Along with playing an ever-increasing role in the integration of other communication networks and expanding in application diversities, the current Internet suffers from serious overuse and congestion bottlenecks. Efficient congestion control is fundamental to ensure the Internet reliability, satisfy the specified Quality-of-Service (QoS) constraints and achieve desirable performance in response to varying application scenarios. Active Queue Management (AQM) is a promising scheme to support end-to-end Transmission Control Protocol (TCP) congestion control because it enables the sender to react appropriately to the real network situation. Analytical performance models are powerful tools which can be adopted to investigate optimal setting of AQM parameters. Among the existing research efforts in this field, however, there is a current lack of analytical models that can be viewed as a cost-effective performance evaluation tool for AQM in the presence of heterogeneous traffic, generated by various network applications. This thesis aims to provide a generic and extensible analytical framework for analyzing AQM congestion control for various traffic types, such as non-bursty Poisson and bursty Markov-Modulated Poisson Process (MMPP) traffic. Specifically, the Markov analytical models are developed for AQM congestion control scheme coupled with queue thresholds and then are adopted to derive expressions for important QoS metrics. The main contributions of this thesis are listed as follows: iii ¿ Study the queueing systems for modeling AQM scheme subject to single-class and multiple-classes Poisson traffic, respectively. Analyze the effects of the varying threshold, mean traffic arrival rate, service rate and buffer capacity on the key performance metrics. ¿ Propose an analytical model for AQM scheme with single class bursty traffic and investigate how burstiness and correlations affect the performance metrics. The analytical results reveal that high burstiness and correlation can result in significant degradation of AQM performance, such as increased queueing delay and packet loss probability, and reduced throughput and utlization. ¿ Develop an analytical model for a single server queueing system with AQM in the presence of heterogeneous traffic and evaluate the aggregate and marginal performance subject to different threshold values, burstiness degree and correlation. ¿ Conduct stochastic analysis of a single-server system with single-queue and multiple-queues, respectively, for AQM scheme in the presence of multiple priority traffic classes scheduled by the Priority Resume (PR) policy. ¿ Carry out the performance comparison of AQM with PR and First-In First-Out (FIFO) scheme and compare the performance of AQM with single PR priority queue and multiple priority queues, respectively.

Queueing theory is widely used in practical queuing applications. It can be applied for specific models of queuing systems, especially the ones that follow the Markovian property. Its purpose is to predict system behaviour in order to be used for performance optimization. In this case study, it was used to evaluate an extended queuing model with agents serving multiple queues. The purpose was to try to capture more variability and input factors into the theoretical model and test its applicability on more extended models. The main objective was to use relevant queuing theory models to estimate the wait time using real contact center data. Different from the theoretical model, the service rates of the system model depended on how many queues an agent served concurrently, which increased the complexity of the model. The obtained results demonstrated some limitations that made the models too restrictive to be applied to a model with multi-skilled agents that were not equally available. Moreover, it was shown that heuristical approaches might be more suitable for more complex queuing systems that are not covered in queueing theory models. / Köteori används i stor utsträckning i praktiska kö-applikationer. Den kan tillämpas för specifika modeller av kö-system, speciellt de som följer Markovegenskapen. Dess syfte är att förutse systembeteende för att kunna användas för prestandaoptimering. I denna fallstudie användes den för att utvärdera en utökad kömodell med agenter som betjänade flera köer. Syftet var att försöka fånga mer variabilitet och inputfaktorer i den teoretiska modellen och testa dess tillämplighet för mer utökade modeller. Huvudmålet var att använda relevanta kö-teorimodeller för att estimera väntetiden med användning av riktiga contact center data. Till skillnad från den teoretiska modellen, betjäningsintensiteten för systemmodellen berodde på hur många köer en agent betjänade samtidigt, vilket ökade komplexiteten av modellen. De erhållna resultaten visade begränsningar som gjorde modellen för restriktiv för att appliceras på en modell med fler-kvalificerade agenter som inte var lika tillgängliga. Utöver detta så visade det sig att heuristiska metoder kan vara mer lämpliga för mer komplicerade system som inte täcks av kö-teori modeller.

