面對通訊與資訊科技的大幅進步、網際網路的蓬勃發展、以及電信自由化帶來的激烈競爭，通訊網路正在進行一個巨大的變革，企圖將原有Circuit Switching 與 Packet Switching 網路整合成一個單一整合型網路－All-IP網路以支援所有的應用服務。All-IP網路受限於封包交換網路原有的特性，有服務品質問題（QoS）有待克服，因此有必要在All-IP網路上提供服務品質管理機制以實現整合型網路的目標。而要提供適當的QoS 管理，其成功之關鍵主要在於是否能提供一個簡單易行之架構。本論文先提出BBQ(Budget-Based QoS)採用以預算為基礎之服務品質管理，以簡化管理、追求效率，不增加管理複雜度為原則。BBQ提供一個高適用性的管理架構和相關的管理工具，可適用於不同的下層網路架構和不同營運目標的網管政策。 本論文為提出在BBQ管理系統中之端對端服務品質解決方案。透過承載服務的概念，端對端之服務由接取網路和骨幹網路之承載服務提供支援。本論文依據分層負責的精神提出一系列的資源規劃及路徑建構方式，提高網路資源運用效率，並可快速的以即時方式建構具服務品質保證的端對端路徑給使用者。骨幹網路由許多核心網路相連而成，各個核心網路各自獨立規劃內部路徑而端對端的路徑規劃只需選擇所欲通過的核心網路即可，計算量可大幅降低，因此可適用即時的路徑建構。路徑規劃之研究重心為如何挑選最佳核心網路路徑，以規劃具服務品質之端對端路徑並可達到資源之最有效利用。 / The advance in communication and information technology and impact of telecommunication liberalization cause a revolution in telecommunication world. It attempt to merge circuit-switching and packet-switching network into one standalone perform. All-IP is one of those candidates. But All-IP also inherit the characteristics of packet-switching network, the problem of transmission quality. So, we need some management system of quality on All-IP network to achieve the ambition of converged network. In this thesis, we propose BBQ management system. BBQ offers a highly adaptive management architecture and some management tools for operators. Those tools can be used in different underlying network layer and can tune network by different network policy. We propose an End-to-End QoS solution in BBQ management system. By the concept of bearer services, we separate the End-to-End service into Backbone and Stub Network bearer services. According to the hierarchical management, we propose an approach for resource and path planning in order to enhance network efficiency and provide End-to-End path with QoS in real time. Because Backbone Network is connected by Core Networks and each Core Network plans its internal path independently, End-to-End path planning is to choose Core Networks to pass through. This scheme could reduce the enormous computation and fit for real time path setup. The key point of path planning is how to choose the better Core Network path to compose End-to-End path with QoS and reach efficient resource utilization.

現有大多數的QoS 繞徑演算法，均假設鏈結上的參數是固定的，而依據固定鏈結參數的條件，尋求最佳的繞徑。實際上鏈結上能保證的服務品質與鏈結實際的負載有關。如果鏈結的所承諾提供的服務品質及所承諾之負載，在最大容量維持不變的條件下，可以由繞徑程式自由選擇的話，繞徑程式可能可以做更好的路徑規劃。 本研究在BBQ (Budget-Based QoS)的架構下，期望藉由事先規劃鏈結上的參數設定，以提供較具彈性的路由規劃，提升整體網路的效能，幫助營運者靈活可擴展的管理網路，以到達營運者的營運目標。由於加入參數的調變，使得繞徑問題變的更為複雜，在本文中，我們設計了兩套不同的爬山式搜尋演算法，經反覆的模擬測試，發現本研究的方式的確能提供較佳的網路規劃。 / Most of current QoS routing algorithms assume fixed network parameters, such as guaranteed service quality and bandwidth capacity. In fact, guaranteed quality of links is load dependent. Assuming that the maximum capacity of each link is limited, if the guaranteed quality and bandwidth capacity can be chosen by a routing algorithm, then the routing algorithm may find better paths. For example, guaranteed quality could be improved by using lower capacity on the link or by using the maximum capacity with poor guarantee quality. Our research is based on BBQ (Budget-Based QoS) management architecture. We preplan the parameters on each link before routing. This method provides flexible routing options thereby increasing the total profit. We tested two different approaches of allocating parameters on each link. After several simulations, we found that these approaches that we developed could result in better network resource utilization and increase profit for the operator.

