No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

The effective performance of computer systems is to a large measure determined by the synergy between the processor architecture, the instruction set and the compiler. In the past, the sequencing of information within processor architectures has normally been synchronous: controlled centrally by a clock. However, this global signal could possibly limit the future gains in performance that can potentially be achieved through improvements in implementation technology. This thesis investigates the effects of relaxing this strict synchrony by distributing control within processor architectures through the use of a novel asynchronous design model known as a micronet. The impact of asynchronous control on the performance of a RISC-style processor is explored at different levels. Firstly, improvements in the performance of individual instructions by exploiting actual run-time behaviours are demonstrated. Secondly, it is shown that micronets are able to exploit further (both spatial and temporal) instructionlevel parallelism (ILP) efficiently through the distribution of control to datapath resources. Finally, exposing fine-grain concurrency within a datapath can only be of benefit to a computer system if it can easily be exploited by the compiler. Although compilers for micronet-based asynchronous processors may be considered to be more complex than their synchronous counterparts, it is shown that the variable execution time of an instruction does not adversely affect the compiler's ability to schedule code efficiently. In conclusion, the modelling of a processor's datapath as a micronet permits the exploitation of both finegrain ILP and actual run-time delays, thus leading to the efficient utilisation of functional units and in turn resulting in an improvement in overall system performance.

University of Technology, Sydney. Faculty of Information Technology. / As IT environments grow in complexity and diversity, their strategic management becomes a critical business issue. Enterprise architectures (EA’s) provide support by ensuring that there is alignment between an enterprise’s business objectives and the IT systems that it deploys to achieve these objectives. While EA is a relatively new discipline, it has already found widespread commercial application. It is likely that EA will receive even more focus as IT environments continue to grow in complexity and heterogeneity. Despite this widespread acceptance of EA as a valuable IT discipline, there are several serious challenges that contemporary EA approaches are yet to overcome. These arise from the fact that currently, there is no unified EA modelling language that is also easy to use. A unified EA modelling language is one that is able to describe a wide range of IT domains using a single modelling notation. Without a unified, easy to use EA modelling language, it is impossible to create integrated models of the enterprise. Instead, a variety of modelling languages must be used to create an EA, leading to enterprise models that are inconsistent, incomplete and difficult to understand. The need to use multiple modelling languages also places a high cognitive load on modellers and excludes non-IT specialists from developing or using these models, even though such people may be the most important stakeholders in an EA program. The research presented in this thesis tackles these problems by developing a metaphor-based approach to the construction of unified EA modelling languages. Contemporary approaches to the understanding of metaphor are surveyed, and it is noted that one way to understand metaphor is to view it as part of a dynamic type hierarchy. This understanding of metaphor is related to the development of enterprise models and it is shown that highly abstract metaphors can be used to provide conceptually unified models of a range of enterprises and their component structures. This approach is operationalised as methodology that can be used to generate any number of unified EA modelling languages. This methodology is then applied to generate a new, unified EA modelling language called ‘LEAN’ (Lightweight Enterprise Architecture Notation). LEAN is evaluated using a mixed-methods research approach. This evaluation demonstrates that LEAN can be used to model a wide range of domains and that it is easy to learn and simple to understand. The application of the theoretical principles and methodology presented in this thesis can be expected to improve the understandability and consistency of EA’s significantly. This, in turn, can be expected to deliver significant tangible business benefits through improved strategic change management that more closely aligns the delivery of IT services with business drivers. The findings in this research also provide fertile ground for further research. This includes the development and comparative evaluation of alternative unified languages, further research into the use of the methodology presented to align architectures at various levels of abstraction, and the investigation of the applicability of this theoretical approach to other, non-IT disciplines.

Shared control is a technique to provide interactive autonomy in a telerobotic task, replacing the requirement for pure teleoperation where the operator's intervention is unnecessary or even undesirable. In this thesis, a geometrically correct theory of shared control for teleoperation is developed using differential geometry. The autonomous function proposed is force control. In shared control, the workspace is commonly partitioned into a "position domain" and a "force domain". This computational process requires the use of a metric. In the context of manifolds, these are known as Riemannian metrics. The switching matrix is shown to be equivalent to a filter which embodies a Riemannian metric form. However, since the metric form is non-invariant, it is shown that the metric form must undergo a transformation if the measurement reference frame is moved. If the transformation is not made, then the switching matrix fails to produce correct results in the new measurement frame. Alternatively, the switching matrix can be viewed as a misinterpretation of a projection operator. Again, the projection operator needs to be transformed correctly if the measurement reference frame is moved. Many robot control architectures preclude the implementation of robust force control. However, a compliant device mounted between the robot wrist and the workpiece can be a good alternative in lieu of explicit force control. In this form of shared control, force and displacement are regulated by control of displacement only. The geometry of compliant devices is examined in the context of shared control and a geometrically correct scheme for shared control is derived. This scheme follows naturally from a theoretical analysis of stiffness and potential energy. This thesis unifies some recent results formulated for robotic hybrid position / force control under the modern framework of differential geometry and Lie groups.