No description available.

The automotive industry is continuously innovating and adapting new technologies. Along with that, the companies work towards maintaining the quality of a hardware product and meeting the customer demands. Before delivering the product to the customer, it is essential to test and approve it for the safe use. The concept remains the same when it comes to a software. Adapting modern technologies will further improve the efficiency of testing a software. The thesis aims to build a machine learning algorithm for the implementation during the software testing. In general, the evaluation of a generated test report after the testing consumes more time. The built algorithm should be able to reduce the time spent and the manual effort during the evaluation. Basically, the machine learning algorithms will analyze and learn the data available in the old test reports. Based on the learnt data pattern, it will suggest the possible root causes for the failed test cases in the future. The thesis report has the literature survey that helped in understanding the machine learning concepts in different industries for similar problems. The tasks involved while building the model are data loading, data pre-processing, selecting the best conditions for each algorithm and comparison of the performance among them. It also suggest the possible future work towards improving the performance of the models. The entire work is implemented in Jupyter notebook using pandas and scikit-learn libraries.

<p dir="ltr">Deep Neural Networks (DNNs) have been deployed on edge devices for numerous applications, ranging from computer vision, speech recognition, and anomaly detection. When deployed on edge devices, dedicated DNN compilers are used to compile DNNs into binaries to exploit instruction set architectures’ (ISAs’) features and hardware accelerators (e.g., NPU, GPU). These DNN binaries on edge devices process sensitive user information, conduct critical missions, and are considered confidential intellectual property.</p><p dir="ltr">From the security standpoint, the ability to reverse engineer such binaries (i.e., recovering the original, high-level representation of the implemented DNN) enables several applications, such as DNN models stealing, gray/white-box adversarial machine learning attacks and defenses, and backdoor detection. However, no existing reverse engineering technique can recover a high-level representation of a DNN model from its compiled binary code.</p><p dir="ltr">In this dissertation, we propose the following pioneering research for reverse engineering DNN on the edge device. (i) We design and implement the first compiler- and ISA-agnostic DNN decompiler, DnD, with the static analysis technique, capable of extracting DNN models from DNN binaries running on CPU-only devices without the hardware accelerator. We show that our decompiler can perfectly recover DNN models from different DNN binaries. Furthermore, it can extract DNN models used by real-world micro-controllers and enable white-box adversarial machine learning attacks against the DNN models. (ii) We design and implement a novel data-driven approach, NeuroScope, based on dynamic analysis and machine learning to reverse engineer DNN binaries. This compiler-independent and code-feature-free approach supports a larger variety of DNN binaries across different DNN compilers and hardware platforms. We demonstrate its capability by using it to reverse engineer DNN binaries unsupported by previous approaches with high accuracy. Moreover, we showcase how NeuroScope can be used to reverse engineer a proprietary DNN binary compiled with a closed-source compiler and enable gray-box adversarial machine learning attacks.</p>

Broadly, agile software development is an approach where code is frequently built, tested and shipped, leading to short release cycles. Extreme version is the DevOps approach where the development, testing and deployment pipelines are merged and software is continuously tested and updated. In this context our work focuses on identifying spots where the participants should be more aware of the performance and offers approaches and tools to improve their awareness with the ultimate goal of producing better software in shorter time. In general, the awareness is raised by testing, documenting, and monitoring the performance in all phases of the development cycle. In this thesis we (1) show a framework for writing performance tests for individual components (e.g. libraries). The tests capture and codify assumptions about the performance into runnable artifacts that simplify repeatability and automation. For evaluation of the performance tests we (2) propose new methods, which can automatically detect performance regressions. These methods are designed with inherent variation of performance data in mind and are able to filter it out in order to detect true regressions. Then we (3) reuse the performance tests to provide the developers with accurate and up-to-date performance API documentation that steer them...