近十年來，由於無線網路的普及與人們對於即時服務的需求提高，導致人們迫切需要更好的服務品質，WiMAX是其中最被看好的一種無線網路傳輸技術。但在WiMAX無線網路中，標準的規格中並未規範connection admission control (CAC)、bandwidth request (BR)、bandwidth allocation、scheduling等機制，在本篇論文中，我們將上述機制設計並實作於MAC layer中。 本論文首先探討在設計connection admission control、bandwidth request、bandwidth allocation、scheduling會遇到的相關參數及相關議題。並進一步提出一個有效的方法以改善目前大部分設計在bandwidth allocation的公平性 (Fairness)及contention bandwidth request等效率差的問題。我們將設計一個MAC Layer co-function，稱之為Dynamic Polling Interval function (DPI function)。利用DPI function設計no contention bandwidth request改善傳統 contention bandwidth request的效率，以及利用DPI function的特性改善bandwidth allocation以及scheduling的公平性。最後我們將利用網路模擬器NS-2 (Network Simulater version 2)與測試實驗架構作不同效能的驗證比較並評估所提方法的有效性。 / Over the past decade, wireless network access and real-time services have become more popular than ever. People are eager to have better quality of service. Among all, WiMAX is one of the best wireless communication technigues . However, WiMAX standard does not specify those mechanisms of connection admision control (CAC)、bandwidth request (BR)、bandwidth allocation and scheduling . In this thesis, we propose the above mechanisms and imcorporate them as MAC layer functions. First, we discuss those related parameters and issues when designing connection admision control、bandwidth request、bandwidth allocation and scheduling. Second, we propose an efficient method to improve the fairness of bandwidth allocation and efficiency of contention bandwidth request. We design a MAC layer co-function called dynamic polling interval function (DPI function). We use the DPI function to design a no contention bandwidth request method to improve the efficiency of traditional bandwidth request method and use the features of DPI function to improve the fairness of bandwidth allocation and scheduling. At last , we use NS-2 (Network Simulator version 2) as our network simulator and compare the result of simulations to prove the efficiency of our proposed methods.

Transmission of video content over wireless access networks (in particular, Wireless Local Area Networks (WLAN) and Third Generation Universal Mobile Telecommunication System (3G UMTS)) is growing exponentially and gaining popularity, and is predicted to expose new revenue streams for mobile network operators. However, the success of these video applications over wireless access networks very much depend on meeting the user’s Quality of Service (QoS) requirements. Thus, it is highly desirable to be able to predict and, if appropriate, to control video quality to meet user’s QoS requirements. Video quality is affected by distortions caused by the encoder and the wireless access network. The impact of these distortions is content dependent, but this feature has not been widely used in existing video quality prediction models. The main aim of the project is the development of novel and efficient models for video quality prediction in a non-intrusive way for low bitrate and resolution videos and to demonstrate their application in QoS-driven adaptation schemes for mobile video streaming applications. This led to five main contributions of the thesis as follows:(1) A thorough understanding of the relationships between video quality, wireless access network (UMTS and WLAN) parameters (e.g. packet/block loss, mean burst length and link bandwidth), encoder parameters (e.g. sender bitrate, frame rate) and content type is provided. An understanding of the relationships and interactions between them and their impact on video quality is important as it provides a basis for the development of non-intrusive video quality prediction models.(2) A new content classification method was proposed based on statistical tools as content type was found to be the most important parameter. (3) Efficient regression-based and artificial neural network-based learning models were developed for video quality prediction over WLAN and UMTS access networks. The models are light weight (can be implemented in real time monitoring), provide a measure for user perceived quality, without time consuming subjective tests. The models have potential applications in several other areas, including QoS control and optimization in network planning and content provisioning for network/service providers.(4) The applications of the proposed regression-based models were investigated in (i) optimization of content provisioning and network resource utilization and (ii) A new fuzzy sender bitrate adaptation scheme was presented at the sender side over WLAN and UMTS access networks. (5) Finally, Internet-based subjective tests that captured distortions caused by the encoder and the wireless access network for different types of contents were designed. The database of subjective results has been made available to research community as there is a lack of subjective video quality assessment databases.

