Smart homes are powered by numerous programmable IoT platforms. Despite tremendous innovations, these platforms often suffer from safety and security issues. One class of defense solutions dynamically enforces safety and security policies, which essentially capture the expected behavior of the IoT system. While many proposed works were built on this runtime approach, they all are under-vetted. The primary reason lies in their evaluation approach. They are mostly self-evaluated in isolation using a virtual testbed combined with manually orchestrated test scenarios that rely on user interactions with the platform's UI. Such hand-crafted and non-uniform evaluation setups are limiting not only the reproducibility but also a comparative analysis of their efficacy results. Closing this gap in the traditional way requires a huge upfront manual effort, which causes the researchers turn away from any large-scale comparative empirical evaluation. Therefore, in this paper, we propose a highly-automated uniform evaluation platform, dubbed VetIoT, to vet the defense solutions that hinge on runtime policy enforcement. Given a defense solution, VetIoT easily instantiates a virtual testbed inside which the solution is empirically evaluated. VetIoT replaces manual UI-based interactions with an automated event simulator and manual inspection of test outcomes with an automated comparator. We developed a fully-functional prototype of VetIoT and applied it on three runtime policy enforcement solutions: Expat, Patriot, and IoTguard. VetIoT reproduced their individual prior results and assessed their efficacy results via stress testing and differential testing. We believe VetIoT can foster future research/evaluation.

翻译：智能家居由大量可编程物联网平台驱动。尽管创新层出不穷，这类平台常面临安全与可靠性问题。一类防御方案通过动态实施安全策略与可靠性策略来运行，这类策略本质上捕捉了物联网系统的预期行为。尽管已有诸多研究基于该运行时方法展开，但这些方案均未经过充分审验。根本原因在于其评估方式：多数方案通过虚拟测试床结合人工编排的测试场景进行独立自评，依赖用户与平台用户界面的交互。这种手工定制、非统一的评估设置不仅限制了可复现性，更阻碍了其有效性结果的比较分析。以传统方式弥合这一差距需要巨大的前期人工投入，导致研究者放弃大规模比较性实证评估。为此，本文提出高度自动化的统一评估平台VetIoT，用于审验依赖运行时策略强制的防御方案。给定防御方案后，VetIoT可快速实例化虚拟测试床，在其内部进行实证评估。VetIoT用自动化事件模拟器替代人工用户界面交互，用自动化比较器替代人工测试结果检验。我们开发了VetIoT的全功能原型，并将其应用于三项运行时策略强制方案（Expat、Patriot与IoTguard）。VetIoT复现了各方案既有的独立结果，并通过压力测试与差异测试评估了其有效性。我们相信VetIoT有望推动未来的研究与评估工作。