DARED: a double-feedback AQM technique for routers supporting real-time multimedia traffic in a best-effort scenario

Abstract

The transmission of real-time multimedia traffic on the Internet is becoming a challenging problem for current research in telecommunications networks. The idea at the basis of this paper consists of using an active queue management (AQM) technique in the router buffers and a rate-control algorithm at the source site which collaborate with each other, in order to meet source QoS requirements, depending on the available resources, simultaneously achieving pro-active control of network congestion. The proposed AQM technique, called dynamically adaptive random early detection (DARED), works on a network architecture characterized by a double feedback: from network to sources, and from sources to network. The paper shows via simulation, how the architecture simultaneously meets different target QoS levels, specified by sources belonging to different classes.