Οι οικονομικές και τεχνολογικές εξελίξεις των τελευταίων ετών, η απελευθέρωση της αγοράς, οι ισχυρές κατά απαίτηση πολυμεσικές υπηρεσίες καθώς και ο αυξημένος αριθμός χρηστών των κινητών δικτύων υποδεικνύουν την αναγκαιότητα της σύγκλισης των δύο δικτυακών τεχνολογιών (κινητή τηλεφωνία και internet) με στόχο την παροχή υπηρεσιών Internet στο περιβάλλον των κινητών επικοινωνιών. Η παραπάνω απαίτηση οδήγησε στην δημιουργία του δικτύου επόμενης γενιάς NGN. Η διπλωματική αυτή εργασία ασχολείται με την μελέτη της αρχιτεκτονικής του επιπέδου υπηρεσιών και την υποστήριξη Ποιότητας Υπηρεσίας (QoS) σε δίκτυα Νέας Γενιάς. Συγκεκριμένα δίνεται έμφαση στις λειτουργίες ελέγχου, σηματοδοσίας και λειτουργιών αρχιτεκτονικής του QoS σε επίπεδο υπηρεσιών. Αρχικά, ορίζουμε την έννοια “QoS” όσον αφορά την οπτική του δικτύου και το χρήστη. Περιγράφονται η αρχιτεκτονική του στρώματος υπηρεσιών του δικτύου καθώς και οι λειτουργίες ελέγχου πόρου και αποδοχής των κλήσεων, που αποτελούν σημαντικό μέρος της αρχιτεκτονικής του NGN. Στη συνέχεια παρουσιάζονται αρχιτεκτονικές που προορίζονται για την παροχή του QoS (IntServ, DiffServ), στη Συμφωνία Στάθμης Παρεχόμενης Υπηρεσίας (SLA), το πρωτόκολλο σηματοδοσίας COPS. Επίσης, δίνουμε έμφαση σε ορισμένες πτυχές (χρονοδρομολόγηση, διαχείριση ουρών) μίας QoS αρχιτεκτονικής, οι οποίες είναι ζωτικής σημασίας όσον αφορά την αποδοτική παροχή Ποιότητας Υπηρεσίας. Στη συνέχεια, με τη χρήση του εργαλείου προσομοίωσης OPNET, διεξάγουμε μια σειρά προσομοιώσεων σε ένα ATM και σε ένα NGN δίκτυο. Τέλος, παραθέτουμε και αναλύουμε τα αποτελέσματα των προαναφερθέντων πειραμάτων. / The concept of an NGN (Next Generation Network) has been introduced to take into consideration the new realities in the telecommunications industry, characterized by factors such as: competition among operators due to ongoing deregulation of markets, explosion of digital traffic, e.g.,increasing use of "the Internet", increasing demand for new multimedia services, increasing demand for a general mobility, convergence of networks and services, etc. This thesis has as subject the architecture of service stratum and presents an overview of standards functions defining the Quality of Service (QoS) in Next Generation Networks (NGNs). Several standards bodies define the QoS control architectures based on their scope of work. Specifically, emphasis is given to control functions, signalling and functional architecture of QoS in service stratum. Firstly, we define the meaning of QoS according to the view of the operator’s network and terminal users. The functional architecture of service stratum and especially the part of resource and admission control functions are described in the main body. Furthermore, architectures as IntServ and DiffServ, SLAs and COPS protocol are used as providers of the QoS. Scheduling and queuing management are necessary to optimize the QoS in NGN networks. In this project, we use OPNET simulator in two scenarios to determine construct and control ATM and NGN networks. Finally, collect the results of the experiments and analyze them.

