Context-aware applications are a class of mobile computer applications where the behaviour of the application depends on the user's current context, for example their location, the current temperature, the proximity of other people or equipment, ... .
The talk will cover a technology, based on SGML, that aims to aid the authorship and use of a wide class of context-aware applications. Some sample applications will be considered, including tourist information, a general form of paging, aids to mobile workers and rolling surveys.
The potential value of allowing algorithms to make random choices first became apparent in the mid-seventies, with the discovery of fast randomised tests for primality. From those origins, an important area of research grew and flourished, to the extent that randomness is now seen as perhaps the most fruitful extension so far made to the classical model of computation. I shall argue that, twenty years on, this line of research is still throwing up many surprises, in the form of randomised algorithms that are not only efficient but also arrestingly simple and wonderfully elegant.
A partial evaluator, given a program and a known ``static'' part of its input data, outputs a residual program in which computations depending only on the static data have been performed.
Ideally the partial evaluator would be a ``black box'' able to extract nontrivial static computations whenever possible; which never fails to terminate; and which always produces residual programs of reasonable size and maximal efficiency, so all possible static computations have been done. Practically speaking, partial evaluators fall short of this goal; they may loop, sometimes pessimise, and/or explode code size.
A partial evaluator is analogous to a spirited horse: while impressive results can be obtained when used well, the user must know what he/she is doing. Our thesis is that this knowledge can be communicated to new users of these tools.
This paper presents a series of examples, concentrating on a quite broad and on the whole successful application area: using specialisation to remove interpretational overhead. It presents a series of examples, both positive and negative, to illustrate the effects of program style on the efficiency and size of the target programs obtained by specialising an interpreter with respect to a known source program.
Hypermedia technology has reached the stage of providing excellent access to distributed multimedia, in particular through the World Wide Web. However, application developers are still faced with many problems when dealing with large-scale systems, such as the authoring effort required to create all the appropriate links, the maintenance of link integrity during the re-organisation of large structures and hypermedia linking to and from third party data. The Microcosm system was developed by the Multimedia Group at the University of Southampton specifically for managing large scale hypermedia resources. This talk will discuss how this project has evolved and which direction(s) it is heading in the future including.
The language Java is enjoying a rapid rise in popularity as an application programming language. For many applications an effective provision of database facilities is required. Here we report on a particular approach to providing such facilities, called "orthogonal persistence". Persistence allows data to have lifetimes that vary from transient to (the best approximation we can achieve to) indefinite. It is orthogonal persistence if the available lifetimes are the same for all kinds of data.
We argue that an efficient implementation of an orthogonally persistent language will yield performance benefits and productivity benefits as well as enhancing security and mobility. A prototype is available and is in use for several projects at a number of sites. The prototype does not have all of the facilities intended for PJava, but is, nevertheless, an interesting vehicle for evaluation and further development.
The talk will cover both the background and the design issues that have led to the current design of PJava. It will then discuss a few aspects of the implementation and plans for further development.
Clocked digital circuit methods have underpinned processor design for over 40 years; thus these methods are now deeply engrained in the psyche of electronic engineers. Consequently most processors are designed with a clocked implementation in mind, even when viewed at an abstract level.
Self-timed logic replaces global clocks with localised handshake signals. This alternative design method is radically different from clocked techniques and suggests novel processor designs. This seminar introduces current self-timed processor architectures and illustrates the benefits and pitfalls of this design method. Recent developments of a self-timed rotary pipeline architecture will be included.
Over the past few years, several wide area network (WAN) testbeds have been deployed to explore the issues involved in the creation of large scale, high performance production WANs.
The MAGIC project demonstrated a high-speed, wide-area IP/ATM network that supports a real-time terrain visualization application and a high-speed distributed storage system. The original MAGIC network extended from Kansas to Minnesota, with 2.4 Gb/s facilities provided by Sprint, US West, and Southwestern Bell.
The AAI testbed interconnects several Department of Defense High Performance Computing (HPC) sites scattered from California to Maryland, at 155 Mb/s rates. In addition, it interconnects the MAGIC testbed to the ATDNet testbed in the Washington, DC, area. Two sites are also equipped with NASA ACTS satellite high-data rate (622 Mb/s) terminals for experiments with alternate modes of connectivity.
Both of these testbeds have provided an opportunity for the early deployment of IP/ATM technology, evaluation of its strengths and shortcomings, and identification of solutions to some of the limitations.
This talk will discuss the MAGIC & AAI testbeds, experiences with these networks, and the lessons learned.