is study has focused on the technical sustainability dimension of sustainable soŸware engineering. Sustainabilitywithin soŸware engineering is something that has recently started to gain traction, where most of theresearch has focused on the environmental dimension.e purpose of this study was to investigate the technicalsustainability dimension, and contribute by continuing to build upon previous research which has suggestedsustainability as a part of non-functional requirements and as a part of soŸware quality.is study was conducted with a qualitative approach. is approach used interviews with people withexperience of working in soŸware development to collect data. e interviews were used to gain in-depthinformation of the dišerent aspects in soŸware development, e.g. development processes.e results, based on the interviews, showed that from a technical sustainability perspective there are issueswith where the focus of the work is in soŸware development and also problems related to how timemanagementis used.e results also showed that there was some lacking in understanding the benets and tradeošs ofdecisions made in the project, for example regarding the soŸware design, which can result in unnecessarycomplexity is acquired.e study found that the culture within soŸware development projects and companies needs to be changedtowards a culture which promotes the development of high quality soŸware. Because non-functional requirementsand soŸware quality are components of technical sustainability, therefore is the technical sustainabilityhigher if the quality is high. It was further found that this culture would be possible if changes were made tothe way prioritisations are made, and the way testing is performed should be extended. Finally it was foundthat the understanding of the benets and tradeošs of decisions and ideas needs to be increased and extendedcollectively.e study concluded that the technical sustainability dimension can be achieved through these changes andthat the problem and solution therefore are not small and local but rather a big and complex issue where thesolution spans several solutions that encompasses dišerent aspects and perspectives.

Fault detection is a key component to minimizing service unavailability. Fault detection is generally handled by a monitoring system. This project investigates the possibility of extending an existing monitoring system to alert based on anomalous patterns in time series.The project was broken up into two areas. The first area conducted an investigation whether it is possible to alert based on anomalous patterns in time series. A hypothesis was formed as follows; forecasting models cannot be used to detect anomalous patterns in time series. The investigation used case studies to disprove the hypothesis. Each case study used a forecasting model to measure the number of false, missed and correctly predicted alarms to determine if the hypothesis was disproved.The second area created a design for the extension. An initial design of the system was created. The design was implemented and evaluated to find improvements. The outcome was then used to create a general design.The results from the investigation disproved the hypothesis. The report also presents a general software design for an anomaly detection system. / Feldetektering är en nyckelkomponent för att minimera nedtid i mjukvarutjänster. Feldetektering hanteras vanligtvis av ett övervakningssystem. Detta projekt undersöker möjligheten att utöka ett befintligt övervakningssystem till att kunna skicka ut larm baserat på avvikande mönster i tidsserier.Projektet bröts upp i två områden. Det första området genomförde en undersökning om det är möjligt att skicka ut larm baserat på avvikande mönster i tidsserier. En hypotes bildades enligt följande; prognosmodeller kan inte användas för att upptäcka avvikande mönster i tidsserier. Undersökningen använde fallstudier till att motbevisa hypotesen. Varje fallstudie använde en prognosmodell för att mäta antalet falska, missade och korrekt förutsedda larm. Resultaten användes sedan för att avgöra om hypotesen var motbevisad.Det andra området innefattade skapadet av en mjukvarudesign för utökning av ett övervakningssystem. En initial mjukvarudesign av systemet skapades. Mjukvarudesignen implementerades sedan och utvärderades för att hitta förbättringar. Resultatet användes sedan för att skapa en generell design. Resultaten från undersökningen motbevisade hypotesen. Rapporten presenterar även en allmän mjukvarudesign för ettanomalitetsdetekteringssystem.

Nowadays it is more clear that the Internet of things (IoT) is not a transient trend but a completely new industry. The internet of things has the capability to enhance current industries (Industry 4.0), as well as to help protecting the environment and people. The latter is the case with the system developed and described in this thesis. The possibilities that IoT brings are due to the interconnection of heterogeneous embedded devices to the internet. This thesis focus on LPWANs (Low Power Wide Area Networks), which is a new set of technologies specifically design for the needs of IoT devices.Due to the recent deploy of NB-IoT (Narrow Band IoT) networks it has become more difficult to know what LPWAN is best for a certain application. Thus, the first half of this thesis involves the comparative study of NB-IoT and LoRaWAN LPWANs. This comparison required an in depth study of each technology, specially on the physical and datalink layers. The comparison briefly displays the main characteristics of each technology and explain the main conclusions in a concise manner. The second part of the thesis describes the development of a GNSS tracker. This tracker will be used on train wagons carrying goods that are dangerous for people and the environment. This thesis report describes the different steps taken, from the requirement specification to the partial development of the software.