Orientador: Nelson Luis Saldanha da Fonseca / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Computação / Made available in DSpace on 2018-08-13T10:35:13Z (GMT). No. of bitstreams: 1 Santi_Julianade_M.pdf: 1658984 bytes, checksum: 8a9f078587406a06815484e4fe057f7d (MD5) Previous issue date: 2008 / Resumo: A utilização eficiente da banda passante em redes de alta velocidade e grandes atrasos, denominadas redes com produto banda-atraso elevado (PBA), tornou-se um grande desafio. Isto ocorre devido aos ajustes do protocolo Transmission Control Protocol (TCP). O High Speed TCP (HSTCP), uma variante do TCP para redes com PBA elevado, emprega ajustes mais agressivos permitindo, assim, que a utilização da banda seja escalável. As políticas de Gerenciamento Ativo de Filas ou Active Queue Management (AQM), monitoram o nível de ocupação das filas nos roteadores e notificam o congestionamento incipiente aos emissores TCP através do descarte/marcação de pacotes. O sistema de controle de congestionamento apresenta natureza de retroalimentação, na qual a taxa de transmissão dos nós fontes é ajustada em função do nível de ocupação da fila. Os controladores AQM determinam a probabilidade de descarte/marcação para maximizar a vazão e minimizar perdas, garantindo, assim, a estabilidade do tamanho da fila independentemente das variações das condições da rede. Neste trabalho, define-se a política de gerenciamento ativo de filas HSTCP-H2 para redes com PBA elevado que utilizam o protocolo HSTCP. Para a derivação de HSTCP­H2: são utilizadas técnicas de Teoria de Controle Ótimo. A principal característica desta política é considerar o atraso do sistema o que permite melhor utilização dos recursos disponíveis. A estabilidade e os objetivos de desempenho do sistema são expressos e solu­cionados através de Desigualdades Matriciais Lineares, permitindo que os parâmetros do controlador possam ser calculados através da solução de um problema convexo simples. Diferentes controladores foram derivados considerando-se diferentes objetivos de de­sempenho, os quais consideram as características de redes com produto banda-atraso elevado. Através de simulações, os desempenhos dos controladores derivados são avalia­dos e a eficácia do controlador que apresentou o melhor desempenho foi comparado com o desempenho da política de AQM RED. São considerados cenários com enlace gargalo único e com múltiplos gargalos. / Abstract: The efficient utilization of bandwidth in high speed and large delay networks, called high bandwidth-delay product networks (BDP), has become a major challenge. This is due to adjustments of the Transmission Control Protocol (TCP). The High Speed TCP HSTCP): a TCP variant to high BDP networks, employs more aggressive adjustments, allowing scalable bandwidth utilization. The Active Queue Management (AQM) policies monitor the queue length in the routers and notify incipient congestion to TCP source by marking or dropping packets. The congestion control system presents intrinsic feedback nature, where the transmission rates of the sources are adjusted according to the level of congestion inferred by the queue occupancy. The AQM controllers determine the dropping marking probability values to maximize throughput and minimize losses, giving guarantees to stabilize the queue length independent of network conditions. In this work, it is defined HSTCP-H2, an active queue management policy to high BDP networks, which adopt the HSTCP as their transport protocol. Optimal control theory is used to conceive HSTCP-H2. The novelty of the proposed approach lies in consider the delay of the system which allows better use of available resources. Furthermore, in the proposed approach, stability and performance objectives are completely expressed as Linear Matrix Inequalities (LMIs), thus requiring the solution of a single convex problem for the computation of the controller parameters. Different controllers are derived considering different design goals, which take into ac­count the characteristics of the high bandwidth-delay product networks. The performance produced by different optimal controllers was investigated. The efficacy of the control­ler with the best performance was then compared to the performance of RED policy. The simulation experiments were carried out using topologies with single and multiple bottleneck. / Mestrado / Redes de Computadores / Mestre em Ciência da Computação