Using context and preferences to implement self-adapting pervasive computing applications

Applications that exploit contextual information in order to adapt their behaviour to dynamically changing operating environments and user requirements are increasingly being explored as part of the vision of pervasive or ubiquitous computing. Despite recent advances in infrastructure to support these applications through the acquisition, interpretation and dissemination of context data from sensors, they remain prohibitively difficult to develop and have made little penetration beyond the laboratory. This situation persists largely due to a lack of appropriately high-level abstractions for describing, reasoning about and exploiting context information as a basis for adaptation. In this paper, we present our efforts to address this challenge, focusing on our novel approach involving the use of preference information as a basis for making flexible adaptation decisions. We also discuss our experiences in applying our conceptual and software frameworks for context and preference modelling to a case study involving the development of an adaptive communication application.

Microanalysis of quenched and self-extinguished aluminium rods burned in oxygen

Promoted ignition tests and quench tests have been conducted and analysed for 3.2 mm aluminum rods in 99.995% oxygen. Tests have been conducted in oxygen pressures varying from 538 kPa to 773 kPa. Samples that self-extinguished or were quenched were selected for further analysis. The microstructure of the selected samples were analysed by electron microscopy, using energy dispersive spectrometry and electron back-scatter techniques, to identify and visualize, respectively, the species present. The grain structures of these samples were etched, viewed and photographed under polarized light by an optical microscope. From the micrographs produced by the post-test analysis, clearly defined boundaries between the oxide and the melted and resolidified metal have been observed. In both the melted and resolidified metal and the oxide layer, significant numbers of gas bubbles, solid inclusions and several diffuse oxide bubbles have been captured during the cooling process. It is concluded that convective movement is occurring within the molten drop and that analysis of quenched samples provides more useful information on the state of the burning droplet than samples allowed to cool slowly to room temperature. Recommendations are made regarding future investigations into aluminum burning, focusing on the transport of reactants through the liquid oxide layer.