Context: Software testing is expensive, labor intensive and consumes lot of time in a software development life cycle. There was always a need in software testing to decrease the testing time. This also resulted to focus on Automated Software Testing (AST), because using automated testing, with specific tools, this effort can be dramatically reduced and the costs related with testing can decrease [11]. Manual Testing (MT) requires lot of effort and hard work, if we measure in terms of person per month [11]. Automated Software testing helps to decrease the work load by giving some testing tasks to the computers. Computer systems are cheap, they are faster and don‘t get bored and can work continuously in the weekends. Due to this advantage many researches are working towards the Automation of software testing, which can help to complete the task in less testing time [10]. Objectives: The main aims of this thesis is to 1.) To systematically classify contributions within AST. 2.) To identify the different benefits and challenges of AST. 3.) To identify the whether the reported benefits and challenges found in the literature are prevalent in industry. Methods: To fulfill our aims and objectives, we used Systematic mapping research methodology to systematically classify contributions within AST. We also used SLR to identify the different benefits and challenges of AST. Finally, we performed web based survey to validate the finding of SLR. Results: After performing Systematic mapping, the main aspects within AST include purpose of automation, levels of testing, Technology used, different types of research types used and frequency of AST studies over the time. From Systematic literature review, we found the benefits and challenges of AST. The benefits of AST include higher product quality, less testing time, reliability, increase in confidence, reusability, less human effort, reduction of cost and increase in fault detection. The challenges include failure to achieve expected goals, difficulty in maintenance of test automation, Test automation needs more time to mature, false expectations and lack of skilled people for test automation tools. From web survey, it is observed that almost all the benefits and challenges are prevalent in industry. The benefits such as fault detection and confidence are in contrary to the results of SLR. The challenge about the appropriate test automation strategy has 24 % disagreement from the respondents and 30% uncertainty. The reason is that the automation strategy is totally dependent on the test manager of the project. When asked “Does automated software testing fully replace manual testing”, 80% disagree with this challenge. Conclusion: The classification of the AST studies using systematic mapping gives an overview of the work done in the area of AST and also helps to find research coverage in the area of AST. These results can be used by researchers to use the gaps found in the mapping studies to carry on future work. The results of SLR and web survey clearly show that the practitioners clearly realize the benefits and challenges of AST reported in the literature. / Mobile no: +46723069909

<p>Static program analyzers are vital tools to produce useful insights about programs without executing these programs. These insights can be used to improve the quality of programs, e.g., detecting defects in programs, or optimizing programs to use fewer resources. However, building static program analyzers that are simultaneously sound, performant, and flexible is notoriously challenging.</p> <p>This dissertation aims to address this challenge by exploring the potential of applying correct-by-construction metaprogramming techniques to build static program analyzers. Metaprogramming techniques manipulate and transform programs as data objects.  In this thesis, we consider static program analyzers as the objects to be manipulated or transformed. We show that metaprogramming techniques can improve our understanding, the construction, flexibility, and performance of program analyzers.</p> <p>We first study the inter-derivation of abstract interpreters. Using off-the-shelf program transformation techniques such as refunctionalization, we demonstrate that big-step abstract interpreters can be mechanically derived from their small-step counterparts, thus building a functional correspondence between two different styles of abstract interpretation.</p> <p>To build high-performance program analyzers, we exploit the first Futamura projection to build compilers for abstract interpretation and symbolic execution. The first Futamura projection states that specializing an interpreter with respect to an input program is a process equivalent to compilation, thus providing a practical way to repurpose interpreters for compilation and code generation. We systematically apply this idea to build program-analysis compilers by writing analyzers as staged interpreters using higher-level abstractions. The staged interpreter can be used for generating sound and performant analysis code given a specific input program. Moreover, the approach enables using abstractions without regret: by using higher-level program abstractions, the analyzer can be written in a way that is close to its high-level specification (e.g. big-step operational semantics), and by compilation, the analyzer is performant since it does not need to pay the runtime overhead of using these abstraction mechanisms.</p> <p>We also develop novel type systems that track sharing and separation in higher-order imperative languages. Such type systems are useful both for general-purpose programming languages and for optimization of domain-specific metaprograms such as those program-analysis compilers.</p> <p><br></p>

<p dir="ltr">Proof assistants based on dependent types are powerful tools for building certified software. In order to verify programs written in a different language, however, a representation of those programs in the proof assistant is required. When that language is sufficiently similar to that of the proof assistant, one solution is to use a <i>shallow embedding</i> to directly encode source programs as programs in the proof assistant. One challenge with this approach is ensuring that any semantic gaps between the two languages are accounted for. In this thesis, we present <i>GSet</i>, a mixed embedding that bridges the gap between OCaml GADTs and inductive datatypes in Coq. This embedding retains the rich typing information of GADTs while also allowing pattern matching with impossible branches to be translated without additional axioms. We formalize this with GADTml, a minimal calculus that captures GADTs in OCaml, and gCIC, an impredicative variant of the Calculus of Inductive Constructions. Furthermore, we present the translation algorithm between GADTml and gCIC, together with a proof of the soundness of this translation. We have integrated this technique into coq-of-ocaml, a tool for automatically translating OCaml programs into Coq. Finally, we demonstrate the feasibility of our approach by using our enhanced version of coq-of-ocaml, to translate a portion of the Tezos code base into Coq.</p>

