Alan Ruddle

An Holistic View of Quality of Service

The growing presence of digital multimedia in everyday life, as evidenced by the continuing increase in Internet usage, is expected to augment or replace traditional modes of social and economic interaction. Face-to-face meetings, paperwork, and office-space are all expected to diminish in importance as cyberspace provides new facilities and interaction modes. In the higher education sector, distributed learning environments (DLEs) are being cited as solutions to the ambitious political goals of better education, wider access and lower costs. DLEs are still in their infancy however, and if they are to succeed in transforming conventional educational practice, they must become quality of service (QoS) aware. The notion of QoS originated in computer networking, but if it is not addressed appropriately in all a systems components then there may be little or no benefit for that systems users, regardless of the network quality. There is a real danger that policy makers in the educational sphere will simply equate the needs of the end users, namely teachers and students, with network-level QoS. Our experience with the TAGS framework for constructing DLEs has shown that QoS must be taken into account from many perspectives before traditional educational processes can be satisfactorily transformed. This paper contributes towards an holistic view of QoS in DLEs by reviewing the network perspective and augmenting it with other, equally important, concerns.

The architecture of a framework for building distributed learning environment

Distributed learning environments (DLEs) are being cited as solutions to the ambitious political goals of better education, wider access and lower costs. The paper outlines the architecture of the TAGS framework (C. Allison et al., 2000) for building DLEs. The main features that the architecture facilitates are strong support for group work, anytime/anywhere access, the input of real world data, event monitoring facilities, authentication, authorisation and protection, and the controlling of delay between user actions and system responses. These features have required the development of an expanded view of quality of service (QoS) parameters. The architecture has been refined over a number of years as the result of collaborative development between users and programmers. In particular, we have found that the bringing together of systems programmers, computer science researchers and subject-specific educationalists has allowed needs to be identified and addressed effectively. We believe that TAGS makes it possible for educationalists to easily build DLEs that meet their needs.

Quality of service issues in distributed learning environments

The growing presence of digital multimedia in everyday life, as evidenced by the continuing increase in Internet usage, is expected to augment or replace traditional modes of social and economic interaction. Face-to-face meetings, paperwork, and office-space are all expected to diminish in importance as cyberspace provides new facilities and interaction modes. In the higher education sector distributed learning environments (DLEs) are being cited as solutions to the ambitious political goals of better education, wider access and lower costs. DLEs are still in their infancy however, and if they are to succeed in transforming conventional educational practice they must become quality of service (QoS) aware. The notion of QoS originated in computer networking but if it is not addressed appropriately in all a system components then there may be little or no benefit for that systems users, regardless of the network quality. There is a real danger that policy makers in the educational sphere will simply equate network level QoS with the needs of the end users, namely teachers and students. Our experience with the TAGS framework for constructing DLEs has shown that QoS must be taken into account from many perspectives before traditional educational processes can be satisfactorily transformed. This paper contributes towards an understanding of QoS issues in distributed learning environments by reviewing the network perspective and augmenting it with other, equally important, concerns.

Analysing the latency of World Wide Web applications

As the Web becomes increasingly used for routine social and economic interactions, the issue of interactive delay, as experienced by end‐users, has become an ongoing cause for concern. This paper reports on an approach of identifying and analysing the factors that contribute towards that delay. The evolution of a structured timing model is described, which forms the basis for decomposing delay into its constituent parts. The four high‐level components identified are the server, network, protocol and client. Methodologies for determining the contribution of each element are presented and their use is illustrated in a case study of the traffic associated with an operational distributed learning environment (DLE) serving six universities over a period of several months. Quality of service (QoS) is key to the success of DLEs and this includes an understanding of their relationship with the underlying networks and systems upon which they are built and deployed. Delay, as experienced by the end user, is one of the key QoS parameters for a DLE. The importance of developing an understanding of the nature of delay in a DLE is twofold: it informs decisions about where to target effort and resources in order to achieve improvements; and it provides techniques that can be used as the basis for an early‐warning and advisory system for QoS aware applications. The latter functionality is illustrated by a novel solution that modifies TCPs congestion avoidance algorithm for the benefit of interactive Web traffic. Copyright

A measurement based approach to TCP congestion control

Congestion is a major Internet problem. TCP has evolved in attemptting to deal with this and now employs congestion control algorithms that effectively limit the bandwidth available to any one connection. However, as TCP is not the only transport protocol used on the Internet, the growth of non-TCP traffic has lead to the formulation of TCP-Friendly formulae. It is recommended that all best effort traffic conform to these formulae, and they act as a benchmark against which the success of any congestion control algorithms can be judged. Unfortunately, key assumptions behind the formulation of TCPs algorithms - low bandwidth, and bulk data transfers - are often no longer valid. In this paper it is shown that the combination of increasing bandwidth and short web transfers means that a significant amount of TCP traffic fails to obtain the bandwidth implied by the TCP-friendly formulae. An alternative, measurement-based approach that predicts a fairer starting window size for a connection is presented and evaluated. Information about the characteristics of particular network paths is dynamically maintained and used to suggest a fairer starting window size for new connections. Minor modifications to TCP allow these suggestions to be used to set start-up control variables, and thereby strengthen the negative correlation between the bandwidth used by a connection and the level of congestion on its network path. The measurement-based approach is realised through the use of a Location Information Server (LIS). The LIS performs centralised passive monitoring of transport headers in order to derive network level path information. Location Information Packets (LIP) are used to communicate suggested start-up variables to local hosts. The design, implementation and evaluation of an LIS, LIP and participating hosts TCP software are presented.

Visualising interactions between TCP's congestion and flow control algorithms

The Transmission Control Protocol, TCP, is used in over 90% of the traffic on the Internet. The effect of the protocol itself on the efficiency and fairness of Internet traffic is accordingly the focus of much study. In particular, the tight timeliness constraints required by Web traffic require that TCP should not unnecessarily introduce delay. TCP continues to evolve, and insights into its behaviour are of great potential value. The work reported offers novel approaches to understanding the subtle interactions between TCPs congestion control and flow control algorithms across a wide range of transfer sizes and congestion regimes. A model of TCPs behaviour is described in the form of a program used to generate data sets. These are then processed to generate 3D graphics that are used to visualise the model. This leads to a re-evaluation of TCPs congestion control algorithms within the context of high bandwidth WANs and a high level of multiplexing. As new access technologies are deployed and more users acquire high bandwidth access to the Internet, the limitation imposed by the needs of competing traffic will reinforce the importance of the fair distribution of network resources.

Realising real time multimedia groupware on the web

Real-time audio and video conferencing has not yet been satisfactorily integrated into web-based groupware environments. Conferencing tools are at best only loosely linked to other parts of a shared working environment. This is mainly due to the special Quality of Service (QoS) requirements that these types of resources demand. This paper describes an approach to overcoming this problem by integrating the management of video and audio conferences into the resource allocation mechanism of an existing web-based groupware framework. The issue of adaptation is discussed and a means of initialising multimedia session parameters based on predicted QoS is described. In addition to linking audio and video media rendering quality to the predicted network QoS, the use of models to reduce bandwidth is also explored, replacing video content with control data. Component technologies utilised include the TAGS groupware framework, the Java Media Framework, a Conference Control Architecture and VRML-based avatars that can intermix with real actors.