Le cadre de cette thèse est l'étude des systèmes contrôlés en réseau sans fil (SCRSF) qui utilise la technologie IEEE 802.15.4. Le premier objectif est d'étudier la pertinence de l'utilisation du réseau de type IEEE 802.15.4 pour les SCRSF puis de proposer et d'évaluer des mécanismes pour garantir la Qualité de Service (QdS) offerte par le réseau au système contrôlé. Nous analysons l'utilisation des slots temporels réservés (GTS) dans le cadre des SCRSF et les contraintes qui en découlent. De plus, nous proposons des mécanismes de gestion de la QdS avec priorité aussi bien pour le mode avec balise que pour le mode sans balise du protocole IEEE 802.15.4. Ces propositions ont été validées par des simulations et une partie de manière analytique. Notre deuxième objectif est de concevoir, d'une manière conjointe, les SCRSF pour pouvoir régler en ligne la QdS offerte par le réseau en fonction de la Qualité de Contrôle (QdC) du système contrôlé. Nous proposons des protocoles d'adaptation en ligne de la QdS du réseau qui prennent en compte la QdC du système contrôlé. Ces protocoles ont été validés par simulations et une implémentation réelle de chacun d'eux est proposée / In this thesis, we study wireless networked control systems (WNCS) which use the IEEE 802.15.4 technology. The first objective is to study the pertinence of the use of the IEEE 802.15.4 for the WNCS, then to propose and evaluate QoS management mechanisms which guarantee the Quality of Service (QoS) offered by network to the controlled system. We analyse the use of the guaranteed temporel slots (GTS) for WNCS and in which conditions it is possible. We propose QoS management mechanisms with priority for both the beacon enabled mode and the non-beacon enabled mode of the IEEE 802.15.4 protocol. These proposals are validated through simulations and partially with analytical approach. The second objective is to design the WNCS so that the QoS offered by the network is adated online depending on the Quality of Control (QoC) on the controlled system. We propose QoS online adaptation protocols which take as parameter the QoC of the system. These protocols are validated through simulations and a realistic implementation of them is proposed

The transport control protocol (TCP) is one of the most heavily used protocols on the Internet, offering a reliable, connection oriented transport service. However, the quality of service (QoS) provided by the TCP protocol deteriorates when it is used over satellite IP networks. With the increased usage of Internet applications by the military in remote geographical regions, there is an increased need to address some of the shortcomings of TCP performance in satellite IP networks. In this research we describe our efforts at designing and testing a performance enhancing proxy (PEP) that can be used improve the QoS provided by the TCP service in large latency networks. We also show how one can use such a proxy to create a new transport service similar to the Available Bit Rate (ABR) service provided by ATM networks without needing ATM infrastructure, this new service offers a connection oriented, reliable, best effort transport service with minimal queuing delay, jitter and throughput variation.

In this thesis, focusing on the quality of service (QoS) improvement using per-packet power control algorithm in Ad Hoc Cognitive Packet Networks (AHCPN). A power control mechanism creates as a network-assisted function of ad hoc cognitive packet-based routing and aims at reducing both energy consumption in nodes and QoS requirements. The suggested models facilitate transmission power adjustments while also taking into account the effects on network performance. The thesis concentrate on three main contributions. Firstly, a power control algorithm, namely the adaptive Distributed Power management algorithm (DISPOW) was adopted. Performance of DISPOW was compared to existing mechanisms and the results showed 27, 13, 9, and 40 percent improvements in terms of Delay, Throughput, Packet Loss, and Energy Consumption respectively. Secondly, the DISPOW algorithm was enhanced, namely a Link Expiration Time Aware Distributed Power management algorithm (LETPOW). This approach periodically checks connectivity, transmission power, interference level, routing overhead and Node Mobility in AHCPN. The results show that LETPOW algorithm improves the performance of system. Results show further improvement from DISPOW by 30,25,30,42 percent in terms of delay, packet loss ratio , path lengths and energy consumption respectively. Finally,Hybrid Power Control Algorithm (HLPCA) has presented is a combination of Link Expiration Time Aware Distributed Power management algorithm (LETPOW) and Load Power Control Algorithm (LOADPOW); deal with cross-layer power control applied for transmitting information across the various intermediate layers. LOADPOW emphasis on the concept of transmission Power, Received Signal Strength Indication (RSSI), and the suitable distance between the receiver and the sender. The proposed algorithm outperforms DISPOW and LETPOW by 31,15,35,34,44 percent in terms of Delay, Throughput, Packet Loss,path length and Energy Consumption respectively. From this work, it can be concluded that optimized power control algorithm applied to Ad-hoc cognitive packet network results in significant improvement in terms of energy consumption and QoS.

