In this paper we describe the design of a QoS driven communications stack in a microkernel operating system environment. The paper focuses on resource management aspects of the design and in particular we deal with CPU scheduling, network resource management and memory management issues. We describe an API at which QoS parameters are supplied by users and present an architecture able to guarantee these QoS requirements with varying degrees of commitment. The architecture includes modules to translate user level QoS parameters into representations usable by the scheduling, network and memory management subsystem. It also incorporates admission tests which determine whether or not a new connection can be accepted given a particular QoS requirement and resource availability. In addition to predictable resource allocation the architecture also incorporates dynamic QoS management which ensures that ongoing commitments continue to be met in the face of variable loading.

Currently, popular operating systems are unable to support the end-to-end real-time requirements of distributed continuous media. Furthermore, the integration of continuous media communications software into such systems poses significant challenges. This paper describes a design for distributed multimedia support in the Chorus micro-kernel operating system environment which provides the necessary soft real-time support while simultaneously running conventional applications. Our approach is to extend existing Chorus abstractions to include QoS configurability, connection oriented communications and real-time threads. The design uses the following key concepts: the notion of a flow to represent QoS controlled communication between two application threads, a close integration of communications and thread scheduling and the use of a split level scheduling architecture with kernel and user level threads. The paper shows how our design qualitatively improves performance over existing micro-kernel facilities by reducing the number of protection domain crossings and context switches incurred.

To support multimedia applications in mobile environments, it will be necessary for applications to be aware of the underlying network conditions and also to be able to adapt their behaviour and that of the underlying platform. This paper focuses on the role of middleware in supporting such adaptation. In particular, we investigate the role of open implementation and reflection in the design of middleware platforms such as CORBA. The paper initially extends CORBA with the concept of explicit binding, where path of communication between objects is represented as first class objects. We then introduce the concept of open bindings which support inspection and adaptation of the path of communications. An implementation of open bindings for adaptive continuous-media interaction is described using the example of adaptive video-on-demand for mobile environments.

There is currently considerable interest in developing multimedia applications in open distributed systems. However, it is now becoming clear that existing architectures for open distributed systems do not support the particular requirements of continuous media types such as digital audio and video. This is particularly the case in the important areas of quality of service support and real-time synchronization. This paper presents results from the Sumo project which aims at supporting continuous media types within the framework defined by the draft Open Distributed Processing standard. The paper advocates the use of synchronous languages within this framework for specifying and implementing real-time synchronization and QoS monitoring. A computational model and the realization of an infrastructure supporting this view are presented.

One of the major requirements of distributed multimedia applications is the need to maintain, often complex, real-time synchronisation constraints. More specifically, it is necessary to be able to manage arbitrary intraand intermedia synchronisation across activities in the distributed environment. Furthermore, it is important that such developments are integrated into emerging object-oriented standards for distributed computing. This paper presents an object-oriented programming model and associated implementation to meet these requirements. The main concepts behind the proposed approach are, firstly, the use of reactive objects for real-time control and synchronisation and, secondly, quality of service controlled bindings for predictable communication between objects. The flexibility of the approach is demonstrated by three contrasting examples of real-time synchronisation. In addition, the relationship of the proposals to hypermedia is examined.

We describe the design of an application platform able to run distributed real-time and multimedia applications alongside conventional UNIX programs. The platform is embedded in a micro-kernel/ PC environment and supported by an ATM based, QoS driven communications stack. We focus in particular on resource management aspects of the design and deal with CPU scheduling, network resource management and memory management issues. An architecture is presented which guarantees QoS levels of both communications and processing with varying degrees of commitment as specified by user level QoS parameters. The architecture uses admission tests to determine whether or not new activities can be accepted and includes modules to translate user level QoS parameters into representations usable by the scheduling, network and memory management subsystems.

In this paper we describe the design of a QoS driven communications stack in a micro- kernel operating system environment. The paper focuses on resource management aspects of the design and in particular we deal with CPU scheduling, network resource management and memory management issues. We describe an API at which QoS parameters are supplied by users and present an architecture able to guarantee these QoS requirements with varying degrees of commitment. The architecture includes modules to translate user level QoS parameters into representations usable by the scheduling, network and memory management subsystem. It also incorporates admission tests which determine whether or not a new connection can be accepted given a particular QoS requirement and resource availability. In addition to predictable resource allocation the architecture also incorporates dynamic QoS management which ensures that ongoing commitments continue to be met in the face of variable loading.

Currently, popular operating systems are unable to support the end-to- end real-time requirements of distributed continuous media. Furthermore, the integration of continuous media communications software into such systems poses significant challenges. This paper describes a design for distributed multimedia support in the Chorus micro-kernel operating system environment which provides the necessary soft real-time support while simultaneously running conventional applications. Our approach is to extend existing Chorus abstractions to include QoS configurability, connection oriented communications and real-time threads. The design uses the following key concepts: the notion of a flow to represent QoS controlled communication between two application threads, a close integration of communications and thread scheduling and the use of a split level scheduling architecture with kernel and user level threads. The paper shows how our design qualitatively improves performance over existing micro-kernel facilities by reducing the number of protection domain crossings and context switches incurred.

Currently, popular operating systems are unable to support the end-to-end real-time requirements of distributed continuous media. Furthermore, the integration of continuous media communications software into such systems poses significant challenges. This paper describes a design for distributed multimedia support in the Chorus micro-kernel operating system environment which provides the necessary soft real-time support while simultaneously running conventional applications. Our approach is to extend existing Chorus abstractions to include QoS configurability, connection oriented communications and real-time threads. The design uses the following key concepts: the notion of a flow to represent QoS controlled communication between two application threads, a close integration of communications and thread scheduling and the use of a split level scheduling architecture with kernel and user level threads. The paper shows how our design qualitatively improves performance over existing micro-kernel facilities by reducing the number of protection domain crossings and context switches incurred.