A Survey on Energy Consumption and Environmental Impact of Video Streaming

The pressing need to reduce greenhouse gas emissions in light of the growing demand for video streaming is addressed. The environmental impact of this trend is considerable, as video streaming accounts for over 65% of global internet data traffic and contributes roughly 3.7% of worldwide greenhouse gas emissions, a figure comparable to that of the airline industry. The survey aims to provide insights into the energy consumption and environmental footprint of video streaming technologies. However, it highlights the challenges in measuring energy usage and carbon emissions due to variability in factors like location, device type, and technological developments. The focus of the survey is on two main phases of video streaming: content provisioning (which includes encoding, storage, and transmission) and content consumption (which involves decoding and displaying the video on end devices).   

Regarding content provisioning, the two factors analysed in terms of their energy consumption are (1) encoding and (2) storage. Reviewed studies on video encoding focus on the impact that codec types and encoding parameters have on energy consumption. Advanced codecs provide higher compression efficiency but consume more energy, while encoding parameters like higher QP (Quantisation Parameter) values, reduced framerate, and faster presets can lower energy use. As for storage, the attention is primarily drawn at CDNs that store the datasets, and their caching strategies. Recent studies focus on energy-efficient caching algorithms, using methods like heuristics, linear programming, differential evolution, and game theory to balance energy savings, quality of experience (QoE), and operational costs. These approaches help determine the ideal number of servers, steer content across multiple CDNs, optimise cache sizes, and incorporate green energy sources to improve sustainability in CDN operations.  

Three critical factors are identified as contributors to energy consumption on end-user devices: (1) content retrieval via the network interface card (NIC), (2) video decoding and (3) video display. In terms of NICs, energy consumption varies significantly based on the network type, signal strength, video resolution, and data throughput. Generally, WiFi is more energy-efficient than LTE, while Ethernet provides energy-saving benefits through its EEE (Energy Efficient Ethernet) standard. Video decoding energy use depends on factors such as the decoding mode (hardware vs. software), codec choice, encoding settings (e.g., resolution and framerate), and any optimisations in the decoding process. Finally, the power consumption of displays is influenced by the technology used (LCD, OLED, AMOLED), with different methods being explored to estimate energy consumption across these display types. The survey discusses open research issues for advancing energy-efficient video streaming. These include optimising encoding and storage processes, developing energy-aware streaming protocols, and encouraging user awareness of energy consumption during streaming choices.