Submitted by Erika Demachki (erikademachki@gmail.com) on 2015-02-05T19:42:01Z No. of bitstreams: 2 Dissertação - Diógenes Antonio Marques José - 2014.pdf: 9980108 bytes, checksum: eea0a39505448f7845d92bdbb4716ff2 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Approved for entry into archive by Erika Demachki (erikademachki@gmail.com) on 2015-02-05T19:42:44Z (GMT) No. of bitstreams: 2 Dissertação - Diógenes Antonio Marques José - 2014.pdf: 9980108 bytes, checksum: eea0a39505448f7845d92bdbb4716ff2 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Made available in DSpace on 2015-02-05T19:42:45Z (GMT). No. of bitstreams: 2 Dissertação - Diógenes Antonio Marques José - 2014.pdf: 9980108 bytes, checksum: eea0a39505448f7845d92bdbb4716ff2 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Previous issue date: 2014-06-05 / This work contributes with an extension to the Optimized Link State Routing protocol (OLSR) called Fuzzy Cost OLSR (OLSR-FC). In order to prevent selfish nodes as well as to improve the traffic flow over Mobile Ad-hoc Networks (MANETs), the routing metrics implemented in OLSR-FC make use of a Fuzzy Inference System (FIS) composed of 8 inference rules. Aiming at the choose of paths with low packet loss, better energy capacity and high connectivity, OLSR-FC implements a procedure of election of routes that takes into account the following parameters: Packet Loss Index (PLI), Residual Energy (RE) and Connectivity Index (CI). The OLSR-FC was evaluated by simulation through the NS- 2, in which two scenarios were implemented: a static one with 10 nodes (in testing phase), and a mobile one with up to 50 nodes. In the former scenario, a comparison was made between OLSR-FC and the original OLSR protocol which results showed that OLSR-FC overcomes OLSR in terms of throughput the packet loss. In the latter scenario, besides the original OLSR protocol, OLSR-FC was also faced up with the OLSR-ETX, OLSR-ML and OLSR-MD extensions in terms of the following performance metrics: throughput, energy consumption, packet loss rate, overhead, delay end-to-end, jitter, and packet delivery rate. In this context, results pointed that OLSR-FC achieved better performance in scenarios with a maximum of 10% of selfish nodes in comparison with every OSLR extension and the OLSR. Besides, by evaluating the main network performance metrics, such as throughput and delivery packet rate, OLSR-FC achieved eleven favorable cases against five cases in comparison with OLSR protocol. / O presente trabalho propõe uma extensão ao protocolo Optimized Link State Routing (OLSR) denominada OLSR Fuzzy Cost (OLSR-FC). A métrica de roteamento utilizada pelo OLSR-FC faz uso de um Sistema de Inferência Fuzzy (SIF), composto por 8 regras de inferência, que tem como objetivos evitar nós egoístas e melhorar o fluxo do tráfego nas redes móveis ad-hoc (MANETs). O critério de escolha de rotas leva em consideração os seguintes parâmetros: Índice de Perda de Pacotes (IPP), Energia Residual (ER) e Índice de Conectividade (IC), o propósito disso, consiste em escolher caminhos que possuam baixa perda de pacotes, melhor capacidade energética e alta conectividade. A proposta foi avaliada por simulação utilizando o simulador de redes NS-2. Foram considerados na avaliação dois cenários, um estático com 10 nós (utilizado na fase de testes) e um móvel com até 50 nós. No cenário estático o OLSR-FC foi comparado ao OLSR e os resultados mostraram que a proposta obtém vantagens com relação às métricas de desempenho vazão e perda de pacotes. No cenário móvel o OLSR-FC, além do OLSR, foi comparado às extensões OLSR-ETX, OLSR-ML e OLSR-MD, com relação as seguintes métricas de desempenho: vazão, consumo de energia, perda de pacotes, overhead, atraso fim-a-fim, jitter e taxa de entrega de pacotes. Nesse contexto, os resultados obtidos mostram que em ambientes com até 10% de nós egoístas o OLSR-FC obtém melhor desempenho que as extensões testadas e com relação ao OLSR, na mesma situação, avaliando as principais métricas de desempenho de redes, como vazão e taxa de entrega de pacotes, o OLSR-FC obteve onze casos favoráveis contra apenas cinco casos do OLSR.

