Intelligent edge-based recommender system for internet of energy applications

TL;DR

This paper presents an innovative edge-based recommender system integrated into the Home-Assistant platform to promote energy efficiency in households, emphasizing privacy preservation and real-time, explainable recommendations.

Contribution

It introduces the first energy-saving recommender system on edge devices, enhancing privacy and providing real-time, explainable energy-saving suggestions within a smart home environment.

Findings

Successful integration into Home-Assistant platform

Provides real-time, explainable energy recommendations

Ensures data privacy by processing locally on edge devices

Abstract

Preserving energy in households and office buildings is a significant challenge, mainly due to the recent shortage of energy resources, the uprising of the current environmental problems, and the global lack of utilizing energy-saving technologies. Not to mention, within some regions, COVID-19 social distancing measures have led to a temporary transfer of energy demand from commercial and urban centers to residential areas, causing an increased use and higher charges, and in turn, creating economic impacts on customers. Therefore, the marketplace could benefit from developing an internet of things (IoT) ecosystem that monitors energy consumption habits and promptly recommends action to facilitate energy efficiency. This paper aims to present the full integration of a proposed energy efficiency framework into the Home-Assistant platform using an edge-based architecture. End-users can visualize their consumption patterns as well as ambient environmental data using the Home-Assistant user interface. More notably, explainable energy-saving recommendations are delivered to end-users in the form of notifications via the mobile application to facilitate habit change. In this context, to the best of the authors' knowledge, this is the first attempt to develop and implement an energy-saving recommender system on edge devices. Thus, ensuring better privacy preservation since data are processed locally on the edge, without the need to transmit them to remote servers, as is the case with cloudlet platforms.