount of traffic, it is important to have mechanisms to safeguard against failures that can render the network unavailable. In this thesis we propose and develop mechanisms to enable interdomain traffic engineering as well as to speed up the restoration time in optical transport networks. We propose a mechanism, called abstracted path information, that enable peering entities to exchange just enough information to engage in QoS and TE operations without divulging all the information about the internal design of the network. We also extend BGP to carry the abstracted information. Our simulations show that BGP could still deliver the same performance with the abstracted information. In this thesis we also develop a method of classifying failures of links or paths. To improve the restoration time we propose that common failures be classified and assigned error type numbers and we develop a mechanism for interlayer communication and faster processing of signalling messages that are used to carry notification signals. Additionally we develop a mechanism of exchanging the failure information between layers through the use of service primitives; that way we can speed up the restoration process. Finally we simulate the developed mechanism for a 24 node Pan American optical transport network. / <p>QC 20100913</p>

Αντικείμενο της παρούσας διπλωματικής εργασίας αποτελεί η μελέτη της τηλεπικοινωνιακής κίνησης σε οπτικά δίκτυα δρομολόγησης μήκους κύματος. Στα πλαίσια της αποτίμησης της απόδοσής τους, υπολογίζεται η πιθανότητα φραγής κλήσης (Call Blocking Probability) για ένα σύνολο δοσμένων διαδρομών. Αφού αρχικά γίνεται αναφορά στην ιστορία και την εξέλιξη της τεχνολογίας των οπτικών ινών και των οπτικών δικτύων, το θέμα της εργασίας επικεντρώνεται στα οπτικά δίκτυα σταθερής δρομολόγησης και σταθερής εναλλακτικής δρομολόγησης. Γίνεται χρήση δύο αναλυτικών μοντέλων υπολογισμού της πιθανότητας απωλειών σε αυτά: το μοντέλο Birman και το μοντέλο Sridharan-Sivarajan, τα οποία υλοποιούνται στη γλώσσα προγραμματισμού C. Τα αποτελέσματα των μεθόδων ανάλυσης για συγκεκριμένα παραδείγματα συγκρίνονται με αυτά της προσομοίωσης, που υλοποιείται με χρήση της γλώσσας προσομοίωσης SIMSCRIPT II.5. Επιπρόσθετα, εξετάζεται η τεχνική της ομοιοκαταχώρησης τηλεπικοινωνιακής κίνησης (Traffic grooming) με προσομοίωση της λειτουργίας οπτικών δικτύων που κάνουν χρήση της τεχνικής αυτής. / The main objective of this diploma thesis is the study of the telecommunication traffic in Wavelength-Routed Optical Networks. In the frames of their performance evaluation, Call Blocking Probabilities for given routes are calculated. Firstly, the history and evolution of optical fibers and optical networks are described. Followingly, the present thesis is focused in Wavelength-Routed Optical Networks for two routing schemes: Fixed Routing and Fixed Alternate Routing. Two analytical models are used for the calculation of blocking probabilities for these optical networks: Birman model and Sridharan-Sivarajan model, which are realized in C programming language. Afterwards, both these analytical techniques and the simulation, which is realised in SIMSCRIPT II.5 simulation language, are applied in particular examples and their equivalent results are compared. In addition, the Traffic-Grooming technique is examined by simulating the operation of optical networks which use this specific technique.

Passive optical networks (PONs) are point-to-multipoint networks where a single Central Office (CO) is connected to a number of downstream Optical Network Units (ONUs) via a single optical fiber by splitting the optical signal with passive splitters. Due to technology advances and increasing bandwidth requirements, these networks have moved to last mile deployment, also known as fiber-to-the-home (FTTH). The planning of these PONs are traditionally done by hand, but automated methods can be used to decrease deployment costs and planning time. Even though a number of methods have been proposed to address this problem through the solving of integer linear programming (ILP) models, they suffer from limited availability, inaccuracies and limited scalability due to the problem complexity. This dissertation focusses on improving the accuracy of these models as well as improving scalability to a point where large-scale problems can be solved feasibly. To address this, a basic model is implemented to capture the network structure and verified accordingly. Results show this model can be solved quickly, but has large discrepancies with real-world plans. Refinements in the form of fiber duct sharing, network constraints, multiple splitter types and economies of scale among others are then incorporated into a refined model and solved. Analysis of the experimental results indicates improved accuracy and lower deployment costs, at the expense of increasing computation effort considerably. Heuristic techniques are then examined to improve computational performance, including an elementary heuristic (ELEM), the Branch Contracting Algorithm (BCA) and problem decomposition. It is demonstrated that through the use of k-means clustering, the refined model can be solved in a fraction of the time while keeping deployment costs comparably low. / Thesis (MIng (Computer and Electronic Engineering))--North-West University, Potchefstroom Campus, 2013

