A Quality of Experience Prediction Model for Smart Home Energy Management Systems

Smart Home Energy Management (SHEM) systems allow to optimize the usage of resources in our houses while making them comfortable to humans. However, in the development of SHEM systems only limited attention has been put to the impact of the system performance on the Quality of Experience (QoE), considering the humans mostly as the final recipient of the service rather than the central component of the whole service. In this paper, we consider a SHEM system that, on the basis of pre-created user profiles, aims to reduce the electricity costs while preserving the QoE perceived by the user (evaluated in terms of perceived user annoyance due to the shift of the appliance's starting time). We focus on the identification of the user profile during the live sessions in the SHEM scenario and on the analysis of the impact of such task in the SHEM system performance. To evaluate the impact of profile selection, we conducted experiments in which 12 people were asked to provide their feedback each time the SHEM system proposed a modified starting time for an appliance with respect to the user's preferences. From experiments results it is found that the SHEM system needs on average just one feedback from the user to find the best user profile.