The Emergence of Hybrid Edge-Cloud Computing for Energy Efficiency in Buildings

Abstract

Edge computing is attracting an increasing attention presently even though most of the building energy efficiency solutions are still using cloud computing for gathering, pre-processing and analyzing energy data. However, edge computing still requires more power in order to be used alone to meet the high computation demand of artificial intelligence based energy saving solutions. Meanwhile, a hybrid edge-cloud architecture can be the best current approach to implement energy efficiency systems. It provides end-users and utility companies with a flexible control of their energy usage footprints, minimizes the cost of cloud hosting, and improves privacy-preservation. Accordingly, in this paper, we present a novel energy efficiency system based on a hybrid edge-cloud computing architecture. To analyze energy and occupancy data collected through different smart meters and occupancy sensors, we use a micro-moment approach to cluster energy observations into different categories representing both normal and abnormal energy usage. Following, a deep micro-moments (deepM2) scheme is deployed to automate the Anomaly Detection task, where a new approach called deepM2-AD is developed. Moving forward, deepM2-AD is implemented on three different architectures, defined as edge-only, cloud-only and hybrid edge-cloud to evaluate their performance and identify their merits and demerits. Overall, the hybrid edge-cloud architecture has presented the best compromise in terms of improving the processing speed, curtailing the cost of cloud hosting, and reducing the communication latency. Therefore, it has a great potential for supporting real-time energy consumption anomaly detection applications that help in minimizing wasted energy. 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.