ATHENA Christian Doppler (CD) Laboratory

Energy Consumption in Video Streaming: Components, Measurements, and Strategies

ICT4S Workshop: Towards Sustainable and Energy-Aware Video Communications (SEViC)

June 5-9 2023 – Rennes, France

Abstract: The rapid growth of video streaming usage is a significant source of energy consumption, driven by improved internet connections and service offerings, the quick development of video entertainment, the deployment of Ultra High-Definition, Virtual and Augmented Reality, as well as an increasing number of video surveillance and IoT applications. To address this challenge, it is essential to understand the various components involved in energy consumption during video streaming, ranging from video encoding to decoding and displaying the video on the end user’s screen. Then, it is critical to measure energy consumption for each component accurately and conduct an in-depth analysis to develop energy-efficient strategies that optimize video streaming. These components are classified into three categories: (i) data centers, which include encoding, packaging, and storage on cloud data centers; (ii) networks, which include core network and access networks; and (iii) end-user devices which involve decoding, players, hardware, etc.

In addition to identifying the primary components of video streaming that affect energy consumption, it is important to conduct a comprehensive analysis of the entire video streaming. It is also essential to balance energy optimization and service quality to ensure that energyefficient strategies are implemented without sacrificing the quality of video streaming services.

This talk aims to provide insights into the components of video streaming that contribute to energy consumption and highlight the challenges associated with measuring their energy usage. I will also introduce the tools that can be used for energy measurements for those components and the possible and associated strategies that lie within energy efficiency. By accurately measuring energy consumption, digital media companies can effectively monitor and control their energy usage, ultimately leading to cost savings and improved sustainability.

Exploring the Energy Consumption of Video Streaming: Components, Challenges, and Opportunities

10th FOKUS Media Web Symposium

June 13–14, 2023 – Berlin, Germany

Abstract: The rapid growth of video streaming usage is a significant source of energy consumption, driven by improved internet connections and service offerings, the quick development of video entertainment, the deployment of Ultra High-Definition, Virtual and Augmented Reality, as well as an increasing number of video surveillance and IoT applications. However, it is essential to note that these advancements come at the cost of energy consumption. To address this challenge, it is essential to understand the various components involved in energy consumption during video streaming, ranging from video encoding to decoding and displaying the video on the end user’s screen. Then, it is critical to accurately measure energy consumption for each component and conduct an in-depth analysis to develop energy-efficient strategies that optimize video streaming. I categorize these components into three categories: (i) data centers, (ii) networks, and (iii) end-user devices.

Data centers: Data centers are responsible for the significant growth in video data traffic, which is estimated to reach over 1,000 TWh of power consumption by 2025. To effectively manage energy consumption in data centers, it is crucial to understand the various components that contribute to it, including encoding process and parameters, resource provisioning, core network, storage, as well as hardware aspects such as cloud platform features and hardware units (e.g., CPU, GPU). By analyzing and optimizing these energy-intensive components, energy-efficient strategies, such as energy management and task schedulers, and energy-efficient codecs can be developed to improve the sustainability of data centers.

Networks: The next category within video streaming is the transmission of video from data centers to end devices through heterogeneous networks. The network energy-intensive components are CDNs, routers, switches, and network channels. To optimize energy efficiency during video transmission, it is essential to manage and optimize energy consumption in each of these components. This can be achieved by implementing efficient routing algorithms, reducing data redundancy, and utilizing power-saving mechanisms in network devices.

End-user devices: The last category of the video streaming is video usage at the end-user device, which has been shown to account for the majority of energy consumption by the decoding hardware and end-user devices. Energy consumption in this category is due to the components such as decoding, players, browsers, codecs, operating systems, and hardware (e.g., CPU, display). Improving the energy efficiency of end-user devices can significantly reduce energy consumption in video streaming. Some examples of achieving this are through the use of more energy-efficient devices such as laptops, tablets, and smartphones or by improving screen display technologies.

In addition to identifying the primary components of video streaming that affect energy consumption, it is important to conduct a comprehensive analysis of the entire video streaming. It is also important to strike a balance between energy optimization and service quality to ensure that energy-efficient strategies are implemented without sacrificing the quality of video streaming services.

