You are here
Gunnar Karlsson. Martin Karsten. Helmut Kneer. Jerzy Kubasik. Wang Kun. Pascal Kurtansky. Nicolas Liebau. Marco Liebsch. Yuanan Liu. Rui J. Hongguang Ma.
A Pull-Based Approach for a VoD Service in P2P Networks
Victor Marques. Andreas Mauthe. Lee McKnight. Telemaco Melia. Eric Melin. Piet Van Mieghem. Jan Mischke. Edmundo Monteiro. Cristian Morariu.
Jose Ignacio Moreno. Simon Oechsner.
- Pricing and QoS - Research Collection.
- The Physiology of Mosquitoes!
- Budget Chapter 2!
Huw Oliver. Maximilian Ott. Piotr Pacyna. Ioanna Papafili. George Pavlou. Dalibor Peric. Charles E. Thomas Plagemann. Bernhard Plattner. Roman Pletka. Mihai Popa. Aiko Pras. Konstantin Pussep. Bruno Quoitin. Peter Racz. Peter Reichl.
- The Scandalous Summer of Sissy LeBlanc!
- Ubuy Lebanon Online Shopping For pano in Affordable Prices.!
- Additive combinatorics.
- Uitgebreid zoeken.
- Agricultural Policies in Emerging Economies 2009: Monitoring and Evaluation.
- Prof. Dr. habil. Michael Menth.
Helmut Reiser. Reinhard Riedl. James Roberts. Gabi Dreo Rodosek. Christian Schaefer. Gregor Schaffrath. Jens B. Schmitt Jens Schmitt. Ralf Schmitz. Michael Shann. Davinder Singh. Vasilios A. Michael Smirnov. Mikhail I. David Songhurst. Sergios Soursos. Rolf Stadler.
Dirk Staehle. George D. Ioannis Stavrakakis. Ralf Steinmetz. Henrik Stormer. Tatsuya Suda. Ahmed N. Denis Trcek. Panos Trimintzios. Bruno Tuffin. Andrei Vancea. Martin Waldburger.mail.cmnv.org/map36.php
Quality of Future Internet Services - COST Action Final Report | Michael Smirnov | Springer
Traffic theory consists in describing through mathematical models the relation between demand, capacity and performance. Section 2 argues that such theory should be increasingly applied to the design of traffic management for the Internet. Distinct modeling approaches are discussed for so-called elastic traffic and streaming traffic, represented in both cases at flow level. It is demonstrated that, under certain assumptions about how the bandwidth is shared, performance is largely insensitive to detailed traffic characteristics, allowing the definition of simple traffic engineering principles.
It is argued finally that existing QoS architectures do not take sufficient account of the insights provided by traffic theory and that an alternative flow-aware networking architecture is necessary. Karlsson et al. The simplest method for applications to obtain the quality of service they need is to over-dimension the network. Despite its simplicity, this solution has several drawbacks: Its financial cost, the implicit promotion of waste, the diversity of the QoS needs of applications, and the different characteristics of the protocols. The model appears in a context in which one recognizes that overdimensioning communication systems is not a good solution.
It is based on the belief that it is possible to obtain good results by applying simple strategies that do not collide with classical IP operational principles and technologies but, instead, are able to make the most out of their virtues. The architecture in the previous section relates to the treatment of traffic within the network. A vital function of traffic management is to restrict the inflow of traffic to the network by means of admission control, as pointed out in Section 2.
There has been an interest during the last five year to simplify measurement-based admission controls by administering the control from the edges of the network and not in each node of the network in the traditional manner.
Section 4 surveys recent proposal of endpoint admission control with an emphasis on probe-based admission control. This particular scheme is based on the transmission of test packets to estimate the quality of the path for a communication session. The session is established only if a specified quality level is reached during the test phase. The service to a user from a public network operator will be regulated in a servicelevel agreement.
It is then of primary importance to provide mechanisms for monitoring the performance related to an SLA. This monitoring is of interest to both users and network service providers. Section 5 proposes a monitoring framework that represents the basis for the collection and distribution of performance data. Monitoring information should be provided by the network to the user application by collecting and appropriately combining performance measures into an SLI that is linked to the SLA.
Monitoring might hence be a meta-service, i. Traffic management at small time scales of the order of packet transmission time consists of buffer management and packet scheduling. Section 6 examines the case of a single node employing the Generalized Processor Sharing GPS scheduling discipline for the provision of deterministic delay guarantees to leaky bucket constrained sources. The problem of call admission control is investigated and an algorithm, which is capable of always producing the optimal weight assignment is presented. In addition, a modification of GPS which is based on the idea of decomposing original sessions into traffic components that can be more efficiently handled by GPS is introduced and a call admission control algorithm for the proposed discipline is discussed.
Effective design of traffic management relies on a sound understanding of the nature of the traffic.
It is also necessary to be able to succinctly describe specific types of traffic using appropriate mathematical models. Section 7 discusses such models for variable rate video traffic using MPEG coding. A two-step approach is proposed consisting of first accounting for the correlation structure at the group of pictures GoP Traffic Management 13 level and then applying known intra-GoP correlation properties to adjust the size distributions of successive I, P and B frames.