Jason Holdsworth

Beyond traditional program slicing

Traditional program slices are based on variables and statements. Slices consist of statements that potentially affect (or are affected by) the value of a particular variable at a given statement. Two assumptions are implicit in this definition: 1) that variables and statements are concepts of the programming language in which the program is written, and 2) that slices consist solely of statements.Generalised slicing is an extension of traditional slicing where variables are replaced by arbitrary named program entities and statements by arbitrary program constructs. A model of generalised slicing is presented that allows the essence of any slicing tool to be reduced to a node marking process operating on a program syntax tree. Slicing tools can thus be implemented in a straight-forward way using tree-based techniques such as attribute grammars.A variety of useful program decompositions are shown to be instances of generalised slicing including: call graph generation, interface extraction, slicing of object-oriented inheritance hierarchies and slices based on type dependences. Examples are also given of how slicing can enhance understanding of formal compiler specifications and aid the creation of subset language specifications.

Mobile augmented reality based context-aware library management system

Mobile augmented reality has gained popularity in recent years due to the technological advances of smartphones and other mobile devices. One particular area in which mobile augmented reality is being used is library management. However, current mobile augmented reality solutions in this domain are lacking in context-awareness. It has been suggested in the literature that agent programming may be suitable at overcoming this problem, but little research has been conducted using modern mobile augmented reality applications with agents. This paper aims to bridge this gap through the development of an agent-based, mobile augmented reality prototype, titled Libagent. Libagent was subjected to five experiments to determine its suitability, efficiency, and accuracy for library management. The results of these experiments indicate that agent-based mobile augmented reality is a promising tool for context-aware library management.

Hand-Held Mobile Augmented Reality for Collaborative Problem Solving: A Case Study with Sorting

Due to the advances in mobile technology, mobile augmented reality has been widely used for many disciplines. The ubiquity nature of mobile augmented reality supports a flexible, engaging and entertaining learning environment. However, most mobile devices are hand-held, and they require multitasking (mobile information processing and learning) that is a major hurdle for learning. This paper investigates the effects of multitasking of hand-held mobile augmented reality for problem solving. We design and implement a robust framework, and conduct a case study of sorting activities with two distinct groups: individual and collaborative pair settings. Experimental results demonstrate that 1) there is no significant difference between two groups in sorting without our proposed system, 2) there is a significant improvement with collaborative sorting with our proposed system. Test statistics confirm that our proposed system significantly improve collaborative pair sorting activities.

Automated Outlier Removal for Mobile Microbenchmarking Datasets

Microbenchmarking is a useful tool for fine-grained performance analysis, and represents a potentially valuable tool in the development of mobile applications and systems. However, the fine-grained measurements of microbenchmarking are inherently susceptible to noise from the underlying operating system and hardware. This noise includes outliers that must be removed in order to produce meaningful results. Existing microbenchmarking implementations utilise only simple mechanisms for removing outliers. In this paper we propose a heuristic for the automated removal of outliers from mobile microbenchmarking datasets. We then simplify this heuristic for use on mobile devices. Empirical evaluation demonstrates that our outlier removal heuristics are effective across microbenchmarking datasets collected from a range of mobile devices. Our simplified heuristic operates in log-linear time, making it suitable for use on resource-constrained mobile devices. The ability to perform outlier removal on-device without the need for post-processing on desktop or server hardware enhances the utility of mobile microbenchmarking tools. Our results present interesting opportunities for further studies across a broader range of device platforms.

Web-Based Collaborative Document Writing for Emergency Management

This paper describes the progress of our research into collaborative document writing for emergency management. Our focus is specifically on the use of web technologies to improve communication and shared knowledge in emergency situations. We present a prototype system built using Differential Synchronisation with flexible locking and user attribution features. Based on the past shortcomings of technologies and documents such as situation reports, we consider the potential benefits of a web-based system with flexible locking and user attribution for collaborative situation report writing.

