Content for TR 26.942 Word version: 19.0.0

1… 4… 4.2… 5… 6… 7… 7.3… 7.4… 7.5… 7.6… 7.7… 7.8… 7.9… 7.10… 8… 9…

There have been significant efforts to better understand and estimate the environmental impacts of media consumption. There is evidence of early attempts to measure energy for media consumption in the UK, USA, EU and globally over the past decade [74], [75], [76], [77], [78].

Several standards setting organisations broadly active in the areas of broadcasting and telecommunications are currently considering energy efficiency and the reduction of climate impact. Likewise, several industry fora are active in this area. This clause documents some of the efforts underway, and references standards and reports currently available.

4.2.2 3GPP p. 16

4.2.2.1 Introduction p. 16

3GPP has undertaken significant efforts to address energy efficiency within mobile networks. Technical reports and specifications that outline methodologies for measuring and reporting energy consumption and efficiency have been developed. This includes the establishment of collection, reporting and exposure procedures at various components of mobile networks, which helps in assessing and optimizing the performance of network elements and services in terms of their energy usage.

4.2.2.2 Collection and exposure of energy consumption information at OAM p. 16

4.2.2.2.1 Introduction p. 16

This clause summarizes TS 28.554 as it relates to the evaluation and collection of energy consumption information by the Operations, Administration and Maintenance (OAM) capability of the 5G System, as specified in 3GPP Release 18 by SA WG5.

4.2.2.2.2 Collection of network energy information by OAM p. 16

Clause 6.7.3 of TS 28.554 defines the Energy Consumption KPI of a Physical Node. The network energy information that can be collected by the OAM capability includes that listed in Table 4.2.2.2.2-1.

Table 4.2.2.2.2-1: Network energy information collected by OAM

KPI category	Description	Reference	Clause
Energy Consumption (EC) information	Energy Consumption of a gNodeB	TS 28.554	6.7.3.4.2
	Energy Consumption of the NG-RAN		6.7.3.4.1
	Energy Consumption of the 5GC		6.7.3.2.1
	Energy Consumption of a 5G Network Function		6.7.3.1.1
	Estimated Energy Consumption of a Virtualized Network Function		6.7.3.1.2
	Energy Consumption of a network slice		6.7.3.3
	Energy Consumption of a Physical Network Function (PNF) as well as other Power, Energy, Environmental (PEE) measurements	TS 28.552	5.1.1.19.2
Energy Efficiency KPIs	Energy Efficiency of the NG-RAN data	TS 28.554	6.7.1
	Energy Efficiency of the 5GC		6.7.4.1
	Energy Efficiency of a network slice		6.7.2

In the case of a Virtualized Network Function (VNF) hosted on a physical node, the energy consumption of the VNF is estimated as a portion of the total energy consumption of the physical node on which the VNF is executing, based on its relative virtual CPU usage, virtual memory usage, virtual disk usage and I/O traffic (all metrics collected from ETSI MANO) as defined in clause 6.3.1.2 of TS 28.554.

4.2.2.2.3 Exposure of network energy information by OAM p. 17

Network energy information may be collected by OAM and exposed (other mechanisms may exist) as defined by the Service-Based Management Architecture (SBMA) in TS 28.533. Any authorised consumer willing to collect such measurements or KPIs is first required by TS 28.533 to create an instance of a performance metrics production job (i.e., an instance of the PerfMetricJob information element - see clause 4.3.31 of TS 28.622) by invoking the createMOI operation of the Provisioning Management Service (MnS) (see clause 11.1.1.1 of TS 28.532).

The consumer is required by TS 28.533 to specify:

Which measurement(s) or KPI(s) it wishes to be collected, e.g. the energy consumption of a 5G NF, etc.
Which network entities (represented by managed objects) it wishes the information to be collected from (e.g., SMF x, UPF y, etc.).
The granularity period (expressed in seconds) it wishes the measurements or KPIs to be reported over.
The reporting method, mainly:
- File-based reporting: Performance data is accumulated for a certain time before it is reported; the data is delivered as a file; the file content encoding is either XML or ASN.1.
- Stream-based reporting: The producer sends the performance data to the consumer as when they are ready. The volume of the performance data reported is expected to be small, and the granularity period is expected to be short. The stream content encoding technique is either GPB or ASN.1.

