The development of IoT techniques and the automation enabled by trigger-action platforms have brought great convenience to human life. However, with time the proliferating IoT devices will eventually overwhelm such platforms. Specifically, the ever-increasing huge amount of linkage rules signifificantly challenge the response rate as well as the effificiency of conflict rule detection, which will lead to terrible user experience and disastrous security issues. In this work, we develop a rule engine system for large-scale trigger-action applications. The key insight is that the linkage rules can be executed independent from the central cloud platform, which only require edge-side gateways and an efficient addressing strategy. The platform-side maintains an global image consisting of mappings between rules and gateway addresses, while edge-side rule engine is designed based on an open source tool, i.e., Drools. The implemented RETE algorithm helps to carry out large-scale linkage rules. In addition, we design a lightweight yet effective conflict detection algorithm that categorizes those rules into several types to downgrade the computation complexity. Experiments prove that the system can reduce the resource consumption of the platform-side and remarkably enhance the response speed. Moreover, our system also achieve a high efficacy when detecting conflicts.

物联网技术的发展和触发-动作平台实现的自动化给人类生活带来了极大的便利。然而，随着时间的推移，激增的物联网设备最终将淹没此类平台。具体来说，不断增加的大量链接规则极大地挑战了响应速度和冲突规则检测的效率，这将导致糟糕的用户体验和灾难性的安全问题。在这项工作中，我们为大规模触发-动作应用程序开发了一个规则引擎系统。关键洞察在于，联动规则可以独立于中央云平台执行，只需要边缘侧网关和高效的寻址策略。平台端维护一个全局镜像，由规则和网关地址的映射组成，而边缘规则引擎是基于开源工具Drools设计的。实现的 RETE 算法有助于执行大规模链接规则。此外，我们设计了一种轻量级但有效的冲突检测算法，将这些规则分为几种类型以降低计算复杂度。实验证明，该系统能够降低平台端的资源消耗，显着提高响应速度。此外，我们的系统在检测冲突时也达到了很高的效率。实验证明，该系统能够降低平台端的资源消耗，显着提高响应速度。此外，我们的系统在检测冲突时也达到了很高的效率。实验证明，该系统可以降低平台端的资源消耗，显着提高响应速度。此外，我们的系统在检测冲突时也达到了很高的效率。