Die Dissertationsschrift diskutiert die Machbarkeit einer modularen, portierbaren Middleware für die automatisierte Ausführung und Dokumentation von Software-Tests mit einer durchgehenden Nachverfolgbarkeit von der Anforderungsspezifikation bis hin zur Dokumentation der Testergebnisse. Es werden die Eigenschaften und Probleme bestehender Testautomatisierungslösungen analysiert und dargelegt. Unter Berücksichtigung dieser Probleme werden neuartige Lösungsansätze entwickelt. Die Neuheiten dieser Arbeit sind der modulare Aufbau der Middleware mit einer unproblematischen Portierbarkeit auf neue Testsysteme in Verbindung mit dem neu erarbeiteten Speicherformat für die Testergebnisse. Es wird die Möglichkeit aufgezeigt, Testfälle sowohl mit graphischer als auch textueller Eingabe zu bearbeiten. Neben den typischen Einsatzbereichen, wie zum Beispiel Hardware In The Loop-Tests (HIL), werden auch weitere Felder, vom Modul-Test bis zum Bandende-Test, abgedeckt. Das Speicherformat der Testergebnisse ermöglicht die Ablage aller wichtigen Informationen zu den Tests, ist flexibel erweiterbar und erlaubt die Generierung von Testreports in unterschiedlichen Zielformaten. Ein weiterer zentraler Punkt ist der automatisierte Austausch von Informationen und Testergebnissen mit verschiedenen Requirementsmanagement-Systemen sowie eine nahtlose Integration in vorhandene Versionsmanagement-Systeme. Basierend auf den theoretischen Ausarbeitungen wurde eine modulare, portierbare Middleware in Form des modularen aufgebauten Testautomatisierungs-Frameworks (modTF) umgesetzt. Anhand der dabei gesammelten Erfahrungen und der Ergebnisse der praktischen Erprobung werden die Vorteile des Frameworks gezeigt. / These PhD discusses the feasibility of a modular, portable middleware for automated execution and documentation of software tests with a continuous traceability from the requirements to the test results. The properties and the problems of existing test automation solutions are analyzed and presented. Based on the problems novel solutions are developed. The novelties of this PhD are the modular structure of the middleware with the easy portability to new test systems in cooperation with the novel storage format for the test results. The possibility for a graphical and textual description of the test cases is shown. Beside the typical applications like Hardware In The Loop tests (HIL) also applications from the module test to the line end test are include. The storage format for the results allows the storage of all needed information according to the test cases in one file. The format is flexible and extendable. The generation of test reports in different target formats is possible. Another imported point is the automated exchange of information and test results with different requirements management systems and the seamless integration in existing version management systems. By means of the collected experiences and the results of the practical proving the advantages of the framework are shown.

Made available in DSpace on 2014-12-17T15:47:58Z (GMT). No. of bitstreams: 1 HeitorMAC_DISSERT.pdf: 3258229 bytes, checksum: 5f7856b140a636bd052147c58ff9dede (MD5) Previous issue date: 2011-03-01 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior / Through the adoption of the software product line (SPL) approach, several benefits are achieved when compared to the conventional development processes that are based on creating a single software system at a time. The process of developing a SPL differs from traditional software construction, since it has two essential phases: the domain engineering - when common and variables elements of the SPL are defined and implemented; and the application engineering - when one or more applications (specific products) are derived from the reuse of artifacts created in the domain engineering. The test activity is also fundamental and aims to detect defects in the artifacts produced in SPL development. However, the characteristics of an SPL bring new challenges to this activity that must be considered. Several approaches have been recently proposed for the testing process of product lines, but they have been shown limited and have only provided general guidelines. In addition, there is also a lack of tools to support the variability management and customization of automated case tests for SPLs. In this context, this dissertation has the goal of proposing a systematic approach to software product line testing. The approach offers: (i) automated SPL test strategies to be applied in the domain and application engineering, (ii) explicit guidelines to support the implementation and reuse of automated test cases at the unit, integration and system levels in domain and application engineering; and (iii) tooling support for automating the variability management and customization of test cases. The approach is evaluated through its application in a software product line for web systems. The results of this work have shown that the proposed approach can help the developers to deal with the challenges imposed by the characteristics of SPLs during the testing process / Com o uso da abordagem de linhas de produto de software (LPSs), v?rios benef?cios s?o alcan?ados quando comparados aos processos de desenvolvimento convencionais que se baseiam na cria??o de um ?nico sistema por vez. O processo de desenvolvimento de uma LPS se diferencia da constru??o tradicional de software, uma vez que apresenta duas etapas essenciais: a engenharia de dom?nio - quando elementos comuns e vari?veis da LPS s?o definidos e implementados; e a engenharia de aplica??o quando uma ou mais aplica??es (produtos espec?ficos) s?o derivadas a partir do reuso dos artefatos criados na engenharia de dom?nio. Durante a elabora??o da LPS, assim como no desenvolvimento convencional de sistemas, a atividade de teste ? fundamental e tem como objetivo a detec??o de defeitos nos artefatos produzidos. Contudo, as caracter?sticas de uma LPS trazem novos desafios a essa atividade e que precisam ser considerados. Diversas abordagens foram propostas para o processo de teste de linhas de produto, mas elas se mostram limitadas ou fornecem diretrizes muito gerais. Outro fator preocupante ? a escassez de ferramentas que auxiliem na implementa??o, aplica??o e acompanhamento dos testes, bem como na ger?ncia e customiza??o de tais artefatos. Com base nesse contexto relacionado ao processo de teste de LPSs, esta disserta??o tem como objetivo propor uma abordagem sistem?tica para o teste de linhas de produto de software. A abordagem oferece: (i) estrat?gias de testes automatizados para LPSs tanto na engenharia de dom?nio quanto de aplica??o; (ii) diretrizes para a implementa??o e reuso de casos de teste automatizados nos n?veis de unidade, integra??o e sistema tanto para a engenharia de dom?nio quanto de aplica??o; e (iii) suporte ferramental para ger?ncia e customiza??o autom?tica de casos de teste usando t?cnicas de deriva??o autom?tica de software. A abordagem ? avaliada atrav?s da sua aplica??o em uma linha de produto para sistemas web. Os resultados deste trabalho mostram que a abordagem proposta pode ajudar os desenvolvedores a lidar com os desafios impostos pelas caracter?sticas das LPSs durante o processo de testes