Depending on the selected reporting method, the consumer collects the measurements or KPIs as follows:

In the case of file-based reporting, the producer sends the notification NotifyFileReady to subscribed consumer(s) when a new file becomes available on the producer for subsequent download by consumer(s).
In the case of stream-based reporting, the producer sends units of streaming data to the consumer by invoking the reportStreamData operation.

4.2.2.3 Collection and exposure of energy consumption information at NF p. 17

TR 23.700-66 studies and identifies potential enhancements to the 5G System (e.g., including network energy-related information exposure, enhancement for subscription and policy control to enable energy efficiency as a service criterion) to improve energy efficiency and to support energy saving in the network.

Three different key issues have been identified in that study:

KI#1: Network energy related information exposure
KI#2: Subscription and policy control to support energy efficiency and energy saving as service criteria
KI#3: 5GS enhancements for network energy saving and efficiency

KI#2 is not in scope of this study. The conclusions of KI#1 and KI#3 in clause 8 of TR 23.700-66 and the normative work following will be used for collection and exposure of Energy Consumption information at Network Functions (NFs) and are summarised as follows:

A new network functionality will be defined to collect and calculate energy-related information and expose it to authorised consumers subject to the network operator's policy:
- If the authorised consumer is a 5GC Network Function, the information exposure granularities that can be configured in this policy will include per application, per UE, per-UE-per-QoS Flow, per PDU session.
- If the authorised consumer is an Application Function, the information exposure granularities that can be configured in this policy will include: per UE, per UE per application, per PDU session.
The energy-related information that can be exposed according to the above exposure granularities will include:
- Energy Consumption information as defined in TS 28.310.
- Renewable energy information defined as energy from renewable non-fossil sources. For example (but not limited to) wind, solar, aerothermal, geothermal, hydrothermal.
A consumer of energy-related information (i.e., 5GC NF or AF) may request different modes of exposure (e.g. periodic reporting or threshold-based reporting) as part of its subscription request.
The new network functionality supporting the calculation of the Energy Consumption information includes the following aspects:
1. OAM: provides the NF/Node-level Energy Consumption information at the gNodeB(s) and UPF(s) serving the UE.
2. OAM: provides the overall data volume of the gNodeB.
3. The information of a) and b) received from OAM could be used by the new network functionality for all the UEs served by the NF/Node.
4. UPF: provides the data volume for the QoS Flow or the Service Data Flow (SDF).
5. When the gNodeB and/or the (I-)UPF(s) which are serving the UE change, the serving gNodeB ID and UPF ID will be sent to the new network functionality through AMF/SMF.
The new network functionality determines the end-to-end energy consumption based on energy consumption per the granularities above at the serving Network Function (i.e. NG-RAN and UPF).
In Release 19, only the energy-related information of user plane communication (not control plane signalling) is supported.
Enhancements to NF discovery and (re-)selection procedures based on energy-related information:
- The NF profile may be extended (e.g. by including the new energy-related information or by reusing existing NF profile parameters) to allow an operator to influence NF discovery and selection based on its energy strategy.
- NF discovery and (re-) selection will be enhanced to consider the energy-related information from the NF profiles and/or discovery requests from the NF consumer.
Enhancements to existing operations and procedures for energy saving and energy efficiency:
- The User Plane path of a PDU session may be adjusted so that it consumes less energy.

The recommendations of the present document focusing on media services will need to be aligned with the conclusions in clause 8 of TR 23.700-66 impacting Application Functions used for media services.

As specified in clauses 5.51 and 6.2.34 of TS 23.501, the new network functionality referred to in point 1 above is called the Energy Information Function (EIF) and has the following responsibilities:

Collect data from OAM and 5GC Network Function(s) to assist in the calculation of energy-related information.
Calculate the energy-related information (including energy consumption information and renewable energy information) of user plane communication.
Expose the calculated energy-related information to authorised consumers.
Expose the energy-related information to authorized northbound consumers via the NEF.

4.2.2.4 UE data collection, reporting and event exposure p. 18

4.2.2.4.1 UE data collection, reporting and event exposure architecture p. 18

A generic architecture for the collection and reporting of UE data is defined in TS 26.531 and the corresponding APIs are specified in TS 26.532 but UE energy consumption has not been considered in these specifications up to and including in Release 18. Subject to study in the present document, a potential solution would be to expand the scope of these specifications to support the collection and exposure of energy consumption information at UE.

