This thesis explores the application of functional programming in new areas and its implementation using new technologies. We show how functional languages can be used to implement solutions to problems in fuzzy logic using a number of languages: Haskell, Ginger and Aladin. A compiler for the weakly-typed, lazy language Ginger is developed using Java byte-code as its target code. This is used as the inspiration for an implementation of Aladin, a simple functional language which has two novel features: its primitives are designed to be written in any language, and evaluation is controlled by declaring the strictness of all functions. Efficient denotational and operational semantics are given for this machine and an implementation is devel- oped using these semantics. We then show that by using the advantages of Aladin (simplicity and strictness control) we can employ partial evaluation to achieve con- siderable speed-ups in the running times of Aladin programs.

The possibility of incorporating Artificial Intelligence (A.I) techniques into Visual Interactive Discrete Event Simulation was examined. After a study the current state of the art, work was undertaken to investigate the usefulness of PROLOG as a simulation language. This led to the development of a working Simulation Engine, allowing simulations to be developed quickly. The way PROLOG facilitated development of the engine indicated a possible usefulness as a medium for controlling external simulations. Tests on the feasibility of this were made resulting in the development of an assembler link which allows PROLOG to remotely communicate with and control procedural language programs resident on a separate microcomputer. Experiments using this link were then made to test the application of A.I. techniques to current visual simulations. Studies were carried out on the controlling of the simulation, the monitoring and learning from a simulation, the use of simulation as a window to expert system performance, and on the manipulation of the simulation. This study represents a practical attempt to understand and develop the possible uses of A.I. techniques within visual interactive simulation. The thesis concludes with a discussion of the advantages attainable through such a merger of techniques, followed by areas in which the research may be expanded.

A computational grid is a hardware and software infrastructure that provides dependable, consistent, pervasive, and inexpensive access to high-end computational capability. An ideal grid environment should provide access to the available resources in a seamless manner. Resource management is an important infrastructural component of a grid computing environment. The overall aim of resource management is to efficiently schedule applications that need to utilise the available resources in the grid environment. Such goals within the high performance community will rely on accurate performance prediction capabilities. An existing toolkit, known as PACE (Performance Analysis and Characterisation Environment), is used to provide quantitative data concerning the performance of sophisticated applications running on high performance resources. In this thesis an ASCI (Accelerated Strategic Computing Initiative) kernel application, Sweep3D, is used to illustrate the PACE performance prediction capabilities. The validation results show that a reasonable accuracy can be obtained, cross-platform comparisons can be easily undertaken, and the process benefits from a rapid evaluation time. While extremely well-suited for managing a locally distributed multi-computer, the PACE functions do not map well onto a wide-area environment, where heterogeneity, multiple administrative domains, and communication irregularities dramatically complicate the job of resource management. Scalability and adaptability are two key challenges that must be addressed. In this thesis, an A4 (Agile Architecture and Autonomous Agents) methodology is introduced for the development of large-scale distributed software systems with highly dynamic behaviours. An agent is considered to be both a service provider and a service requestor. Agents are organised into a hierarchy with service advertisement and discovery capabilities. There are four main performance metrics for an A4 system: service discovery speed, agent system efficiency, workload balancing, and discovery success rate. Coupling the A4 methodology with PACE functions, results in an Agent-based Resource Management System (ARMS), which is implemented for grid computing. The PACE functions supply accurate performance information (e. g. execution time) as input to a local resource scheduler on the fly. At a meta-level, agents advertise their service information and cooperate with each other to discover available resources for grid-enabled applications. A Performance Monitor and Advisor (PMA) is also developed in ARMS to optimise the performance of the agent behaviours. The PMA is capable of performance modelling and simulation about the agents in ARMS and can be used to improve overall system performance. The PMA can monitor agent behaviours in ARMS and reconfigure them with optimised strategies, which include the use of ACTs (Agent Capability Tables), limited service lifetime, limited scope for service advertisement and discovery, agent mobility and service distribution, etc. The main contribution of this work is that it provides a methodology and prototype implementation of a grid Resource Management System (RMS). The system includes a number of original features that cannot be found in existing research solutions.

