Toward user-centric feature composition for the Internet of Things
This work addresses usability issues in IoT systems for end-users, but it is incremental as it builds on existing feature-based approaches.
The paper tackles the challenge of managing feature interactions in home automation by proposing a runtime composition mechanism using priority, and demonstrates its effectiveness through three examples and a user study on comprehension.
Many user studies of home automation, as the most familiar representative of the Internet of Things, have shown the difficulty of developing technology that users understand and like. It helps to state requirements as largely-independent features, but features are not truly independent, so this incurs the cost of managing and explaining feature interactions. We propose to compose features at runtime, resolving their interactions by means of priority. Although the basic idea is simple, its details must be designed to make users comfortable by balancing manual and automatic control. On the technical side, its details must be designed to allow meaningful separation of features and maximum generality. As evidence that our composition mechanism achieves its goals, we present three substantive examples of home automation, and the results of a user study to investigate comprehension of feature interactions. A survey of related work shows that this proposal occupies a sensible place in a design space whose dimensions include actuator type, detection versus resolution strategies, and modularity.