In the commonly used method of bolting to secure parts of equipment and structure, the bolts must be tightened to an adequate preload force. Failure to do so could affect the integrity of the structure, as well as the efficient running of the site and, crucially, employees’ safety. In this project, we consider the use of artificial intelligence (AI) techniques to analyse maintenance videos and identify the unwanted loosening of bolts over time in order that they might be used as additional tools in a continuous maintenance plan. We found that accuracy levels of up to 97% could be achieved in identifying bolt rotation with our proposed machine learning-based triplet loss architecture. The use of gradient-weighted class activation mapping (Grad-CAM) visualisations to identify areas of the image where change had occurred enabled us to test how robust our model was to noise in the data. This explanation may assist users in safety-critical environments guiding them to the problem, and helping mitigate the black-box nature of machine learning algorithms. Whilst the accuracy of the models varies depending on the rotational angle of the bolt, we clearly demonstrate that triplet loss is a good basis for performing change detection in industrial settings. Furthermore, Grad-CAM has shown to be a useful technique to help a user understand the decisions made by the network and allow them to see where unwanted rotation has occurred.

Load balancing dynamically optimizes cloud resources and performance, and enhances the performance of \napplications deployed on cloud. We have chosen to investigate the class of cloud deployed web-based threetier business applications. There is a problem with load balancing for this class of applications when they \nsuffer from overload due to sudden flash crowds and resource failures. We propose a novel weight assignment \nload balancing algorithm to address this problem. Our approach utilises five carefully selected server metrics \nto efficiently distribute load among virtual machines. First, we validated our novel algorithm by comparing it \nwith a baseline load-balancing algorithm and round-robin algorithm. Then, we experimentally evaluated our \nsolution, by varying the number of user requests and carefully measuring response times and throughput. The \nexperiments were performed on a private cloud environment testbed running OpenStack. Our experimental \nresults show that our approach improves the response time of user requests by 5.66% compared to the baseline \nalgorithm and 15.15% compared to round-robin algorithm in flash crowd scenario. In addition, while handling \nbetween 110% to 190% overload, our approach improved response times in all scenarios. Consequently, our \nnovel algorithm outperforms the baseline and round-robin algorithms in overload conditions.

Web-based business applications commonly experience user request spikes called flash crowds. Flash crowds \nin web applications might result in resource failure and/or performance degradation. To alleviate these challenges, this class of applications would benefit from a targeted load balancer and deployment architecture of \na multi-cloud environment. We propose a decentralised system that effectively distributes the workload of \nthree-tier web-based business applications using geographical dynamic load balancing to minimise performance degradation and improve response time. Our approach improves a dynamic load distribution algorithm \nthat utilises five carefully selected server metrics to determine the capacity of a server before distributing \nrequests. Our first experiments compared our algorithm with multi-cloud benchmarks. Secondly, we experimentally evaluated our solution on a multi-cloud test-bed that comprises one private cloud, and two public \nclouds. Our experimental evaluation imitated flash crowds by sending varying requests using a standard exponential benchmark. It simulated resource failure by shutting down virtual machines in some of our chosen \ndata centres. Then, we carefully measured response times of these various scenarios. Our experimental results \nshowed that our solution improved application performance by 6.7% during resource failure periods, 4.08% \nand 20.05% during flash crowd situations when compared to Admission Control and Request Queuing benchmarks.

Agile methods are a well-established paradigm in the software development field. Agile adoption has contributed to improving software quality. However, software products are vulnerable to security challenges and susceptible to cyberattacks. This study aims to improve security of software products when using an agile software development process. A multi-methods qualitative research approach was adopted in this study. First, we conducted semi-structured interviews with 23 agile practitioners having varied years of cybersecurity experiences. An approach informed by grounded theory methodology was adopted for data analysis. Second, we developed a novel practice-based agile software development process model derived from the results of the data analysis conducted. Third, we validated the model through a focus group comprising five senior agile cybersecurity professionals to evaluate its relevancy and novelty. The study has identified 26 security practices, organized into the six - software development life-cycle phases: planning, requirements, design, implementation, testing, and deployment. We have mapped the practices onto four swim lanes each representing an agile role. The self-organizing team is exclusively involved in three security practices, the security specialist in nine, penetration tester in one and the DevOps team collaborates on one with the security specialist. There are also seven practices that are collaboratively performed by the self-organizing team and the security specialist. Each of the practices in the model was examined during the validation phase of the study. There are two contributions in this study. First, the paper proposes a novel practice-based model comprising of 26 security practices mapped to agile roles. Second, we propose a new practice, in response to an observed lack of collaborative ceremonies, to disseminate awareness of and hence compliance with security standards.