Made available in DSpace on 2014-12-17T15:48:03Z (GMT). No. of bitstreams: 1 AlexandreSGV_DISSERT.pdf: 2732563 bytes, checksum: ab07c81d7e941ed2d721415180865feb (MD5) Previous issue date: 2012-05-25 / Conselho Nacional de Desenvolvimento Cient?fico e Tecnol?gico / Software Products Lines (SPL) is a software engineering approach to developing software system families that share common features and differ in other features according to the requested software systems. The adoption of the SPL approach can promote several benefits such as cost reduction, product quality, productivity, and time to market. On the other hand, the SPL approach brings new challenges to the software evolution that must be considered. Recent research work has explored and proposed automated approaches based on code analysis and traceability techniques for change impact analysis in the context of SPL development. There are existing limitations concerning these approaches such as the customization of the analysis functionalities to address different strategies for change impact analysis, and the change impact analysis of fine-grained variability. This dissertation proposes a change impact analysis tool for SPL development, called Squid Impact Analyzer. The tool allows the implementation of change impact analysis based on information from variability modeling, mapping of variability to code assets, and existing dependency relationships between code assets. An assessment of the tool is conducted through an experiment that compare the change impact analysis results provided by the tool with real changes applied to several evolution releases from a SPL for media management in mobile devices / Linhas de Produtos de Software (LPS) consiste em um paradigma de desenvolvimento de software, no qual fam?lias de sistemas compartilham caracter?sticas comuns e tornam expl?citas outras caracter?sticas que variam de acordo com o sistema final sendo considerado. Esta abordagem oferece benef?cios ao desenvolvimento de software como redu??o de custos, qualidade do produto final, produtividade e tempo de desenvolvimento reduzido. Por outro lado, a abordagem imp?e novos desafios para a atividade de evolu??o dos artefatos que modelam e implementam a LPS. Trabalhos de pesquisa recentes prop?em abordagens com suporte automatizado de ferramentas de an?lise de impacto de mudan?a no contexto de evolu??o de LPSs. Tais abordagens s?o baseadas em t?cnicas de an?lise de impacto de mudan?as e rastreabilidade de artefatos, por?m apresentam limita??es quanto ? an?lise de impacto de mudan?as em variabilidades de granularidade fina, bem como ? customiza??o dos tipos e estrat?gias de an?lise realizadas. Esta disserta??o prop?e uma ferramenta de an?lise de impacto de mudan?a, denominada Squid Impact Analyzer, que utiliza uma estrat?gia de estimativa de impacto baseada em informa??es de caracter?sticas, mapeamento de tais caracter?sticas em artefatos de c?digo, e depend?ncia existente entre artefatos de implementa??o. A ferramenta ? avaliada atrav?s da condu??o de experimentos que realizam a quantifica??o de m?tricas de cobertura, precis?o e m?dia harm?nica nos resultados de buscas de an?lise de impacto de mudan?a da ferramenta proposta em contraposi??o ?s mudan?as reais realizadas nos artefatos de diversas vers?es de evolu??o de uma LPS para gerenciamento de m?dias em dispositivos m?veis. A ferramenta foi desenvolvida com base em uma infraestrutura que serve de base para a instancia??o de ferramentas de an?lise de propriedades de c?digo de LPSs, e que ? tamb?m parte da contribui??o da disserta??o