No abstract

Since software systems need to be continuously available under varying conditions, their ability to evolve at runtime is increasingly seen as one key issue. Modern programming frameworks already provide support for dynamic adaptations. However the high-variability of features in Dynamic Adaptive Systems (DAS) introduces an explosion of possible runtime system configurations (often called modes) and mode transitions. Designing these configurations and their transitions is tedious and error-prone, making the system feature evolution difficult. While Aspect-Oriented Modeling (AOM) was introduced to improve the modularity of software, this paper presents how an AOM approach can be used to tame the combinatorial explosion of DAS modes. Using AOM techniques, we derive a wide range of modes by weaving aspects into an explicit model reflecting the runtime system. We use these generated modes to automatically adapt the system. We validate our approach on an adaptive middleware for homeautomation currently deployed in Rennes metropolis.

Abstract-In the landscape of cloud computing, the competition between providers has led to an ever growing number of cloud solutions offered to consumers. The ability to run and manage multi-cloud systems (i.e., applications on multiple clouds) allows exploiting the peculiarities of each cloud solution and hence optimising the performance, availability, and cost of the applications. However, these cloud solutions are typically heterogeneous and the provided features are often incompatible. This diversity hinders the proper exploitation of the full potential of cloud computing, since it prevents interoperability and promotes vendor lock-in, as well as it increases the complexity of development and administration of multi-cloud systems. This problem needs to be addressed promptly. In this paper, we provide a classification of the state-of-the-art of cloud solutions, and argue for the need for model-driven engineering techniques and methods facilitating the specification of provisioning, deployment, monitoring, and adaptation concerns of multi-cloud systems at design-time and their enactment at run-time.

Abstract. Constructing and executing distributed systems that can adapt to their operating context in order to sustain provided services and the service qualities are complex tasks. Managing adaptation of multiple, interacting services is particularly difficult since these services tend to be distributed across the system, interdependent and sometimes tangled with other services. Furthermore, the exponential growth of the number of potential system configurations derived from the variabilities of each service need to be handled. Current practices of writing low-level reconfiguration scripts as part of the system code to handle run time adaptation are both error prone and time consuming and make adaptive systems difficult to validate and evolve. In this paper, we propose to combine model driven and aspect oriented techniques to better cope with the complexities of adaptive systems construction and execution, and to handle the problem of exponential growth of the number of possible configurations. Combining these techniques allows us to use high level domain abstractions, simplify the representation of variants and limit the problem pertaining to the combinatorial explosion of possible configurations. In our approach we also use models at runtime to generate the adaptation logic by comparing the current configuration of the system to a composed model representing the configuration we want to reach.

The ThingML approach, which was inspired by UML, addresses the challenges of distribution and heterogeneity in the Internet of Things. This model-driven, generative approach has been continuously evolved and applied to cases in different domains, including a commercial e-health solution.The challenge of writing, deploying, and evolving software for the Internet of Things (IoT) is often underestimated. 1 From a software engineering viewpoint, IoT applications have two main characteristics. The first is distribution over a large range of processing nodes. The second is high heterogeneity of the processing nodes and the protocols used between them. A rich collection of software engineering literature exists for each characteristic. However, not much research has addressed the combination of the two, which presents three specific challenges.First, in large IoT applications, distribution occurs across many heterogeneous nodes, which play different complementary roles to process data close to the sources and respond in a decentralized way. Most research in distributed systems, telecommunications, or even sensor networks has focused on scaling to very high numbers of nodes but has dealt with only a few types and roles among those nodes (for example, clientserver, peer tracker, sensor gateway server, and so on). 2 This homogeneity keeps the software complexity manageable because it allows developing and deploying similar code on large sets of nodes. But this isn't the case with IoT software, which must scale in not only the number of nodes but also the number of types of nodes and platforms, from microcontrollers up to the cloud.Second, for IoT systems to deliver their full potential, developers must leverage the diverse resources and decentralized computing power that heterogeneous nodes provide. Many tools and techniques hide this heterogeneity to enable developers to write applications that execute the same way on different platforms (OSs, web browsers, mobile devices, and so on). 3 In IoT applications, part of the heterogeneity isn't accidental and must remain explicit and exploitable by developers. This way, developers can, for example, take advantage of IoT nodes' deep-sleep modes, which typically differ among platforms, to save power.Third, in the IoT vision, applications are no longer isolated, proprietary silos of devices and software. 4 Rather, they must combine things readily available in the environment: some generic, some applicationspecific, and some legacy things. So, IoT applications are meant to run on shared hardware, and developers should pay special attention to avoiding unwanted interactions between applications.Consequently, industrializing IoT applications proves challenging at all development phases and requires teams of developers with a broad range of competences from electronics and microcontrollers to cloud and data mining. The ThingML (Internet-of-Things Modeling Language) approach aims to facilitate collaboration between service developers and platform experts so that together they can prod...

Aspects have gained attention in the earlier steps of the software life-cycle leading to the creation of numerous adhoc Aspect-Oriented Modeling (AOM) approaches. These approaches mainly focus on architecture diagrams, class diagrams, state-charts, scenarios or requirements and generally propose Aspect-Oriented composition mechanisms specific to a given kind of models defined by its own meta-model. Recently, some generic AOM approaches propose to extend the notion of aspect to any domain specific modelling language (DSML). In this trend, this paper presents GeKo. GeKo has the following properties. i) It is a generic AOM approach easily adaptable to any DSML with no need to modify the domain meta-model or to generate domain-specific frameworks. ii) It keeps a graphical representation of the weaving between an aspect model and the base model. iii) It is a tool-supported approach with a clear semantics of the different operators used to define the weaving. GeKo relies on the definition of mappings between the different views of an aspect, based on the concrete (graphical) syntax associated to the DSML. To illustrate GeKo, we derive, from the Arcade Game Maker Pedagogical Product Line, a new product in which new features are woven into the Product Line models.

Cyber Physical Systems (CPS) offer new ways for people to interact with computing systems: every thing now integrates computing power that can be leveraged to provide safety, assistance, guidance or simply comfort to users. CPS are long living and pervasive systems that intensively rely on microcontrollers and low power CPUs, integrated into buildings (e.g. automation to improve comfort and energy optimization) or cars (e.g. advanced safety features involving car-to-car communication to avoid collisions). CPS operate in volatile environments where nodes should cooperate in opportunistic ways and dynamically adapt to their context. This paper presents µ-Kevoree, the projection of Kevoree (a component model based on models@runtime) to microcontrollers. µ-Kevoree pushes dynamicity and elasticity concerns directly into resource-constrained devices. Its evaluation regarding key criteria in the embedded domain (memory usage, reliability and performance) shows that, despite a contained overhead, µ-Kevoree provides the advantages of a dynamically reconfigurable component-based model (safe, fine-grained, and efficient reconfiguration) compared to traditional techniques for dynamic firmware upgrades.

This paper discusses preliminary work on modeling and validation dynamic adaptation. The proposed approach is on the use of aspect-oriented modeling (AOM) and models at runtime. Our approach covers design and runtime phases. At design-time, a base model and different variant architecture models are designed and the adaptation model is built. Crucially, the adaptation model includes invariant properties and constrains that allow the validation of the adaptation rules before execution. During runtime, the adaptation model is processed to produce a correct system configuration that should be executed.