The Grid provides a novel platform where the scientific and engineering communities can share data and computation across multiple administrative domains. There are several key services that must be offered by Grid middleware; one of them being the Grid Information Service( GIS). A GIS is a Grid middleware component which maintains information about hardware, software, services and people participating in a virtual organisation( VO). There is an inherent need in these systems for the delivery of reliable performance. This thesis describes a number of approaches which detail the development and application of a suite of benchmarks for the prediction of the process of resource discovery and monitoring on the Grid. A series of experimental studies of the characterisation of performance using benchmarking, are carried out. Several novel predictive algorithms are presented and evaluated in terms of their predictive error. Furthermore, predictive methods are developed which describe the behaviour of MDS2 for a variable number of user requests. The MDS is also extended to include job information from a local scheduler; this information is queried using requests of greatly varying complexity. The response of the MDS to these queries is then assessed in terms of several performance metrics. The benchmarking of the dynamic nature of information within MDS3 which is based on the Open Grid Services Architecture (OGSA), and also the successor to MDS2, is also carried out. The performance of both the pull and push query mechanisms is analysed. GridAdapt (Self-adaptive Grid Resource Monitoring) is a new system that is proposed, built upon the Globus MDS3 benchmarking. It offers self-adaptation, autonomy and admission control at the Index Service, whilst ensuring that the MIDS is not overloaded and can meet its quality-of-service,f or example,i n terms of its average response time for servicing synchronous queries and the total number of queries returned per unit time.

The conventional multilayer feedforward network having continuous-weights is expensive to implement in digital hardware. Two new types of networks are proposed which lend themselves to cost-effective implementations in hardware and have a fast forward-pass capability. These two differ from the conventional model in having extra constraints on their weights: the first allows its weights to take integer values in the range [-3,3] only, whereas the second restricts its synapses to the set {-1,0,1} while allowing unrestricted offsets. The benefits of the first configuration are in having weights which are only 3-bits deep and a multiplication operation requiring a maximum of one shift, one add, and one sign-change instruction. The advantages of the second are in having 1-bit synapses and a multiplication operation which consists of a single sign-change instruction. The procedure proposed for training these networks starts like the conventional error backpropagation procedure, but becomes more and more discretised in its behaviour as the network gets closer to an error minimum. Mainly based on steepest descent, it also has a perturbation mechanism to avoid getting trapped in local minima, and a novel mechanism for rounding off 'near integers'. It incorporates weight elimination implicitly, which simplifies the choice of the start-up network configuration for training. It is shown that the integer-weight network, although lacking the universal approximation capability, can implement learning tasks, especially classification tasks, to acceptable accuracies. A new theoretical result is presented which shows that the multiplier-free network is a universal approximator over the space of continuous functions of one variable. In light of experimental results it is conjectured that the same is true for functions of many variables. Decision and error surfaces are used to explore the discrete-weight approximation of continuous-weight networks using discretisation schemes other than integer weights. The results suggest that provided a suitable discretisation interval is chosen, a discrete-weight network can be found which performs as well as a continuous-weight networks, but that it may require more hidden neurons than its conventional counterpart. Experiments are performed to compare the generalisation performances of the new networks with that of the conventional one using three very different benchmarks: the MONK's benchmark, a set of artificial tasks designed to compare the capabilities of learning algorithms, the 'onset of diabetes mellitus' prediction data set, a realistic set with very noisy attributes, and finally the handwritten numeral recognition database, a realistic but very structured data set. The results indicate that the new networks, despite having strong constraints on their weights, have generalisation performances similar to that of their conventional counterparts.

Active learning is considered by many academics as an important and effective learning strategy. Students can improve the quality of their work by developing their higher cognitive skills through reflection on their own ideas, and through practice of analytic and evaluative skills. Assessment is a tool for learning, but traditional assessment methods often encourage surface learning, rather than deep learning which is an approach to developing higher cognitive skills. Peer assessment is one of the successful approaches, which can be used to enhance deep learning. It is a method of motivating students, involving students discussing, marking and providing feedback on other students' work. Although it is often used in the context of essays, it has seldom been applied to computer programming courses. The skill of writing good software includes understanding different approaches to the task, and stylistic and related considerations - these can be developed by evaluation of other programmers' solutions. As part of a study investigating the extent that peer assessment can promote deep learning to develop the higher cognitive skills in a programming course, a novel web-based peer assessment tool has been developed. - The process used is novel, since students are engaged not only in marking each other's work, but also in evaluating the quality of marking of their peers. - This system is designed to provide anonymity for the whole process, in order to ensure that the process is fair, and to encourage students to discuss without embarrassment by using an anonymous communication device (ACD) in a variety of roles (script authors, marker, and feedback marker). In this thesis, we describe and compare the learning theory and tools, which are relevant in learning computer programming. Deep learning, which can be described using the six categories of learning in Bloom's taxonomy, is discussed. Other peer assessment software tools are compared and discussed. The design and implementation of a novel web-based peer assessment system (with anonymous communication device) are described, and set in the context of the learning theories. The results of evaluating the tools through several experiments involving large programming classes and an essay writing module are reported. In this thesis, we also propose a new variation of Bloom's taxonomy, which is appropriate to describe the skills required for tasks such as programming. The results indicate that this approach to web-based peer assessment has successfully helped students to develop their higher cognitive skills in learning computer programming, and peer assessment is an accurate assessment method in a programming course.

No description available.

