With the accelerated advancement of IoT, diverse devices are ubiquitously deployed in environments. Building on this, Web of Things (WoT) further integrates fragmented device services and provides unified interfaces using standardized Web technologies, promoting the development and deployment of WoT applications to sense and regulate the environment. However, disparate WoT applications independently control devices in the WoT environment, causing interference among devices and with the environment. This results in device behaviors that deviate from user expectations, causing violations of the user's desired environment properties. The intricate interplay of applications, user activities, and environment changes makes identifying and resolving potential violations a complex task. In this paper, we introduce EnvGuard, an environment-centric approach for property description, violation identification, and resolution in WoT environment. EnvGuard proposes a conceptual schema to model the relationship between device services and environment context, and automatically extends the conceptual schema into a specific environment representation based on device and space information. Furthermore, EnvGuard employs a template-based approach, enabling users to describe spatial and temporal properties based on the abstract device effects on the environment, and translating properties description into formal expressions. EnvGuard adopts a hybrid model checking method to respectively identify the spatial and temporal violations, and a resolution strategy that align with user intention is proposed to resolve violations. We evaluate EnvGuard through user studies and our proposed dataset, which is constructed by collecting real-world data from a laboratory WoT environment and manually labeling ten types of violations. The results confirm the usability, feasibility and efficiency of EnvGuard.

翻译：摘要：随着物联网技术的快速发展，各类设备被广泛部署于环境中。在此基础上，物联网络（Web of Things，WoT）进一步整合碎片化的设备服务，并利用标准化Web技术提供统一接口，从而推动WoT应用的开发与部署，以实现对环境的感知与调控。然而，不同的WoT应用独立控制环境中的设备，导致设备之间及设备与环境之间产生干扰，使得设备行为偏离用户预期，进而违反用户所需的环境性质。应用、用户活动与环境变化之间的复杂交互，使得识别并解决潜在违规行为成为一项艰巨任务。本文提出EnvGuard，一种面向WoT环境、以环境为中心的性质描述、违规识别与解决方法。EnvGuard通过概念模式建模设备服务与环境上下文之间的关系，并基于设备与空间信息自动将概念模式扩展为特定环境表示。此外，EnvGuard采用基于模板的方法，使用户能够基于设备对环境的抽象影响描述时空性质，并将性质描述转化为形式化表达式。EnvGuard采用混合模型检测方法分别识别空间与时间违规行为，并提出一种符合用户意图的解决策略以消除违规。我们通过用户研究与自建数据集对EnvGuard进行评估，该数据集通过采集实验室WoT环境中的真实数据并手动标注十种违规行为构建而成。实验结果证实了EnvGuard的可用性、可行性与高效性。