Submitted by camilla martins (camillasmmartins@gmail.com) on 2016-12-13T13:25:05Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Tese_CarimboCanalRetorno.pdf: 3013619 bytes, checksum: 8b15e2a109bce37be96c990d534d3870 (MD5) / Rejected by Edisangela Bastos (edisangela@ufpa.br), reason: on 2016-12-15T12:07:14Z (GMT) / Submitted by camilla martins (camillasmmartins@gmail.com) on 2016-12-15T13:42:20Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Tese_CarimboCanalRetorno.pdf: 3013619 bytes, checksum: 8b15e2a109bce37be96c990d534d3870 (MD5) / Rejected by Edisangela Bastos (edisangela@ufpa.br), reason: on 2016-12-15T14:01:08Z (GMT) / Submitted by camilla martins (camillasmmartins@gmail.com) on 2016-12-15T14:26:04Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Tese_CarimboCanalRetorno.pdf: 3013619 bytes, checksum: 8b15e2a109bce37be96c990d534d3870 (MD5) / Approved for entry into archive by Edisangela Bastos (edisangela@ufpa.br) on 2016-12-19T14:06:09Z (GMT) No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Tese_CarimboCanalRetorno.pdf: 3013619 bytes, checksum: 8b15e2a109bce37be96c990d534d3870 (MD5) / Made available in DSpace on 2016-12-19T14:06:09Z (GMT). No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Tese_CarimboCanalRetorno.pdf: 3013619 bytes, checksum: 8b15e2a109bce37be96c990d534d3870 (MD5) Previous issue date: 2006-12-15 / Esta tese de doutorado propõe um framework para controle e gerenciamento da rede de retorno da Televisão Digital a partir de quatro componentes: o Provedor de Televisão Digital, o Terminal Interativo, o Provedor de Canal de Retorno e o Provedor de Conteúdo. Juntos, interagem com base em um mecanismo de sinalização que propaga restrições de QoS associadas à programação interativa corrente pelo sistema de carrossel de dados. O conjunto também provê um sistema de admissão que investiga, periodicamente, a condição do canal de retorno autorizando ou restringindo acesso em função das necessidades das aplicações interativas. Também é flexível a ponto de viabilizar aplicações tanto em larga escala, quanto localmente uma vez que permite mobilidade ao gestor do serviço. Uma aplicação de e-gov/ehealth foi desenvolvida em Java XleTView para consolidar o desempenho do sistema. Realizou-se a avaliação utilizando técnicas de aferição e modelagem de acordo com a escala. Na última milha, avaliaram-se tecnologias de baixo custo e que não requerem tarifação periódica como é o caso de redes locais sem fio. Em escala maior, o sistema foi modelado e seu desempenho avaliado em diferentes cenários com auxílio do simulador de redes Network Simulator. / This doctoral thesis proposes a framework has controling and management the return network in Digital Television from four components: Digital Television Provider, Interactivity Terminal, Return Channel Provider, and Content Provider. Together, they interact based on a signaling mechanism that propagates QoS restrictions associated to the current interactive programming through the data carrousel system. The set of them also provide an admission control that investigates, periodically, the return channel status allowing or restricting the access according to the necessities of the interactive applications. Moreover, the system is so flexible that admit applications as in large scale as in small scale (local), thanks to the mobility of the service provider. An e-gov/e-health application was development in Xlet Java to evaluate the performance of the system. The performance analysis was proceeded using measurement and modeling techniques according to the system scale. In the last mille, some technologies of low cost and without periodically billing were tested such as wireless local area network. In large scale, the system was evaluated in different scenarios by means of simulation using the Network Simulator.