Tolerance measure is an important part of engineering, however, to date the system of applying this important technology has been left to the assessment of the engineer using appropriate guidelines. This work offers a major departure from the trial and error or random number generation techniques that have been used previously by using a knowledge-based system to ensure the intelligent optimisation within the manufacturing system. A system to optimise manufacturing tolerance allocation to a part known as Knowledge-based Automatic Tolerance Analysis (KATA) has been developed. KATA is a knowledge-based system shell built within AutoCAD. It has the ability for geometry creation in CAD and the capability to optimise the tolerance heuristically as an expert system. Besides the worst-case tolerancing equation to optimise the tolerance allocation, KATA's algorithm is supported by actual production information such as machine capability, types of cutting tools, materials, process capabilities etc. KATA's prototype is currently able to analyse a cylindrical shape workpiece and a simple prismatic part. Analyses of tolerance include dimensional tolerance and geometrical tolerance. KATA is also able to do angular cuts such as tapers and chamfers. The investigation has also led to the significant development of the single tolerance reference technique. This method departs from the common practice of multiple tolerance referencing technique to optimise tolerance allocation. Utilisation of this new technique has eradicated the error of tolerance stackup. The retests have been undertaken, two of which are cylindrical parts meant to test dimensional tolerance and an angular cut. The third is a simple prismatic part to experiment with the geometrical tolerance analysis. The ability to optimise tolerance allocation is based on real production data and not imaginary or random number generation and has improved the accuracy of the expected result after manufacturing. Any failure caused by machining parameters is cautioned at an early stage before an actual production run has commenced. Thus, the manufacturer is assured that the product manufactured will be within the required tolerance limits. Being the central database for all production capability information enables KATA to opt for several approaches and techniques of processing. Hence, giving the user flexibility of selecting the process plan best suited for any required situation.

The design of an application-specific integrated circuit of a parallel array processor is considered for recursive least squares by QR decomposition using Givens rotations, applicable in adaptive filtering and beamforming applications. Emphasis is on high sample-rate operation, which, for this recursive algorithm, means that the time to perform arithmetic operations is critical. The algorithm, architecture and arithmetic are considered in a single integrated design procedure to achieve optimum results. A realisation approach using standard arithmetic operators, add, multiply and divide is adopted. The design of high-throughput operators with low delay is addressed for fixed- and floating-point number formats, and the application of redundant arithmetic considered. New redundant multiplier architectures are presented enabling reductions in area of up to 25%, whilst maintaining low delay. A technique is presented enabling the use of a conventional tree multiplier in recursive applications, allowing savings in area and delay. Two new divider architectures are presented showing benefits compared with the radix-2 modified SRT algorithm. Givens rotation algorithms are examined to determine their suitability for VLSI implementation. A novel algorithm, based on the Squared Givens Rotation (SGR) algorithm, is developed enabling the sample-rate to be increased by a factor of approximately 6 and offering area reductions up to a factor of 2 over previous approaches. An estimated sample-rate of 136 MHz could be achieved using a standard cell approach and O.35pm CMOS technology. The enhanced SGR algorithm has been compared with a CORDIC approach and shown to benefit by a factor of 3 in area and over 11 in sample-rate. When compared with a recent implementation on a parallel array of general purpose (GP) DSP chips, it is estimated that a single application specific chip could offer up to 1,500 times the computation obtained from a single OP DSP chip.

In recent years, wireless networks with Internet services have become more and more popular. Technologies which integrate Internet and wireless networks have extended traditional Internet applications into a more flexible and dynamic environment. This research work investigates the technology that supports the connection between a Mobile Ad Hoc Wireless Network (MANET) and the Internet, which enables the current wireless Internet technologies to provide a ubiquitous wireless life style. With detailed analysis of the existing wireless Internet technologies and MANETs regarding their features and applications, the demand and lack of research work for an application to provide Internet connection to MANET is indicated. The primary difficulty for MANET and Internet connection is that the dynamic features of MANET do not suit the traditional connection methods used in infrastructure wireless networks. This thesis introduces new concept of the 'Gateway Awareness' (GAW) to the wireless devices in the MANET. GAW is a new routing protocol designed by the author of this thesis, at the University of Warwick. Based on GAW, an inclusive definition for the connection method, which supports the Internet connection and keeps the independency of routing in MANET, is addressed. Unlike other research work, this method supports the MANET and Internet communication in both directions. Furthermore, it explores possible ways of using the Internet as an extension for wireless communications. The GAW routing method is developed from destination sequenced distance vector (DSDV) routing protocol. However, it defines a layer of wireless nodes (known as GAWNs) with exclusive functions for the Internet connection task. The layer of GAWNs brings a new set of route update and route selection method. Simulations show that the GAW routing method provides quality Internet connection performance in different scenarios compared with other methods. In particular, the connection is completed with minimum effect on the independent MANET while the routing efficiency and accuracy is guaranteed.