Passive optical networks (PONs) are point-to-multipoint networks where a single Central Office (CO) is connected to a number of downstream Optical Network Units (ONUs) via a single optical fiber by splitting the optical signal with passive splitters. Due to technology advances and increasing bandwidth requirements, these networks have moved to last mile deployment, also known as fiber-to-the-home (FTTH). The planning of these PONs are traditionally done by hand, but automated methods can be used to decrease deployment costs and planning time. Even though a number of methods have been proposed to address this problem through the solving of integer linear programming (ILP) models, they suffer from limited availability, inaccuracies and limited scalability due to the problem complexity. This dissertation focusses on improving the accuracy of these models as well as improving scalability to a point where large-scale problems can be solved feasibly. To address this, a basic model is implemented to capture the network structure and verified accordingly. Results show this model can be solved quickly, but has large discrepancies with real-world plans. Refinements in the form of fiber duct sharing, network constraints, multiple splitter types and economies of scale among others are then incorporated into a refined model and solved. Analysis of the experimental results indicates improved accuracy and lower deployment costs, at the expense of increasing computation effort considerably. Heuristic techniques are then examined to improve computational performance, including an elementary heuristic (ELEM), the Branch Contracting Algorithm (BCA) and problem decomposition. It is demonstrated that through the use of k-means clustering, the refined model can be solved in a fraction of the time while keeping deployment costs comparably low. / Thesis (MIng (Computer and Electronic Engineering))--North-West University, Potchefstroom Campus, 2013

Optical WDM network is the suitable transport mechanism for ever increasing bandwidth intensive internet applications. The WDM technique transmits the data over several different wavelengths simultaneously through an opticalfiber and the switching is done at wavelength level. The connection between the source and destination is called the light path. Since the WDM network carries huge amount of tra c, any failure can cause massive data loss. Therefore protecting the network against failure is an important issue. Maintaining high level of service availability is an important aspect of service provider. To provide cost effective service, all-optical network is the suitable choice for the service provider. But in all optical network, the signals are forced to remain in optical domain from source to destination. In the firrst part of the thesis, we deal the physical layer impairments (PLIs) aware shared-path provisioning on a wavelength routed all-optical networks. As the signal travels longer distances, the quality of the signal gets degraded and the receiver may not be able to detect the optical signal properly. Our objective is to establish a light path for both the working path and protection path with acceptable signal quality at the receiver. We propose an impairment aware integer linear programming (ILP) and impairment aware heuristic algorithm that takes into account the PLIs. The ILP provides the optimal solution. It is solved using IBM ILOG CPLEX solver. It is intractable for large size net-work. Therefore we propose the heuristic algorithm for large size network. It is evaluated through discrete-event simulation. But the algorithm provides only the suboptimal solution. To know the performance of this algorithm, the simulation result is compared with the optimal solution. We compute total blocking probability, restoration delay, computation time, and connection setup delay with respect to network load for the heuristic algorithm. We compare the performance of shared-path protection with dedicated-path protection and evaluate the percentage of resource saving of shared-path protection over the dedicated-path protection. In the second and third part of the thesis, we address the issues related to flexible-grid optical networks. In wavelength routed optical network, the bandwidth of each wavelength is fixed and rigid. It supports coarse grained tra c grooming and leads to ancient spectrum utilization. To overcome this, flexible-grid optical networks are proposed. It supports flexible bandwidth, and ne grained tra c groom In the second part of the thesis, we address the routing and spectrum allocation (RSA) algorithm for variable-bit-rate data tra c for flexible-grid optical networks. The RSA problem is NP-complete. Therefore a two-step heuristic approach (routing and spectrum allocation) is proposed to solve the RSA problem. The first step is solved by using a classical shortest path algorithm. For the second step we propose two heuristic schemes for frequency-slot allocation: (i) largest number of free frequency-slot allocation scheme and (ii) largest number of free frequency-slot maintaining scheme. As the network load increases, the spectrum is highly fragmented. To mitigate the fragmentation of the spectrum, we propose a xed-path least-fragmentation heuristic algorithm which fragments the spectrum minimally. It also supports varying-bit-rate tra c and also supports dynamic arrival connection requests. Through extensive simulations the proposed algorithms have been evaluated. Our simulation results show that the algorithms perform better in terms of spectrum utilization, blocking probability, and fraction of fragmentation of the spectrum. The spectrum utilization can reach up to a maximum of 92% and that only 71% of the spectrum is fragmented under maximum network load condition. Finally in the third part of the thesis, we discuss PLIs-aware RSA for the transparent exible-grid optical network. In this network, not only the optical signal expected to travel longer distance, but also to support higher line rates, i.e., data rate is increased up to 1 Tb/s. In such a high data rate, the optical signals are more prone to impairments and noises. As the transmission distance increases, optical signals are subject to tra-verse over many bandwidth-variable wavelength cross connects (BV-WXC) and multiple fibber spans due to which the PLIs get accumulated and are added to the optical signal. These accumulated impairments degrades the signal quality to an unacceptable level at the receiver, the quality of transmission falls below the acceptable threshold value, and the receiver may not be able to detect the signal properly. Therefore our objective is to develop an impairment aware RSA algorithm which establishes the QoT satisfied empathy based on the available resources and the quality of the signal available at the receiver. We formulate the PLIs-RSA problem as an ILP that provides an optimal solution. The optimal solution is obtained by solving the ILP using IBM ILOG CPLEX optimization solver. Since ILP is not efficient for large-size networks, we propose a heuristic algorithm for such a large-size networks. The signal power is measured at the receiver and the connection is established only when the signal power lies above the threshold value. The heuristic algorithm is evaluated through discrete-event simulation. It gives the sub-optimal solution. The simulation result is compared with optimal solution. The result shows that heuristic algorithm performs closer to the ILP. We compute the total blocking probability versus the network load for different spectrum allocation schemes. Total blocking probability is the sum of frequency-slot blocking probability and QoT blocking probability. We compute spectrum efficiency for the proposed algorithm. We also compare our algorithm with the existing routing and spectrum allocation algorithm, and the result shows that our algorithm outperforms the existing algorithms in terms of blocking probability and spectrum utilization.