Dans cette thèse, nous étudions plusieurs méthodes pour la mise en place de la QoS<br />(Quality of Service) dans un réseau IP. Avant d'entamer nos travaux de recherche, nous<br />dévoilons tout d'abord les différents mécanismes de QoS que nous étudierons dans la thèse.<br />MPLS (Multi Protocol Label Switching), DiffServ (Differentiated Services) et les algorithmes<br />d'ordonnancement formeront la base du réseau coeur que nous utiliserons. En accord avec<br />plusieurs travaux, nous définissons trois classes de service à différencier dans le réseau. La<br />première classe comporte les flux voix. Elle nécessite un délai faible et une gigue réduite. Les<br />flux de "données critiques", qui requièrent un taux de perte faible et un délai borné, forme la<br />deuxième classe. La troisième classe, qui regroupe les applications telles que le transfert de<br />fichiers ou l'échange de courriers électroniques, n'exige aucune condition particulière du<br />réseau.<br />Dans un premier temps, nous modélisons les flux sortant/entrant d'un réseau mobile<br />sans fil. D'abord, nous modélisons les lois d'inter arrivée des paquets d'un flux agrégé au<br />niveau de la couche MAC (Medium Access Control) entrant au réseau UTRAN (UMTS<br />Terrestrial Radio Access Network), le réseau d'accès à l'UMTS (Universal Mobile<br />Telecommunication Service). Le protocole CDMA (Code Division Multiple Access) utilisé<br />dans ce réseau propose un accès différent en fonction de la QoS demandée. Ensuite, nous<br />déterminons la loi caractérisant l'inter arrivée des paquets sortant d'un réseau WiFi (Wireless<br />Fidelity) utilisant la couche MAC 802.11 de base. Pour ces deux réseaux, nous proposons<br />différents modèles d'agrégation de flux voix, Web, transfert de fichiers ou un multiplexage de<br />ces différentes classes. Nous mesurons l'adéquation de deux modèles de loi de distributions<br />aux traces créées. La première distribution est le processus MMPP (Markov Modulated<br />Poisson Process) qui représente un modèle Markovien. Nous expérimentons deux valeurs du<br />nombre d'états: 2 et 4. La seconde loi que nous considérons est la loi Gaussienne. Nos<br />résultats montrent que d'une part, le type des flux agrégés et d'autre part, le réseau utilisé,<br />influent tous les deux sur le modèle produit.<br />Dans un second temps, nous développons les équations qui déterminent les<br />probabilités stationnaires d'une file d'attente implémentant l'ordonnanceur GPS (Generalised<br />Processor Sharing) avec trois classes de service. En utilisant le mécanisme DiffServ pour<br />différencier les flux, nous mesurons la QoS à la sortie d'une file unique utilisant WRR<br />(Weighted Round Robin), un des algorithmes qui approximent GPS. Nous traçons alors les<br />différentes courbes de délai et de taux de perte observés à la sortie de cette file en fonction de<br />la pondération et de la charge créée par chacune des classes. Nous appliquons les différentes<br />conclusions du choix des paramètres que nous tirons d'un seul serveur à un réseau entier. De<br />plus, nous ajoutons l'ingénierie de trafic de MPLS pour quantifier le gain mesuré par chaque<br />politique. A partir de ce travail, nous avons pu généraliser nos constatations qui deviennent<br />valable aussi bien sur une file que dans un réseau.<br />Dans un troisième temps, nous développons une méthode d'adaptation dynamique du<br />routage. Nous la proposons afin de palier aux variations de la distribution du délai sur les<br />liens qui forment le chemin de bout en bout. Ce mécanisme se base sur les techniques de<br />tomographie des réseaux afin d'estimer la distribution du délai sur les différents tronçons des<br />chemins observés. Si le délai moyen sur la route utilisée reste supérieur d'un seuil ε pendant<br />un temps τ au délai moyen d'un autre chemin, le mécanisme déclenche alors la procédure de<br />modification du chemin emprunté. C'est l'utilisation du protocole MPLS associé à ce<br />mécanisme qui permet une modification souple et rapide des itinéraires.