[pt] Sistemas impactados por Problemas de Projeto (PPs) podem se tornar difíceis de manter e evoluir. A identificação de PPs pode ocorrer por meio de múltiplos sintomas, tais como code smells. Após tal identificação, pode-se remover os PPs por meio de refatorações. No entanto, decidir onde e como refatorar é uma tarefa desafiadora. Assim, técnicas de recomendação de refatoração têm sido propostas. Apesar disso, ainda há pouco consenso sobre quais requisitos devem ser atendidos por elas. Nesta tese, estamos propondo quatro requisitos empiricamente identificados que tais técnicas devem seguir. Primeiro, cada PP geralmente está relacionado a vários tipos de sintomas no código-fonte e eles devem ser considerados juntos para gerar recomendações. Além disso, uma técnica de recomendação deve permitir a seleção de contextos específicos para refatoração. Quarto, também deve-se considerar as funcionalidades modificadas para criar recomendações úteis. Finalmente, os desenvolvedores nem sempre conduzem as refatorações mais eficazes na prática, muitas vezes inconscientemente, resultando na remoção incompleta de PPs. Assim, eles precisam de assistência para remover os PPs. Existem apenas técnicas que atendem parcialmente aos requisitos mencionados. Sendo assim, nós propomos a técnica OrganicRef. OrganicRef destina-se a ajudar os desenvolvedores na remoção de PPs em seus contextos de interesse. OrganicRef encontra as funcionalidades dos elementos de código usando um algoritmo de modelagem de tópicos. Em seguida, ele coleta múltiplos tipos de sintomas que afetam os elementos do código. Para recomendar refatorações, OrganicRef combina heurísticas baseadas em regras e baseadas em funcionalidades. OrganicRef também aplica otimização baseada em busca para derivar várias recomendações possíveis. Para avaliar o OrganicRef, realizamos um estudo experimental com seis projetos de software. Os resultados mostraram que as recomendações do OrganicRef melhoram significativamente a qualidade dos elementos refatorados. / [en] Software projects impacted by Design Problems (DPs) may become difficult to maintain and evolve. The identification of DPs may occur through symptoms such as code smells. After such identification, developers can remove the DPs through refactorings. However, deciding where and how to refactor is a challenging task. Thus, several refactoring recommendation techniques have been proposed. Nevertheless, there is still little consensus on which requirements must be satisfied by them. In this thesis, we are proposing four empirically identified requirements that any DP removal technique should follow. First, each single DP is usually related with multiple types of symptoms in the source code and they should be considered altogether for generating recommendations. Second, a recommendation technique should allow the selection of possible candidate contexts for refactoring. Fourth, the technique should consider the features of undergoing changes to create useful recommendations. Finally, developers do not always conduct the most effective refactorings in practice, quite often unconsciously, resulting in the incomplete removal of DPs. Thus, they need assistance to remove DPs. There are techniques partially fulfilling the aforementioned requirements, though none satisfactorily meets them all. Thus, we propose the OrganicRef technique. OrganicRef is intended to help developers in removing DPs in their contexts of interest. OrganicRef finds the contexts by capturing the features affecting relevant code elements using a topic modeling algorithm. Then, it collects multiple symptom types affecting the code elements. To recommend effective refactorings, OrganicRef combines rulebased and feature-driven heuristics. It also uses search-based optimization to derive multiple possible recommendations. To evaluate OrganicRef, we conducted an empirical study with six open source projects. Results showed that OrganicRef recommendations significantly improves the design of refactored elements.

