Automatic trigger generation for end user written rules for home automation

Abstract

To customize the behavior of a smart home, an end user writes rules. When an external event satisfies a rule's trigger, the rule's action executes; for example, when the temperature is above a certain threshold, then window awnings might be extended. End users often write incorrect rules. This paper's technique prevents a certain category of errors in the rules: errors due to too few triggers. It statically analyzes a rule's actions to automatically determine a set of necessary and sufficient triggers. We implemented the technique in a tool called TrigGen and tested it on 96 end-user written rules for openHAB, an open-source home automation platform. It identified that 80% of the rules had fewer triggers than required for correct behavior. The missing triggers could lead to unexpected behavior and security vulnerabilities in a smart home.