Efficient hierarchical clustering for single-dimensional data using CUDA

Hierarchical clustering is a widely-used and well-researched clustering technique. The classical algorithm for agglomerative hierarchical clustering is prohibitively expensive for use with large datasets. Numerous algorithms exist to improve the efficiency of hierarchical clustering for various linkage metrics, and for large datasets. Recent research has proposed approaches for improving the efficiency of hierarchical clustering through parallelism. The newest approaches utilise GPGPU technologies, which facilitate massive parallelism on commodity consumer hardware. Existing GPGPU implementations fail to maximise the number of merges that can be performed in parallel, and feature high use of memory. These limitations prevent existing implementations from achieving the full performance offered by GPGPU. In this paper, we propose a novel GPGPU algorithm for hierarchical clustering of single-dimensional data. Our proposed algorithm exploits the unique characteristics of one-dimensional data to maximise merge parallelism and significantly reduce memory requirements. Validation demonstrates that our proposed algorithm produces equivalent results to the classical algorithm for both the single-linkage and complete-linkage metrics. Benchmarking results show that our algorithm scales well to large datasets, and offers a substantial speed-up over the classical algorithm. Future work will look to extend our proposed approach to larger datasets with higher dimensions.

An input-centric performance model for computational offloading of mobile applications

Abstract Computational offloading frameworks are a widely-researched technology for optimising mobile applications through the use of cloud resources. Existing frameworks fail to fully account for the effect of input data characteristics on application behaviour. Comprehensive timing models exist in the literature, but feature information requirements and performance overheads that preclude use on mobile devices. In this paper, we propose a conceptual model for an input-centric view of application performance. Our proposed model simplifies the existing count-and-weights and pipeline timing models to significantly reduce their information and processing requirements, facilitating use on resource-constrained mobile devices. Our proposed model also utilises symbolic execution techniques to account for the effects of application input data characteristics. Validation with both synthetic and real device datasets demonstrates that our model provides an extremely accurate approximation of the count-and-weights model. Results demonstrate the predictive power of our model for linear execution paths with no loops or recursion. Further work with improved symbolic execution techniques may look to expand application of our proposed model to real-world use cases. The proposed input-centric approach provides a promising foundation for incorporating a deeper level of application-specific knowledge into computational offloading framework cost models, with the potential to contribute to higher-quality offloading decisions.

Untangling the Edits: User Attribution in Collaborative Report Writing for Emergency Management

This paper describes the progress of our ongoing research into collaborative report writing for emergency management using web technologies to improve communication and shared knowledge in emergency situations. Specifically, the work presented in this paper focuses on the development of a user attribution framework that enhances the Differential Synchronisation (diffsync) technique by exploiting its diff operation. This technique improves real-time collaborative editing for emergency management by combining the benefits of user attribution with diffsync features such as convergence, scalability, and robustness to poor network environments. As a proof of concept, we implement a prototype collaborative system and report results of simulations to test scalability, efficiency, and correctness. Further, we consider the potential benefits of this framework for web-based collaborative report writing in the context of emergency management.

Context-Aware Mobile Augmented Reality for Library Management

Mobile augmented reality applications aim to simplify everyday tasks by providing additional information to increase efficiency. There are many challenges facing the field, including efficient and accurate methods for providing context-awareness. It is believed that research in the field of multi-agent systems can assist such applications in reducing the complexity of everyday tasks. To determine the benefits of using software agents to support mobile augmented reality, we develop a prototype titled Libagent for library management tasks. Our experimental results indicate that Libagent provides benefits to users by reducing errors.

Graph traversal and graph transformation

Graph traversal algorithms are important since graphs are a common data structure in which information is distributed. None of the existing algorithmic paradigms focuses on graph traversal. This article introduces enNCE substitution as an extension to eNCE substitution. The relationship between enNCE substitution and eNCE substitution is explored. Moreover, an enNCE graph transformation system is defined and then used to generate depth-first and breadth-first graph traversal. Thus, enNCE graph transformation shows potential as a fundamental concept for a traversal-oriented algorithmic paradigm.