Cloud-based IoT Analytics for the Smart Grid: Experiences from a 3-year Pilot T. Hasan, P. Kikiras, A. Leonardi AGT International Hilpertstraße 35 64295 Darmstadt, Germany {thasan, pkikiras, ale- onardi}@agtinternational.com H. Ziekow * Furtwangen University Robert-Gerwig-Platz 1 78120 Furtwangen, Germany zie@hs-furtwangen.de J. Daubert TU Darmstadt / CASED Mornewegstr. 32 64293 Darmstadt, Germany joerg.daubert@cased.de ABSTRACT The transformation of electrical grids into smart-grid is seen as one of the major technological challenges of our times and at the same time as one of the key domains for Internet of Things (IoT). Smart-home technologies and corresponding analytics are an integral part of many use cases in this field. In this paper we present a cloud-based test bed for capturing and analyzing smart-home data and report on experiences from a 3 year pilot with a cloud-based system. We discuss on real-world challenges that we encountered throughout the pilot - e.g. related to big data volumes and data quality - and describe corresponding technical solutions. Categories and Subject Descriptors H.4 [Information Systems Applications]: Miscellaneous General Terms Internet of Things Keywords Smart-home, IoT, Analytics 1. INTRODUCTION In recent years, environmental and economic considerations have fueled investments in renewable energy sources. For instance, member states of the European Union have set the so-called 20-20-20 goals, to obtain at least 20 percent of their electricity from renewable sources by 2020. Germany even strives for 80 percent renewable sources in the energy mix by 2050. However, this change in the energy supply has severe implications on the operation of electrical grids and the corresponding ICT infrastructure. Decentralized * Main part of the work was done while at AGT Interna- tional. production (e.g. with solar panels on roof tops) causes the need to expand grid management infrastructures to the low voltage level, i.e. the level of streets and houses. However, it remains an open question how technical solutions for grid management of on this local level should be designed. Smart-home technology is seen as a key enable for better understanding and managing the energy consumption on in the low voltage grid. To better understand the correspond- ing challenges and solutions for data capturing and process- ing, the German federal government has provided funding for the PeerEnergyCloud research project [12]. This paper reports on a cloud-based test bed for capturing and analyz- ing sensor data from smart-homes that was developed and piloted throughout this project. The pilot was conducted using smart-home hardware packages with sensors for cap- turing device specific energy consumption as well as sensors for environmental conditions (e.g. room temperature). In total we used 60 packages that were installed in various se- tups and different homes. Overall, we observed installations for up to three years and captured about 12 Billion sensor measurements in total. The emphasis of this paper is on the technical solutions that we developed for capturing and an- alyzing the data as well as on the challenges that we faced in the real-world deployment. Key contributions of the paper are the following: • We provide a solution architecture and details on tech- nical components for a test bed that supports cloud- based service on top of smart-home sensors. • We discuss the real-world challenges for data capturing and analysis that we derived from a 3 year deployment in the smart home/smart grid domain. • We present technical solutions that respond to the challenges that we encountered throughout the pilot. The remainder of the paper is structured as follows. Section 2 discusses relevant related work. Section 3 provides a high level overview of the solution architecture and different com- ponents of our test bed and Section 4 discusses operational issues that we faced when piloting the system. The subse- quent sections address technical details of the key compo- nents in our test bed and the mechanisms for addressing the TRIDENTCOM 2015, June 24-25, Vancouver, Canada Copyright © 2015 ICST DOI 10.4108/icst.tridentcom.2015.259694