Abstract Non‐functional requirements define qualities of the software system that ensure effectiveness while embedding any constraints and restrictions on the design. A challenge rises with agile implementation in handling non‐function requirements in regulated environments. Thus, a practitioner's perceptions of agile method tailoring are described in relation to inter‐team boundaries and non‐functional requirements. The research comprises 18 practitioner interviews from two multinational agile software development companies. Interviews were recorded, transcribed, and analysed using an approach informed by grounded theory and information flow models were used to compare and contrast interactions of processes. It was discovered that one of the case study companies managed non‐functional requirements as artefacts in their agile methodology, while the other company reverts to conventional plan‐based software development practices of documentation, timeline estimations, and safety critical requirements. This research creates a detailed comparison of these contrasting approaches. The main contribution of this study is a set of proposed recommendations to deal with non‐functional requirements in a regulated environment using agile techniques. The introduction of two new artefacts, Documentation Work Item and Safety Critical Work Item, is recommended and it is accompanied with an illustrative example, to transform the handling of documentation and safety critical requirements in a more agile way.

Organisations tailor agile methods to fit their specific needs. There are two recognised approaches to agile method tailoring: Contingency Factors and Method Engineering. Spotify has developed its own agile culture, which has become influential among agile proponents. The Spotify model has formed the basis of agile methods used in numerous other organisations. We have identified a lack of research into the tailoring approach used in the Spotify model.

DevOps is described as a software engineering culture and philosophy that utilises cross-functional teams to build, test and release software faster and more reliably through automation. Research shows that its adoption speeds up software delivery time, improve quality, security, and collaboration in software development. One controversial issue has been whether DevOps is an organisation-wide culture or a job description. As DevOps is an emerging concept, its definitions and best practices are still hazy, making its implementation in practice less informed and somewhat risky. The rising trend of DevOps adoption among software development practitioners therefore heightens the need for in-depth investigation into its implementation.This paper seeks to contribute to the above by critically examining DevOps implementation in practice through an exploratory case study, based on interviews with 11 industry practitioners across nine organisations. Transcripts of interviews were coded and analysed using a method informed by Grounded Theory. This study provides an empirical taxonomy of DevOps implementation, describing developers' interaction with On-premises Ops, Outsourced Ops, DevOps teams, and DevOps bridge teams. We present a novel mapping of the approaches to on-premises and cloud-based deployments, and identified the facilitators of DevOps practices in the different modes. We further identified three distinct groups of activities in the fourth mode: provisioning and maintenance of physical systems, function virtualisation and creation of automated pipelines, and development, deployment and maintenance of applications, which may have given rise to the implementation of DevOps as bridge teams. Interviewees claimed these distinctions allowed developers to focus on delivering value for the business.

Agile software development has become increasingly common in the context of large-scale organisations. Typically, software organisations tailor agile methods to fit their needs and ultimately maximise success. The size of the organisation, business goals, and operative models are some examples of factors for which agile methods are tailored. Spotify model is introduced to facilitate the development of a very large-scale project with a Business-to-Consumer (B2C) model, but mission-critical large-scale projects with Business-to-Business (B2B) model are not addressed by the model. Hence, a question that imposes itself is: What are practitioner perceptions of agile tailoring when using the Spotify model? In this paper, we conduct a longitudinal embedded case study to investigate practitioner perceptions of agile method tailoring on a large-scale mission-critical project in B2B environment. The case study lasted over 21 months during which 14 semi-structured interviews were conducted. To analyse the collected data, the Grounded Theory (GT) is adopted. As a result, we identify 44 tailored practices and attributes for B2B product development. Based on this tailoring, 4 influential factors on "Spotify Tailoring" have been derived. These derived factors are worth considering for other organisations concerned with agile method tailoring for large-scale mission-critical projects in B2B context.

Syntactic parsing is a time-consuming task in natural language processing particularly where a large number of text files are being processed. Parsing algorithms are conventionally designed to operate on a single machine in a sequential fashion and, as a consequence, fail to benefit from high performance and parallel computing resources available on the cloud. We designed and implemented a scalable cloud-based architecture supporting parallel and distributed syntactic parsing for large datasets. The main architecture consists of a syntactic parser (constituency and dependency parsing) and a MapReduce framework running on clusters of machines. The resulting cloud-based MapReduce parsing is able to build a map where syntactic trees of the same input file have the same key and collect into a single file containing sentences along with their corresponding trees. Our experimental evaluation shows that the architecture scales well with regard to number or processing nodes and number of cores per node. In the fastest tested cloud-based setup, the proposed design performs 7 times faster when compared to a local setup. In summary, this study takes an important step toward providing and evaluating a cloud-hosted solution for efficient syntactic parsing of natural language data sets consisting of a large number of files.

