Retour au site du LAAS-CNRS

Abstract

In the context of a reconfigurable transport protocol framework, we propose a QoS-aware Transport Protocol (QSTP), specifically designed to operate over QoS- enabled networks with bandwidth guarantee. QSTP combines QoS-aware TFRC congestion control mechanism, which takes into account the network-level band- width reservations, with a Selective ACKnowledgment (SACK) mechanism in order to provide a QoS-aware transport service that fill the gap between QoS enabled network services and QoS constraint applications. We have developed a prototype of this protocol in the user-space and conducted a large range of measurements to evaluate this proposal under various network conditions. Our results show that QSTP allows applications to reach their negotiated QoS over bandwidth guaranteed networks, such as DiffServ/AF network, where TCP fails. This protocol appears to be the first reliable protocol especially designed for QoS network architectures with bandwidth guarantee.

Mots-Clés / Keywords

Abstract

TCP-friendly rate control (TFRC) is a congestion control technique that trade-offs responsiveness to the network conditions for a smoother throughput variation. We take advantage of this trade-off by calculating the rate gap between the theoretical TCP throughput and the smoothed TFRC throughput. Any rate gain from this rate gap is then opportunistically used for video coding. We define a frame complexity measure to determine the additional rate to be used from the rate gap and then perform a rate negotiation to determine the target rate for the encoder and the final sending rate. Results show that although this method has a more aggressive sending rate compared to TFRC, it is still TCP-friendly, does not contribute too much to network congestion and achieves a reasonable video quality gain over the conventional method.

Abstract

The behaviour of the TCP AIMD algorithm is known to cause queue length oscillations when congestion occurs at a router output link. Indeed, due to these queueing variations, end-to-end applications experience large delay jitter. Many studies have proposed efficient Active Queue Management(AQM) mechanisms in order to reduce queue oscillations and stabilize the queue length. These AQM are mostly improvements of the Random Early Detection (RED) model. Unfortunately, these enhancements do not react in a similar manner for various network conditions and are strongly sensitive to their initial setting parameters. Although this paper proposes a solution to overcome the difficulties of setting these parameters by using a Kohonen neural network model, another goal of this study is to investigate whether cognitive intelligence could be placed in the core network to solve such stability problem. In our context, we use results from the neural network area to demonstrate that our proposal, named Kohonen-RED (KRED), enables a stable queue length without complex parameters setting and passive measurements.

Abstract

This paper presents a novel elearning platform called IREEL. IREEL is a virtual laboratory allowing students to drive experiments with real Internet applications and endtoend protocols in the context of networking courses. This platform consists in a remote network emulator offering a set of predefined applications and protocol mechanisms. Experimenters configure and control the emulation and the endsystems behavior in order to perform tests, measurements, and observations on protocols or applications operating under controlled specific networking conditions. A set of endtoend mechanisms, mainly focusing on transport and applicationlevel protocols, are currently available. IREEL is scalable and easy to use thanks to an ergonomic web interface.