Predicting Energy Consumption of TVsinVideo Streaming Using Machine Learning Techniques

The climate crisis highlights the environmental impact of information and communication technologies (ICT), necessitating sustainable practices to reduce carbon emissions. In compliance with the Corporate Sustainability Reporting Directive (CSRD), large European companies are required to disclose their social and environmental risks, enhancing transparency regarding corporate sustainability performance. With video streaming increasingly leading Internet traffic, research about CO2 emission in video streaming has become critical, especially for the content providers and end-users. In this study, we provide a machine learning-based model to predict the energy consumption of TVs while playing videos. The energy consumption predicted by our model is a means to estimate the CO2 emission of the client side in the streaming chain. To acquire energy data of TVs, we deploy a framework to automatically monitor the energy consumption of TVs. The experimental results show that our model can accurately predict energy consumption with a high correlation to actual measurements, achieving $R^{2}$ values of up to 0.988 in scenarios with known TV models.