Stefan Dresler

Multipath FEC scheme for the ATM adaptation layer AAL5

One approach to facilitate statistical multiplexing of bursty sources in ATM networks is dispersion of the traffic over independent paths. It has been shown within the literature that by dispersion of traffic over disjoint paths it is possible to increase the Quality of Service (QoS) (see e.g., (Maxemchuck, 1993), (Lee, 1993)). In multipath schemes without additional error control, the resulting QoS depends on the worst-case path. By using a multipath scheme with FEC (Forward Error Correction), it is possible to compensate cell losses. Using error correcting codes like Reed-Solomon-Erasure Codes (RSE Codes, (McAuley, 1990)), these schemes add redundancy and disperse the traffic using several transmission paths (Ding, 1995).

A Native ATM API Suited for Multimedia Communication

ATM, the asynchronous transfer mode, was developed to meet two main goals. Firstly, it provides a flexible transmission technology. Secondly, it offers a set of new services to the user [Part94]. In order to take advantage of ATMs strengths, one might want to use its services directly. To be able to do so, in particular to be able to integrate access to ATM services into own applications, one is in need of a powerful ATM application programming interface (API). In this paper, we first present the design and implementation of such an ATM API for Digital Alpha workstations running Digital UNIX. Moreover, we describe how the basic ATM service is enhanced by setting an implementation of SandiaXTP (Xpress Transport Protocol) on top of it. Finally, we give an overview over a monitor for this ATM environment designed to compute quality of service (QoS) parameters from protocol information, detect QoS violations by comparing them to desired values, and display the results graphically.

High performance group communication services in ATM networks

Advanced applications, such as distributed multimedia applications, require efficient communication subsystems providing a variety of services. Existing communication systems face increasing difficulties in fulfilling these requirements. In particular, the efficient provision of reliable group communication services in ATM-Networks remains a major unresolved issue. This paper presents a novel framework for support of multipoint communication in ATM networks. Two adaptation layer protocols are presented that provide reliable multicast services. The first one, called RLMCP (Reliable Lightweight Multicast Protocol), is a simple and efficient adaptation layer protocol for the Service Specific Convergence Sublayer of AAL5. It uses a frame-based ARQ scheme and is suitable for virtual connections with low cell loss rates. The second one, called RMC-AAL (Reliable Multicast ATM Adaptation Layer), features cell-based ARQ and cell-based FEC. A new network element, called the Group Communication Server (GCS), is presented for implementing the adaptation layer protocols in network nodes. It allows for hierarchical multicast error control and support of heterogeneous scenarios. The framework permits to select the combination of error control mechanisms most suitable for the requirements of a specific communication scenario. The functionality of end systems and group communication servers is described, and a basic implementation architecture is presented. Based on this architecture, approximations for the processing delays are presented when the different error control schemes are applied. Finally, the influence of the different error control schemes onto the selection of an appropriate memory management strategy is investigated.

ATM multipeer communication using a two-layer architecture

Applications deploying distributed database updates, distributed simulations, shared whiteboards, and forthcoming distributed interactive multi-user games have communication needs not satisfied by most available protocols or communication subsystems. These applications are different from classical applications in that they have potentially many senders and receivers. Although there has been a lot of research on unicast and multicast protocols, only little work has been done in the area of multipeer communication, i.e. communication realizing an M:N communication service. Even fewer results are available for such a service on connection oriented technologies like ATM (asynchronous transfer mode). This paper describes the current situation, proposes a novel two-layer architecture especially suitable for ADSL and cable networks, which bridge the gap between LANs and the WAN, gives a first analysis, and presents open issues.

Sub-Layer-Hopping: Optimierte Staukontrolle für IP-Multicast

Die Multicast-Kommunikation im Internet ist derzeit Gegenstand zahlreicher Forschungsaktivitaten. Konzentrierte sich in jungster Vergangenheit die uberwiegende Zahl der Arbeiten auf die Bereitstellung skalierbarer Mechanismen zur Fehlerbehebung, so ruckt derzeit die Entwicklung geeigneter Algorithmen zur Staukontrolle in den Mittelpunkt des Interesses. Eine besondere Herausforderung besteht dabei in der Unterstutzung heterogener Kommunikationsgruppen, innerhalb derer die Empfanger unterschiedliche Stausituationen erfahren. Ein vom Konzept her vielversprechender Ansatz ist der sogenannte Layered-Multicast, bei dem der eigentliche Datenstrom in mehrere Unterstrome (Layers) aufgeteilt wird. Bisher scheiterte die praktische Umsetzung dieses vielversprechenden Ansatzes jedoch an der Tragheit heutiger Routingprotokolle. Diese sind nicht in der Lage, auf die notwendigen Wechsel in der Gruppenmitgliedschaft in einer akzeptablen Zeit zu reagieren. Vor diesem Hintergrund wird im vorliegenden Artikel ein Verfahren entwickelt und bewertet, welches trotz der genannten Beschrankungen erstmals die praktische Umsetzung des Layered Multicast in heutigen Netzen ermoglicht.

Native ATM interfaces in C and Java: implementation and experiences

ATM, the asynchronous transfer mode, represents a flexible transmission technology, which offers a set of new services to the user. In order to take advantage of ATMs strength and to be able to integrate access to ATM services into applications, a powerful ATM application programming interface (API) is needed. At our institute, we have implemented three different variants of such an API for accessing native ATM functions. In this paper we first present the design and implementation of an ATM API in C for Digital Alpha workstations running Digital UNIX. We then describe a Java ATM API offering the socket programming environment Java programmers are used to, and finally a set of Java functions accessing the Windows NT Winsock 2 interface. We contrast the different philosophies and architectural approaches with each other and give throughput numbers.