In this research, we consider the impact of spectrum fragmentation in optical single-/multi-path routing transmission on the efficiency of the elastic optical networks. O-OFDM multicarrier transmission is a promising technique that makes it possible to choose just an adequate portion of available spectrum to satisfy the requested capacity. This involves focusing on the work to reduce the fragmentation effects by dynamically updating and controlling the minimum bandwidth allocation granularity. that serves the light path requests over multipath networks. We adopt linear and nonlinear dynamic mechanisms, which are denoted as LDAɡ and NLDAɡ that are proportional to the optical link/path bandwidth fragmentation status. Simulation results show that the minimum bandwidth granularity dynamic adaptation based on the optical path fragmentation status offers improved performance over fixed minimum bandwidth allocation granularity with respect to the bandwidth blocking probability, the throughput, the network bandwidth utilization and the number of path splitting.

As redes ópticas têm evoluído de forma contínua dentro de um paradigma de aumento das taxas de transmissão e extensão dos enlaces, devido à demanda crescente de banda em função do crescimento do tráfego da Internet. Além disso, atualmente, diversas propostas vêm sendo implementadas visando torná-las mais dinâmicas e flexíveis. Uma destas propostas que atualmente está no âmbito de pesquisa e desenvolvimento refere-se às redes ópticas definidas por software (Software Defined Optical Network, SDON). Nas SDONs, o plano de controle é desacoplado do plano de encaminhamento de dados possibilitando que controladores remotos configurem em tempo real diversos parâmetros dos canais ópticos, tais como a taxa de transmissão, o formato de modulação, a largura do espectro, entre outros. Nestas redes, o sistema de controle torna-se bastante complexo, uma vez que diversos parâmetros têm que ser ajustados de forma dinâmica e autônoma, ou seja, com a mínima intervenção humana. O emprego de técnicas de inteligência computacional em tal controle possibilita a configuração autônoma dos parâmetros dos equipamentos com base em dados coletados por monitores de rede e o aprendizado, a partir de eventos passados, visando a otimização do desempenho da rede. Esta arquitetura de controle constitui um novo paradigma na evolução das redes ópticas, as denominadas Redes Ópticas Cognitivas. A escolha de uma técnica de inteligência computacional adequada para tomada de decisão em redes ópticas é importante para se obter vantagens no uso da cognição. Esta técnica deve possibilitar o aprendizado e ainda minimizar a complexidade computacional, uma vez que a configuração dos parâmetros da rede deve ocorrer em tempo real. Neste contexto, esta tese investiga o uso do algoritmo de agrupamento Fuzzy C-Means (FCM) para aprendizado e tomada de decisão em redes ópticas flexíveis de próxima geração. FCM possibilita a geração automática de regras com base na experiência adquirida no meio de operação (aprendizado) e a tomada de decisão a partir destas regras. Uma comparação de desempenho entre os algoritmos FCM e CBR (Case-Based Reasoning) é apresentada. O algoritmo CBR foi escolhido para esta comparação devido a ter sido utilizado recentemente, com sucesso, em redes ópticas cognitivas. Por fim, um conceito de rede óptica cognitiva é apresentado. / Optical networks have evolved continuously increasing the transmission rate and the extension of links due to the increased bandwidth consuming. Moreover, currently, several proposals are under development to make the next generation optical network more dynamic and flexible. The term \"flexible\" refers to the ability of dynamically adjust the parameters of the optical network such as modulation format, transmission rate, optical bandwidth and others, according with the quality of transmission of each lightpath. In this scenario, a Software Defined Optical Network (SDON) emerges as a new optical network paradigm, where the control plane is decoupled from the data plane, enabling remote controllers to configure network equipment from different hardware vendors, which allows a degree of software programmability to the