La recherche en apprentissage automatique peut être vue comme une quête vers l’aboutissement d’algorithmes d’apprentissage de plus en plus généraux, applicable à des problèmes de plus en plus réalistes. Selon cette perspective, le progrès dans ce domaine peut être réalisé de deux façons: par l’amélioration des méthodes algorithmiques associées aux problèmes existants, et par l’introduction de nouveaux types de problèmes. Avec le progrès marqué du côté des méthodes d’apprentissage machine, une panoplie de nouveaux types de problèmes d’apprentissage ont aussi été proposés, où les hypothèses de problèmes existants sont assouplies ou généralisées afin de mieux refléter les conditions du monde réel. Le domaine de l’apprentissage en continu (Continual Learning) est un exemple d’un tel domaine, où l’hypothèse de la stationarité des distributions encourues lors de l’entrainement d’un modèles est assouplie, et où les algorithmes d’apprentissages doivent donc s’adapter à des changements soudains ou progressifs dans leur environnement. Dans cet ouvrage, nous introduisons les hiérarchiées de problèmes, une application du concept de hiérarchie des types provenant des sciences informatiques, au domaine des problèmes de recherche en apprentissage machine. Les hierarchies de problèmes organisent et structurent les problèmes d’apprentissage en fonction de leurs hypothéses. Les méthodes peuvent donc définir explicitement leur domaine d’application, leur permettant donc d’être partagées et réutilisées à travers différent types de problèmes de manière polymorphique: Une méthode conçue pour un domaine donné peut aussi être appli- quée à un domaine plus précis que celui-ci, tel qu’indiqué par leur relation dans la hierarchie de problèmes. Nous démontrons que ce système, lorsque mis en oeuvre, comporte divers bienfaits qui addressent directement plusieurs des problèmes encourus par les chercheurs en apprentissage machine. Nous démontrons la viabilité de ce principe avec Sequoia, une infrastructure logicielle libre qui implémente une hierarchie des problèmes en apprentissage continu. Nous espérons que ce nouveau paradigme, ainsi que sa première implémentation, pourra servir à unifier et accélérer les divers efforts de recherche en apprentissage continu, ainsi qu’à encourager des efforts similaires dans d’autres domaines de recherche. Vous pouvez nous aider à faire grandir l’arbre en visitant github.com/lebrice/Sequoia. / Research in Machine Learning (ML) can be viewed as a quest to develop increasingly general algorithmic solutions (methods) for increasingly challenging research problems (settings). From this perspective, progress can be realized in two ways: by introducing better methods for current settings, or by proposing interesting new settings for the research community to solve. Alongside recent progress in methods, a wide variety of research settings have also been introduced, often as variants of existing settings where underlying assumptions are removed to make the problem more realistic or general. The field of Continual Learning (CL), for example, consists of a family of settings where the stationarity assumption is removed, and where methods as a result have to learn from environments or data distributions that can change over time. In this work, we introduce the concept of problem hierarchies: hierarchical structures in which research settings are systematically organized based on their assumptions. Methods can then explicitly state their assumptions by selecting a target setting from this hierarchy. Most importantly, these structures make it possible to easily share and reuse research methods across different settings using inheritance, since a method developed for a given setting is also directly applicable onto any of its children in the hierarchy. We argue that this simple mechanism can have great implications for ML research in practice. As a proof-of-concept of this approach, we introduce Sequoia, an open-source research framework in which we construct a hierarchy of the settings and methods in CL. We hope that this new paradigm and its first implementation can help unify and accelerate research in CL and serve as inspiration for future work in other fields. You can help us grow the tree by visiting github.com/lebrice/Sequoia.

No description available.

No description available.

No description available.

