Open Access

An Information on Demand teleservice that was developed at the German National Research Center for Information Technology (GMD) provides remote access to multimedia information consisting of audio, video, and text [jonas et al. 94]. It uses a bidirectional narrowband message link between the end user and the service provider, and a unidirectional broadband data link from the service provider to the end user. Since the IoD teleservice is used across a satellite connection (among others), it turned out to be necessary to implement an access protocol that is optimized for the access of real-time multimedia data across a long-delay high-bandwidth link, a long fat pipe [jacobsen et al. 92]. This paper introduces the MediaService Protocol (MSP) and describes a prototype implementation (version 0.6).

stract Expected real-time applications like video telephony, video conferences and document retrieval can either take advantage of human’s ability to cope with missing pictures or bad audio quality, or of — time consuming, therefore inconvenient — retransmission.

The term "virtual studio" refers to real-time 3D graphics systems used to render a virtual set in sync with live camera motion. As the camera pans and zooms, the virtual set is redrawn from the correct perspective. Using blue room techniques, actors in front of the real camera are then “placed in” the virtual set. Current virtual studio systems are centralized – the blue room, cameras, renderers etc. are located at a single site. However distributed configurations offer significant economies such as the sharing of expensive rendering equipment among many sites. This paper describes early expe- riences of the DVP1 project in the realization of a distributed virtual studio. In particular we de- scribe the first video production using a distributed virtual studio over ATM and make observations concerning network QOS requirements.

Multipoint data-communications is among the hot topics of communication research and development. A lot of studies and ideas have been presented, the vast majority focusing on a homogenous environment in terms of physical network, communication protocol stacks, coding schemes and/or service qualities. First straight-forward implementations –Steve Deering‘s IP multipoint on the MBone being the most popular one– already give an idea of the capabilities of a multipoint environment.

Fast and seamless handoff procedures for IP-based mobile communication networks have recently been a major field of research. Various methods have been proposed that focus on micro-mobility suppport, i.e., on mobility management within a single administrative domain.

For efficient deployment of new reliable multicast applications in heterogeneous mobile Internet environments, appropriate retransmission strategies are proposed. The focus is the minimization of the protocol overhead for reliable transport taking into account behaviour in mobile networks (oss and handover behaviour) and application requirements (such as carousel file transfer, one-to-many download and media streaming combined with recording). The proposed techniques are designed as building blocks for localized multicast error control supported by access routers. Considering IETF RMT standardization work, the discussed retransmission approaches can be used for flexible configuration of tree-based reliable multicast protocols in converged wired and wireless Internet environment. The implementation developed for the European project DAIDALOS [1] is based on Linux IPv6 environment. Simulations in ns2 focusing on the benefits of the proposed multicast retransmission schemes for particular application scenarios are presented.

Bei der Datenübertragung im Internet ist es wünschenswert, daß unterschiedliche Datenpackete unterschiedlich behandelt werden können, damit für die zeitkritischen Anwendungen wie Internet-Telefonie die größere Priorität gewährleisten könnte. Erst dann kann ein Netzbetreiber seinen Kunden verschiedene Dienstqualitäten anbieten. Zwei grundsätzliche Mechanismen existieren, um dem Netzwerk solche Anforderungen mitzuteilen: explizit, indem vor der eigentlichen Datenübertragung Kotrollnachrichten ausgetauscht werden; implizit, indem jedes Datenpacket eine Kennzeichnung erhält. Der zweite, Differentiated Services(DS), Ansatz ist Thema dieses Referates. Bei DS wird jedem Packet den sogenannten DS-Codepoint (DSCP) im IP-Header zugewiesen, und dementsprechend erfährt ein Datenpaket eine bestimmte Behandlung durch einen DS-fähigen Router. Drei unterschiedliche Dienstklassen wurde bisher spezifiziert: Best-Effort entspricht dem derzeit im Internet verwendeten Mechanismus; Premium-Service entspricht einer virtuellen Mietleitung und soll eine rasche Weiterleitung von Paketen gewährleisten; Datenpakete mit Assured Forwarding Service (AFS) sollen besser behandelt werden als Best-Effort. Bei AFS wurden insgesamt drei Verlustklassen und vier Weiterleitungsklassen definiert. Es wird auch verschiedene Typ von DS-Router vorgestellt und diskutiert. Der praktische Einsatz erfolgte bisher immer in Laborumgebung (Uni Bern, Uni Karlsruhe, EPFL, Forschungszentrum NEC Research), größere Feldversuche stehen noch aus.

Dynamic configuration and adaptation of resources for QoS-aware applications in heterogeneous access network environment (UMTS, WIMAX, WLAN DVB-T, DVB-H) using automated tools is a challenge today. The focus of this paper is a toolkit for intelligent management of resource allocation in heterogeneous network infrastructures based on policies of different actors (network operator, service providers and users). Policy based management of resources for QoS-aware applications (Video-on-Demand, Mobile TV) dependent on network capabilities, context learning and preferences of the policy actors is proposed, which enhances the current state-of-the-art and IETF standardisation. The policy management toolkit includes components for policy specification, adaptation and enforcement, which are interacting using policy repository. The design allows the automated resource adaptation for QoS based applications based on context information and hierarchical dependencies of policy actors. A learning component is integrated in order to discover the context considering measurement and monitoring data. The policy management tookit is discussed, emphasising on ontology driven policy repository design, context learning and flexible scenario-oriented management interfaces for policy specifications.

Zukünftige Netze sollen in der Lage sein, eine Vielzahl verschiedener Dienste zu unterstützen. Dabei wird angenommen, dass sie oft keine statische Netzwerkstruktur haben und sich daher selbsttätig konfigurieren und automatisch an wechselnde Anforderungen sowie Netzwerksituationen und -änderungen anpassen sollen. Es ist wünschenswert, dass die benötigte Dienstqualität (QoS), im Besonderen für zeitkritische Anwendungen wie VoIP, automatisch gewährleistet werden kann. Das Netz soll zukünftigen Anforderungen gerecht werden, auch wenn diese zum Zeitpunkt der Erstellung noch nicht relevant oder bekannt sind. In diesem Dokument wird ein möglicher Ansatz des noch jungen Forschungsprojekts Self-NET für die Erfüllung der genannten Anforderungen skizziert und zur Diskussion gestellt.

The MoMoSat service will enable mobile end-users to view, manage, annotate, and communicate mapbased information in the field. The handled information exists of a huge volume of raster (satellite or aerial images) and vector data (i.e. street networks, cadastral maps or points of interest), as well as text-specific geo-referenced textual notes (the so-called 'GeoNotes') and real-time voice.