The introduction of cloud computing has changed the provisioning and consumption of IT resources. Software-as-a-Service (SaaS) applications provide functionality using internet connectivity and enable centralised management of executable code resources. Multi-tenancy is an architectural pattern for sharing a single instance of executable software while isolating the data and business process serving each tenant. In a multi-tenant application, a tenant is a logical grouping of end-users who share common code, data and process instances. Multi-tenancy can cut across all the layers of an application software architecture. Conventionally, multi-tenancy is implemented as a flat structure. All tenants are treated in the same way by application software. However, large enterprises are hierarchically organised with defined boundaries between business and functional units. A SaaS solution, aimed at large enterprise users, needs to reflect these hierarchical boundaries to eliminate duplication of functional software but enforce data and business process separation. This paper introduces a novel hierarchical multi-tenancy architecture for an enterprise-scale business-to-business (B2B) cloudhosted software service application. Organisational hierarchy plays a vital role in grouping end users into tenants and subtenants in our software. A new reference architectural style for implementing hierarchical multi-tenant application is presented here for the first time. We use a case study approach to empirically evaluate the latency of this architecture in comparison to known flat multitenancy patterns. Our experimental evaluation supports the hypothesis that the hierarchical multi-tenancy approach meets the needs of our application and improves its performance.

High staff turnover has a negative impact on software development productivity and product quality. Further, offshore outsourcing has a widely held reputation for particularly poor employee retention. Interestingly, in-house sites (regardless of location) do not suffer such high levels of staff turnover.

A multitenant cloud-application that is designed to use several components needs to implement the required degree of isolation between the components when the workload changes. The highest degree of isolation results in high resource consumption and running cost per component. A low degree of isolation allows sharing of resources, but leads to degradation in performance and to increased security vulnerability. This paper presents a simulation-based approach operating on computational metaheuristics that search for optimal ways of deploying components of a cloud-hosted application to guarantee multitenancy isolation When the workload changes, an open multiclass Queuing Network model is used to determine the average number of component access requests, followed by a metaheuristic search for the optimal deployment solutions of the components in question. The simulation-based evaluation of optimization performance showed that the solutions obtained were very close to the target solution. Various recommendations and best practice guidelines for deploying components in a way that guarantees the required degree of isolation are also provided.

Multi-tenancy in cloud computing describes the extent to which resources can be shared while guaranteeing isolation among components (tenants) using these resources. There are three multi-tenancy patterns: shared, tenant-isolated and dedicated component patterns. These patterns have not previously been formally specified. In order to create a precise definition and verify each pattern, we formally specify each pattern using the Z language. To validate the interpretation of our formal description, We empirically evaluate each pattern using the data-tier of a cloud hosted distributed content management application, WordPress, deployed in a Docker container. Experimental results show that the dedicated pattern successfully managed larger numbers of tenants with fewer unhandled request errors. The shared and tenant isolated patterns exhibited larger number of unhandled request errors when the number of tenants increased. We present a selection algorithm to choose suitable multi-tenancy pattern for cloud deployment of content management system.

Globally distributed group projects are becoming an attractive and increasingly common feature in computer science education. They provide opportunities for students to engage in activities that enhance both their technical skills and wider professional competencies with concomitant benefits for graduate employability. There have been some previous attempts to investigate these projects in terms of theories of technology use and collaborative learning, and this paper continues this process by examining globally distributed group projects from the perspective of salient issues in the fields of computer-supported cooperative work (CSCW) and computer-supported collaborative learning (CSCL). After detailing CSCW models that discuss the dimensions that characterise interaction and technology use in groups, we examine aspects of group learning from the perspective of theories of CSCL. Issues of cooperation versus collaboration, motivation for learning and models of group cognition are discussed in the context of two specific group projects. Analysis of these examples allows us to characterise behaviour within groups and provide insights that can facilitate the formation and effective development of project teams. This has important educational implications for the success of these distributed group projects.

Universities face many challenges when creating opportunities for student experiences of global software engineering. We provide a model for introducing global software engineering into the computing curriculum. Our model is based on a three year collaboration between Robert Gordon University, UK and the International Institute for IT Bangalore, India. We provide evidence based on student feedback from three cohorts of virtual team who never met face to face. We found potential employers were supportive of global software engineering in university curricula. We identify four key principles for global software engineering student projects: reconcile contrasting assessment demands between institutions, create a detailed joint timetable to reconcile teaching calendars, provide a project management framework to support phased delivery and carefully manage project scope.

Software development has become increasingly globalized because of technological innovation, the evolution of work and business processes, as well as prevailing educational systems and national policies [1]. This paper addresses the challenges faced when creating realistic opportunities for students to participate in team working on a globally distributed software engineering project. We examine the experience of one set of students who undertook a single semester software development project, the technical objective of which was to create a product using online collaboration tools [2]. The students involved were studying at two widely-separated universities: Robert Gordon University (RGU), UK, and the International Institute for Information Technology, Bangalore (IIIT-B), India. A classroom-based action research approach has been used to evolve and reflect on the project [3]. Detailed student feedback was canvassed using an open-ended questionnaire. The students report favourably on acquisition of employability, collaboration and professional skills. A number of challenges remain in areas of project management, the use of collaborative technology and of scaling the project to larger classes, while retaining a manageable supervision overhead.

