safetap.pdf (1.18 MB)

SafeTAP: An Efficient Incremental Analyzer for Trigger-Action Programs

Download (1.18 MB)
posted on 16.06.2021, 19:17 by McKenna McCall, Faysal Hossain Shezan, Abhishek Bichhawat, Camille Cobb, Limin Jia, Yuan Tian, Cooper Grace, Mitchell Yang

Home automation rules that allow users to connect smart home devices using trigger-action programs (TAP) can interact in subtle and unexpected ways. Determining whether these rules are free of undesirable behavior is challenging; so researchers have developed tools to analyze rules and assist users. However, it is unclear whether users need such tools, and what help they need from such tools. To answer this question, we performed a user study where half of the participants were given our custom analysis tool SafeTAP and the other half were not. We found that users are not good at finding issues in their TAP rules, despite perceiving such tasks as easy.

The user study also indicates that users would like to check their rules every time they make rule changes. Therefore, we designed a novel incremental symbolic model checking (SMC) algorithm, which extends the basic SMC algorithm of SafeTAP. SafeTAPΔ only performs analysis caused by the addition or removal of rules and reports only new violations that have not already been reported to the user. We evaluate the performance of SafeTAPΔ and show that incremental checking on average improves the performance by 6X when adding new rules.


Usage metrics