In this talk, my objective is to provide insights into the components of video streaming that contribute to energy consumption and highlight the challenges associated with measuring their energy usage. I will also introduce the tools that can be used for energy measurements for those components and the possible and associated strategies that lie within energy efficiency. By accurately measuring energy consumption, digital media companies can effectively monitor and control their energy usage, ultimately leading to cost savings and improved sustainability.

The first International ACM Green Multimedia Systems Workshop, hosted and organized as part of the 14th ACM Multimedia Systems Conference, took place on Saturday, 10 June 2023 in the beautiful city of Vancouver, Canada. This workshop served as a crucial forum for researchers in multimedia systems to present and share their latest research findings, with a specific focus on energy consumption and greenhouse gas emissions in multimedia systems.

The workshop featured eight high-quality technical presentations, with an impressive acceptance rate of 66.7%. These presentations included both full papers and short papers, showcasing a diverse range of innovative approaches and solutions related to green video streaming. It was a significant opportunity for multimedia researchers to come together and delve into this timely and important topic.

The workshop was the participation of  experts and researchers in the field of green video streaming. Researchers from universities including Fraunhofer FOKUS, Friedrich-Alexander-Universität Erlangen-Nürnberg, and university of Klagenfurt presented their research papers, shared their valuable insights and research findings, further enriching the discussions. Researchers from leading companies such as Synamedia, Amazon, and Ateme also presented their research papers and  actively participated in the workshop.

Presentations in GMSys: 

VE-Match: Video Encoding Matching-based Model for Cloud and Edge Computing Instances
Samira Afzal, Narges Mehran (Alpen-Adria-Universität Klagenfurt, Austria); Sandro Linder (Bitmovin, Klagenfurt, Austria); Christian Timmerer, Radu Prodan (Alpen-Adria-Universität Klagenfurt, Austria)

Studying Green Video Distribution as a Whole
Burak Kara, Gwendal Simon (Synamedia); Bruno Tuffin (Inria); Jerome Vieron (Synamedia); Ali C. Begen (Ozyegin University)

End-to-end Optimizations for Green Streaming
Robert Seeliger, Stefan Pham, Stefan Arbanowski (Fraunhofer FOKUS)

Audience Aware Streaming: New Dynamics in OTT distribution
Jan Outters, Mickael Raulet (Ateme S.A.)

Green Video Complexity Analysis for Efficient Encoding in Adaptive Video Streaming 
Vignesh V Menon (Alpen-Adria-Universität Klagenfurt); Christian Feldmann (Bitmovin); Klaus Schoeffmann (Alpen-Adria-Universität Klagenfurt); Mohammed Ghanbari (University of Essex); Christian Timmerer (Alpen-Adria Universität Klagenfurt)

Energy Efficiency Improvements in Software-Based Video Encoding
Jan De Cock (Synamedia)

Video Decoding Energy Reduction using Temporal-Domain Filtering
Christian Herglotz, Matthias Kränzler, Robert Ludwig, André Kaup (Friedrich-Alexander-Universität Erlangen-Nürnberg)

The analysis of DASH manifest optimizations
Yongjun Wu (Amazon)

The gathering of researchers, industry professionals, and experts in the field showcased the increasing importance and impact of green video streaming. The workshop served as a catalyst for knowledge dissemination, collaboration, and the development of sustainable solutions in multimedia systems. It significantly contributed to advancing the understanding and development of green video streaming technologies, while addressing the environmental impact of digital media consumption.

Acknowledgments:
We would like to express our heartfelt gratitude to all the workshop participants, speakers, technical program committees, authors, and attendees who played a crucial role in the success of this event. Special thanks go to the ACM organizing committee, particularly Mohamed Hefeeda and Shervin Shirmohammadi, for their invaluable support in providing a platform to host this significant workshop. Additionally, we extend our appreciation to GAIA for their generous technical sponsorship, which greatly contributed to the smooth organization and execution of the workshop.