A higher degree of automation is the future, and IoT and Edge Computing are enabling us to be the dwellers of futuristic energy efficient homes.
The energy demand in the housing sector has significantly increased over the past few decades. However, by integrating the Internet of Things (IoT) and some widely used machine learning algorithms 'smart buildings' can reduce their energy consumption by responding intelligently to its dwellers needs.
An ecosystem of connected devices, sensors, actuators is broadly called as Internet of Things (IoT). With the current developments in web technology and the onset of Industry 4.0, IoT is expected to be a well penetrated technology in future of industries. Smart cities that are exploring the concepts of IoT are already under development, and the Smart Home market is expected to reach $25B by 2022 as predicted by Gartner.
Recent research advancements are exploring the possibility to offer more with AI on the edge. AI on the edge suggests running Machine Learning (ML) and/or Deep Learning (DL) models on edge computing devices present at the network's edge to offer more smart features within the vicinity of the application being developed.
The demand for 'useful energy' has surged 2.3% over the past year in the building sector due to rapid development and enhanced lifestyle. Globally, the energy consumption at the commercial and residential buildings share one-third the demand for useful energy, and demand is accelerating due to increase in population and higher consumption by HVAC systems. To address the concerns of climate change and to reduce of energy wastage, household energy conservation has piqued interest.
The energy performance of buildings is reliant on several dependencies like surrounding weather, building type and the energy usage pattern of the households. Researchers around the world are interested in the possibility of exploring the Internet of Things (IoT) and some widely used machine learning algorithms to create a predictive model that can be used for forecasting the indoor temperature buildings.
The prediction of indoor temperatures and automatic control of high energy consuming devices over a network can reduce the overall energy consumption for heating and cooling of buildings, regardless of the building's construction.
Therefore, features like learning dwellers preferences are an essential need in future smart buildings.
Understanding the occupants behaviour (OB) is of paramount importance for indoor room temperature prediction. The determining part of OB depends on an enormous variance in the energy usage cycle and the thermal comfort requirements varying as per person, which is mainly affected by the occupants interaction with the thermal control systems (Thermostat and HVAC setting parameters), building components (fenestrations, drapes, roof styling) and energy appliance usage. Several studies have also revealed that the net consumption of a building severely depends on the work-style approach taken by the occupants.
An example implementation is calculating the prediction of the temperature of the conditioned space through a time series solution with the aid of traditional machine learning algorithms like Neural Networks, Random Forest and Support Vector Machine (SVM).
The prediction task can be done on an edge node that uses concepts of Virtual IoT devices, to provide a harmonized, inter-operable, light-weight and secure solution for smart building purposes.
In my own research it turned out that specifically the 'Ensemble Random Forest' Machine Learning model was a great fit for tasks like this at the network's edge, and can provide a good accuracy in predicting indoor temperature in advance and enable the efficient control of HVAC systems in a smart building.
This in turn could offer a comfortable, automated and smart solution to building dwellers.
The futuristic homes from the sci-fi movies we all are so aware of are actually possible. Now the question is "are we ready to accept and live in that future?"
An Edge Computing Architecture [1]
[1] Chakraborty T, Datta SK. Home automation using edge computing and internet of things. In2017 IEEE International Symposium on Consumer Electronics (ISCE) 2017 Nov 14 (pp. 47-49). IEEE.
[2] Paul D, Chakraborty T, Datta SK, Paul D. IoT and Machine Learning Based Prediction of Smart Building Indoor Temperature. In2018 4th International Conference on Computer and Information Sciences (ICCOINS) 2018 Aug 13 (pp. 1-6). IEEE.
Wevolver 2023