Search

Improving road safety is a major concern for most car manufacturers. In recent years, the development of Advanced Driver Assistance Systems (ADAS) has subsequently seen a tremendous boost. The development of such systems requires complex testing to ensure vehicle’s safety and reliability. Performing road tests tends to be dangerous, time-consuming, and costly. Hence, a large part of testing for ADAS has to be carried out using physics-based simulation platforms, which are able to emulate a wide range of virtual traffic scenarios and road environments. The main difficulties with simulation-based testing of ADAS are: (1) the test input space is large and multidimensional, (2) simulation platforms provide no guidance to engineers as to which scenarios should be selected for testing, and hence, simulation is limited to a small number of scenarios hand-picked by engineers, and (3) test executions are computationally expensive because they often involve executing high-fidelity mathematical models capturing continuous dynamic behaviors of vehicles and their environment. The complexity of testing ADAS is further exacerbated when many ADAS are employed together in a self-driving system. In particular, when self-driving systems include many ADAS (i.e., features), they tend to interact and impact one another’s behavior in an unknown way and may lead to conflicting situations. The main challenge here is to detect and manage feature interactions, in particular, those that violate system safety requirements, hence leading to critical failures. In practice, once feature interaction failures are detected, engineers need to devise resolution strategies to resolve potential conflicts between features. Developing resolution strategies is a complex task and despite the extensive domain expertise, these resolution strategies can be erroneous and are too complex to be manually repaired. In this dissertation, in addition to testing individual ADAS, we focus on testing self-driving systems that include several ADAS. In this dissertation, we propose a set of approaches based on meta-heuristic search and machine learning techniques to automate ADAS testing and to repair feature interaction failures in self-driving systems. The work presented in this dissertation is motivated by ADAS testing needs at IEE, a world-leading part supplier to the automotive industry. In this dissertation, we focus on the problem of design time testing of ADAS in a simulated environment, relying on Simulink models. The main research contributions in this dissertation are: - A testing approach for ADAS that combines multi-objective search with surrogate models to guide testing towards the most critical behaviors of ADAS, and to explore a larger part of the input search space with less computational resources. - An automated testing algorithm that builds on learnable evolution models and uses classification decision trees to guide the generation of new test scenarios within complex and multidimensional input spaces and help engineers interpret test results. - An automated technique that detects feature interaction failures in the context of self-driving systems based on analyzing executable function models typically developed to specify system behaviors at early development stages. - An automated technique that uses a new many-objective search algorithm to localize and repair errors in the feature interaction resolution rules for self-driving systems.

Beyond its functional requirements, architectural design, the quality of a software system is also defined by the degree to which it meets its non-functional requirements. The complexity of managing these non-functional requirements is exacerbated by the fact that they are potentially conflicting with one another. For cloud-based software, i.e., software whose service is delivered through a cloud infrastructure, other constraints related to the features of the hosting data center, such as cost, security and performance, have to be considered by system and software designers. For instance, the evaluation of requests to access sensitive resources results in performance overhead introduced by policy rules evaluation and message exchange between the different geographically distributed components of the authorization system. Duplicating policy rule evaluation engines traditionally solves such performance issues, however such a decision has an impact on security since it introduces additional potential private data leakage points. Taking into account all the aforementioned features is a key factor to enhance the perceived quality of service (QoS) of the cloud as a whole. Maximizing users and software developers satisfaction with cloud-based software is a challenging task since trade-off decisions have to be dynamically taken between these conflicting quality attributes to adapt to system requirements evolution. In this thesis, we tackle the challenges of building a decision support method to optimize software deployment in a cloud environment. Our proposed holistic method operates both at the level of 1) Platform as a service (PaaS) by handling software components deployment to achieve an efficient runtime optimization to satisfy cloud providers and customers objectives 2) Guest applications by making inroads into the design of applications to enable the design of secure systems that also meet flexibility, performance and cost requirements. To thoroughly investigate these challenges, we identify three main objectives that we address as follows: The first objective is to achieve a runtime optimization of cloud-based software deployment at the Platform as a service (PaaS) layer, by considering both cloud customers and providers constraints. To fulfill this objective, we leverage the models@run.time paradigm to build an abstraction layer to model a cloud infrastructure. In a second step, we model the software placement problem as a multi-objective optimization problem and we use multi-objective evolutionary algorithms (MOEAs) to identify a set of possible cloud optimal configurations that exhibit best trade-offs between conflicting objectives. The approach is validated through a case study that we defined with EBRC1, a cloud provider in Luxembourg, as a representative of a software component placement problem in heterogeneous distributed cloud nodes. The second objective is to ameliorate the convergence speed of MOEAs that we have used to achieve a run-time optimization of cloud-based software. To cope with elasticity requirements of cloud-based applications, we improve the way the search strategy operates by proposing a hyper-heuristic that operates on top of MOEAs. Our hyper-heuristic uses the history of mutation effect on fitness functions to select the most relevant mutation operators. Our evaluation shows that MOEAs in conjunction with our hyper-heuristic has a significant performance improvement in terms of resolution time over the original MOEAs. The third objective aims at optimizing cloud-based software trade-offs by exploring applications design as a complementary step to the optimization at the level of the cloud infrastructure, tackled in the first and second objectives. We aimed at achieving security trade-offs at the level of guest applications by revisiting current practices in software methods. We focus on access control as a main security concern and we opt for guest applications that manage resources regulated by access control policies specified in XACML2. This focus is mainly motivated by two key factors: 1) Access control is the pillar of computer security as it allows to protect sensitive resources in a given system from unauthorized accesses 2) XACML is the de facto standard language to specify access control policies and proposes an access control architectural model that supports several advanced access requirements such as interoperability and portability. To attain this objective, we advocate the design of applications based on XACML architectural model to achieve a trade-off between security and flexibility and we adopt a three-step approach: First, we identify a lack in the literature in XACML with obligation handling support. Obligations enable to specify user actions that have to be performed before/during/after the access to resources. We propose an extension of the XACML reference model and language to use the history of obligations states at the decision making time. In this step, we extend XACML access control architecture to support a wider range of usage control scenarios. Second, in order to avoid degrading performance while using a secure architecture based on XACML, we propose a refactoring technique applied on access control policies to enhance request evaluation time. Our approach, evaluated on three Java policy-based systems, enables to substantially reduce request evaluation time. Finally, to achieve a trade-off between a safe security policy evolution and regression testing costs, we develop a regression-test-selection approach for selecting test cases that reveal faults caused by policy changes. To sum up, in all aforementioned objectives, we pursue the goal of analysing and improving the current landscape in the development of cloud-based software. Our focus on security quality attributes is driven by its crucial role in widening the adoption of cloud computing. Our approach brings to light a security-aware design of guest applications that is based on XACML architecture. We provide useful guidelines, methods with underlying algorithms and tools for developers and cloud solution designers to enhance tomorrow’s cloud-based software design. Keywords: XACML-policy based systems, Cloud Computing, Trade-offs, Multi-Objective Optimization