Background:

Internet of Things (IoT) enables the interaction among objects to provide services to their users. Areas such as eHealth, smart energy, and smart buildings have been benefiting from the IoT potential. However, the development of IoT systems is still complex because it deals with a highly dynamic, volatile, and heterogeneous environment. These characteristics require discovering devices, managing these devices’ context, and self-adapt their behavior.

Goal

: In this work, we propose a self-adaptive IoT infrastructure to support multiple facets, i.e., the contextual discovery of smart objects, the context management, and the self-adaptation process of the development of these systems.

Methods

: We evaluated the proposed infrastructure by developing a smart building application with and without it. The evaluation focused on four issues: the feasibility of integrating the context management through middleware platforms with adaptation based on workflows in a request/response communication model, the impact of our infrastructure on the development of self-adaptive IoT systems considering cyclomatic complexity and coupling code metrics; the impact of using contextual filters on the orchestrator of self-adaptation; and the impact on the quality of the self-adaptation.

Results

: The results suggest that: (i) it is feasible to use the proposed infrastructure in the development of self-adaptive IoT systems; (ii) there is a reduction in the cyclomatic complexity and the coupling with our approach, (iii) there is a considerable decrease in the number of rules evaluated at runtime, (iv) our infrastructure reduces the execution time of the adaptations when using contextual filters, and (v) the self-adaptation process was effective when using the orchestrator of self-adaptations.

Conclusion

: With these results, we observed that the proposed multifaceted infrastructure could reduce the complexity related to the development of IoT systems, in addition to optimizing their self-adaptation process.

中文翻译：

---

自适应物联网系统的多方面基础架构

背景：

物联网（IoT）使对象之间的交互能够为用户提供服务。eHealth，智能能源和智能建筑等领域已从物联网潜力中受益。但是，物联网系统的开发仍然很复杂，因为它要处理高度动态，易变且异构的环境。这些特征需要发现设备，管理这些设备的上下文并自适应其行为。

目标

：在这项工作中，我们提出了一种自适应的IoT基础架构，以支持多个方面，即智能对象的上下文发现，上下文管理以及这些系统开发的自适应过程。

方法

：我们通过开发带有或不带有智能建筑应用程序的基础设施，对提议的基础设施进行了评估。评估集中在四个问题上：通过中间件平台将上下文管理与基于请求/响应通信模型中基于工作流的适应性进行集成的可行性，考虑到循环复杂性和耦合代码的基础设施对自适应IoT系统开发的影响指标; 使用上下文过滤器对自适应流程的影响；以及对自适应质量的影响。

结果

：结果表明：（i）在开发自适应物联网系统中使用拟议的基础设施是可行的；（ii）降低了循环复杂性，并与我们的方法相结合，（iii）运行时评估的规则数量大大减少，（iv）我们的基础结构减少了使用上下文时适应的执行时间过滤器；（v）使用自适应编排器时，自适应过程很有效。

结论

：有了这些结果，我们观察到，所提出的多方面基础设施除了可以优化其自适应过程之外，还可以降低与物联网系统开发相关的复杂性。