This paper explores practitioner descriptions of agile method tailoring in large-scale offshore or outsourced enterprise projects. Specifically, tailoring of the scrum master role is investigated. The scrum master acts as a facilitator for software development teams, nurturing adherence to agile practices and removing impediments for team members. But in large projects, scrum masters often work together in geographically distributed teams. Scrum masters use sprint planning to avoid development tasks that overlap team boundaries, coordinate status and effort across teams, and integrate code bases. The study comprises 8 international companies in London, Bangalore and Delhi. Interviews with 46 practitioners were conducted between February 2010 and May 2012. A grounded theory research method was used to identify that the scrum master role comprises six activities: process anchor, stand-up facilitator, impediment remover, sprint planner, scrum of scrums facilitator, and integration anchor. This systematic description of activities in scrum master teams extends our understanding of practitioner perspectives on agile process tailoring in large enterprises. Understanding these activities will help coaches guide large scale agile teams.

This paper explores practitioner descriptions of agile method tailoring in large-scale offshore or outsourced enterprise projects. Specifically, tailoring of the product owner role is discussed. The product owner identifies and prioritizes customer requirements. But in globalized projects, the product owner must reconcile large numbers competing business interests and generate prioritized requirements for many development teams. The study comprises 8 international companies in London, Bangalore and Delhi. Interviews with 46 practitioners were conducted between February 2010 and May 2012. A grounded theory approach was used to identify that product owner teams comprise nine roles: Groom, Prioritizer, Release Master, Technical Architect, Governor, Communicator, Traveler, Intermediary and Risk Assessor. These product owner roles arbitrate between conflicting customer requirements, approve release schedules, make architectural design decisions, provide technical governance and disseminate information across teams. Understanding these roles will help agile coaches guide large scale agile teams.

Resource limitations present several challenges to diffusion of ICT skills in developing (non-OECD) countries. Trainee teachers join colleges of teacher education with little or no prior exposure to technologies such as mobile phones or computers. Distinctive models are a required to help understanding of basic ICT skills learning in this context. A developmental model of basic ICT skills acquisition is presented. The model links learner skill level acquisition with pre-requisite infra-structural requirements called resource milestones. The model is applied to six out of the 22 teacher education institutions in Ethiopia and schools in and around the zonal capital Debre Birhan. It is found that none of the surveyed primary schools, Grade 1 to 8, use ICT in teaching and learning. A higher education preparatory school and two teacher training institutions have high speed Internet access. In the teacher training institutions the Internet is used to teach skills such as Web browsing to trainee teachers. The Skill Levels and Resource Milestones model provides a detailed roadmap for funding agencies and institutions to plan their investments. The Institutions surveyed have some way to go before students can develop more advanced information

Voting algorithms are used to arbitrate between the variant results in fault-tolerant systems. Traditional voters produce incorrect outputs in multiple error conditions. This paper introduces a class of voters, called predictor voters, which can resolve some multiple error conditions. These voters use analysis of a sequence of results in cyclic systems to select the most likely correct variant result as the voter output. Large discontinuities between successive results in cyclic systems are indicative of faults. The voting algorithms have the effect of filtering discontinuities to improve availability. Three different versions of predictor voters are described. Fault-injection simulation tests are used to investigate their safety and availability performance in triple error scenarios. Experimental results show that predictor voters give safety behaviour between majority and median voters. Predictor voters with order three and above give higher availability than the median voter. Predictor voters are suitable for use in systems in which some incorrect outputs can be tolerated in order to maintain functionality over a long period of time.

Voting algorithms are used to provide an error masking capability in a wide range of highly dependable commercial & research applications. These applications include N-Modular Redundant hardware systems and diversely designed software systems based on N-Version Programming. The most sophisticated & complex algorithms can even tolerate malicious (or Byzantine) subsystem errors. The algorithms can be implemented in hardware or software depending on the characteristics of the application, and the type of voter selected. Many voting algorithms have been defined in the literature, each with particular strengths and weaknesses. Having surveyed more than 70 references from the literature, a functional classification is used in this paper to provide taxonomy of those voting algorithms used in safety-critical applications. We classify voters into three categories: generic, hybrid, and purpose-built voters. Selected algorithms of each category are described, for illustrative purposes, and application areas proposed. Approaches to the comparison of algorithm behavior are also surveyed. These approaches compare the acceptability of voter behavior based on either statistical considerations (e.g., number of successes, number of benign or catastrophic results), or probabilistic computations (e.g., probability of choosing correct value in each voting cycle or average mean square error) during q voting cycles.