[pt] Uma Linha de Produtos de Sistemas Multi-agentes (LP-SMA) define uma arquitetura de linha de produto de software (LPS) cujo projeto e implementação são realizados usando agentes de software para compreender features comuns e variáveis. O desenvolvimento de LP-SMA pode ser feito através de plataformas específicas de SMA e técnicas de implementação. Exemplos de tais técnicas são: frameworks orientados a objetos, compilação condicional, arquivos de configuração e programação orientada a aspectos (POA). Todavia, os estudos empíricos existentes não focam na abordagem de LP-SMA, considerando diferentes técnicas de implementação e plataformas específicas de desenvolvimento de SMA. Nesse contexto, este trabalho apresenta uma comparação sistemática de diferentes técnicas de implementação de variabilidades de features de agentes no domínio de LP-SMA. Esta comparação sistemática envolveu o uso de duas plataformas de desenvolvimento de SMA (JADE e Jadex) e das técnicas de implementação: compilação condicional, arquivos de configuração e POA. Neste estudo, um conjunto de métricas de software foram utilizadas para avaliar atributos de qualidade, tais como modularidade e estabilidade. Para a realização deste estudo, duas LP-SMAs foram desenvolvidas. A primeira foi a LP-SMA do Expert Committee, uma linha de produto de sistemas de gerenciamento de conferências. A segunda foi a LP-SMA OLIS, que provê diversos serviços pessoais para os usuários. Os dados coletados durante a realização dos estudos empíricos permitiram o relato de um conjunto de lições aprendidas. / [en] A Multi-agent System Product Line (MAS-PL) defines a Software Product Line (SPL) architecture whose design and implementation are accomplished using software agents to address its common and variable features. The MAS-PL development can be performed through MAS specific platforms and implementation techniques. Examples of such techniques are: object-oriented frameworks, conditional compilation, configuration files and aspect-oriented programming (AOP). However, the existing empirical studies do not focus on MAS-PL approach, considering different implementation techniques and MAS specific platforms. In this context, this work presents a systematic comparison of different variability implementation techniques of agent features in the MAS-PL domain. This systematic comparison involved the use of two platforms of MAS development (JADE and Jadex) and implementation techniques conditional compilation, configuration files and AOP. In this study, a suite of software metries were used to evaluate quality attributes, such as modularity and stability. In order to perform this study, two MAS-PLs were developed. The first one was the Expert Commitee MAS-PL, a product line of conference management systems. The second one as the OLIS MAS-PL, which provides several personal services to the users. The collected data during the accomplished empirical studies allowed to report a set of lessons learned.

Made available in DSpace on 2016-08-17T14:52:46Z (GMT). No. of bitstreams: 1 Geovane bezerra.pdf: 1597415 bytes, checksum: 1a51bd2c6aed9c1c6ee46b8369083350 (MD5) Previous issue date: 2003-03-20 / This work proposes a collection of architectural patterns for the development of multi-agent systems. Main architectural concerns like communication, cooperation and coordination mechanisms between the agents of the society are particularly analyzed in each described pattern. The generation of new patterns through the extension or composition of the proposed ones is also approached. The methodology for pattern extraction is based on the study of frequently used architectures of multi-agent systems, like blackboard and federative architectures, and on mechanisms of cooperation and coordination usually identified in such architectures, like master-slave, meeting and negotiator mechanisms, as well. Pattern description is based on AUML and KQML. Main AUML diagrams, like agent packages, agent diagrams and agent interaction diagrams are used to represent the structure and behavior of the society. In those diagrams, agent interactions are represented as KQML performatives. Proposed patterns are validated through the construction of three case studies related to the development of multi-agent systems for information retrieval and filtering. In these case studies, the architectures RETSINA, AMALTHAEA and ABARFI are analyzed and the reuse or potential application of proposed patterns is identified. / Este trabalho propõe uma coleção de padrões arquiteturais para o desenvolvimento de sistemas multiagente. Os principais problemas arquiteturais tais como comunicação, cooperação e mecanismo de coordenação entre os agentes são analisados e descritos em cada padrão. A geração de novos padrões pela extensão ou composição dos padrões propostos também é abordada. A metodologia utilizada para extração de padrões está baseada no estudo de arquiteturas de sistemas multiagente freqüentemente utilizadas, como as arquiteturas quadro-negro e federativas e nos mecanismos de cooperação e coordenação geralmente identificados em tais arquiteturas, como os mecanismos mestre-escravo, reunião e negociador. A descrição dos padrões está baseada em AUML e KQML. Os principais diagramas da AUML, como o diagrama de pacotes, o diagrama de agente e o diagrama de interação são usados para representar a estrutura e o comportamento da sociedade. Nesses diagramas, a representação das interações entre os agentes é feita com a utilização das performatives de KQML. Os padrões propostos são validados através da construção de três estudos de caso relacionados ao desenvolvimento de sistemas multiagente para recuperação e filtragem de informações. Nesses estudos de caso, são analisadas as arquiteturas RETSINA, AMALTHAEA e ABARFI e é identificado a potencial reutilização ou aplicação dos padrões propostos.