network. In SDON, the control plane needs to include functionalities to operate autonomously, i.e, with minimal human intervention. The use of the computational intelligence techniques in such control plane enables the autonomous operation and learning based on past events, in order to optimize the network performance. This architecture represents a new paradigm in the evolution of optical networks, resulting in so-called Cognitive Optical Networks. The choice of a computational intelligence technique for learning and decisionmaking in such optical networks is essential to bring advantages with the use of cognition. This technique should minimize the computational complexity, since the configuration of the network parameters must occur in real time.In this context, this thesis investigates the use of Fuzzy C-Means clustering algorithm (FCM) for learning and decision-making in the software defined optical networks context. FCM enables the automatic generation of rules, based on the experience gained during the network operation. Then, these rules are used by the control plane to take decisions about the lightpaths\' configuration. A comparison of performance between the FCM and the CBR (Case-Based Reasoning) algorithm. CBR algorithm was chosen because it has been successfully used in cognitive optical networks. Finally, we propose a concept for optical cognitive network.

O principal objetivo deste trabalho é analisar a arquitetura das redes ópticas, discutir o estado da arte dos protocolos necessários ao seu desenvolvimento, e avaliar os vários esforços empregados para a padronização destas por diferentes instituições. Conceitualmente, redes ópticas são aquelas nas quais a tecnologia dominante na camada física é a fibra óptica. Tais redes originaram-se no início dos anos 80 com a utilização dos cabos de fibras ópticas monomodo, tendo evoluído com a introdução de uma nova base tecnológica composta por novos tipos de fibras ópticas, amplificadores ópticos, sistemas DWDM e componentes ópticos, como multiplexadores e demultiplexadores add/drop e cross-connects ópticos. O desenvolvimento destes componentes possibilitará a evolução dos simples enlaces DWDM ponto a ponto para a rede totalmente óptica. Neste trabalho é analisado como as aplicações interagirão com esta camada física. Tal interação tem provado ser um desafio dentro da atual arquitetura das redes de dados, no intuito de disponibilizar soluções que habilitem o transporte de um grande volume de tráfego de forma eficiente. Neste sentido é analisado o emprego de tecnologias como o MPLS (Multiprotocol Label Switching) e sua evolução, o Generalized MPLS (GMPLS). Finalmente apresenta-se a versão do ITU para arquitetura de redes ópticas, e a conseqüente inserção do GMPLS nesta padronização / The main objective of this work is to analyze the evolution trends for the architecture of the optical networks, discuss the state of the art protocols required for its development and evaluate the various efforts being carried out by different institutions in order to achieve standardization. In this context, optical networks are those in which the dominant technology in the physical layer is the optical fiber. Such networks arose in the early eighties with the use of the monomodo fiber optic cables and were further developed with the introduction of a new technological base composed by new types optical fibers, optical amplifiers, DWDM systems and optical components such as add/drop multiplexers and optical cross-connects. The introduction of these components will make possible the evolution from the simple point-to-point DWDM connections to the all optical networks. In this work we analyze how the applications will interact with this physical layer. Such interaction has been proving to be a challenge within the framework for the current architecture of data networks, aiming at offering solutions to enable the transport of a great volume of traffic in an efficient manner. With this goal in mind, the use of technologies such as MPLS (Multiprotocol Label Switching) and its evolution, the Generalized MPLS (GMPLS) is discussed. Finally, we present the ITU version for the architecture optical networks, addressing its merging with the GMPLS