The main principle of the reference architecture defined in clause 4 of TS 26.531 and reproduced in Figure 4.2.2.4.1-1 below is the addition of an intermediary Application Function named the Data Collection AF which is used collected UE data reports from data collection clients and Application Servers, and to synthesise from those reports a set of events which are exposed to event consumer subscriber, such as the Network Data Analytics Function (NWDAF) or an Event Consumer AF deployed in the Application Service Provider.

Copy of original 3GPP image for 3GPP TS 26.942, Fig. 4.2.2.4.1-1: Reference architecture for data collection and reporting

Figure 4.2.2.4.1-1: Reference architecture for data collection and reporting
(⇒ copy of original 3GPP image)

Data collection and reporting functionality in the Data Collection AF is provisioned at reference point R1 by a Provisioning AF of the Application Service Provider that may be deployed either inside or outside the trusted domain. The purpose of the Data Collection AF is to receive UE data reports from one of three possible sources:

Directly from the UE. In this case the Direct Data Collection Client is responsible for collecting relevant data in the UE (typically from a UE Application using a suitable API at reference point R7) and for sending data reports to the Data Collection AF via reference point R2.
Indirectly from the UE. In this case, an Application Service Provider collects data from UE Applications privately via reference point R8 and employ an Indirect Data Collection Client subfunction to then send data reports to the Data Collection AF via reference point R3.
From an Application Server that has been used to deliver media to/from a UE. Application Server instances (AS) inside or outside the trusted domain may also collect data and report it to the Data Collection AF via reference point R4.

The Data Collection AF aggregates and filters UE data that is reported to it. The processed UE data is exposed by the Data Collection AF to the NWDAF in the form of data reporting event notifications via reference point R5. Certain UE data may also be exposed in the form of data reporting events by the Data Collection AF to an Event Consumer AF residing in the Application Service Provider via reference point R6.

When they are deployed in different trust domains, the interactions between the system actors of the UE data collection, reporting and event exposure architecture may be mediated through the NEF, as illustrated by various collaboration scenarios defined in annex A of TS 26.531.

4.2.2.4.2 UE data collection, reporting and event exposure for 5G Media Streaming p. 20

The instantiation of the UE data collection, reporting and event exposure architecture in the 5GMS System is defined in clause 4.7 of TS 26.501 and the reference architecture for this instantiation is reproduced in Figure 4.2.2.4.2-1.

Copy of original 3GPP image for 3GPP TS 26.942, Fig. 4.2.2.4.2-1: Data collection and reporting architecture instantiation for 5G Media Streaming

Figure 4.2.2.4.2-1: Data collection and reporting architecture instantiation for 5G Media Streaming
(⇒ copy of original 3GPP image)

Three existing 5GMS reference points are reused in this instantiation: M1 (for provisioning UE data collection, reporting and event exposure), M5 (for media session handling) and M6 (for interacting with the Media Session Handler).

The Provisioning AF of the Application Service Provider is not instantiated in the 5GMS architecture. Data collection and reporting is instead provisioned using the procedures using M1 defined in TS 26.501.
The Data Collection AF for 5G Media Streaming is instantiated in the 5GMS AF.
The Direct Data Collection Client for 5G Media Streaming is instantiated in the Media Session Handler. This takes logical responsibility for the UE data collection activities of the Metrics Collection & Reporting and Consumption Collection & Reporting subfunctions and the subsequent reporting of this UE data via reference point M5. It also takes logical responsibility for the logging of ANBR-based Network Assistance invocations by the Network Assistance subfunction and their subsequent reporting to the Data Collection AF instantiated in the 5GMS AF via reference point R2.

NOTE:
The use of M5 instead of R2 is due to the definition of some procedures before the definition of this instantiation. M5 was already used for QoE metrics and consumption reports, or invocations of the downlink dynamic policy procedures and invocations of the AF-based downlink Network Assistance procedures. Thus, for these procedures, M5 is used instead of R2.
The Indirect Data Collection Client is not instantiated in the 5GMS architecture. Indirect reporting of UE data is outside the scope of 5G Media Streaming. Thus, R8 is not instantiated in the 5GMS architecture.
The role of the AS in the abstract reference architecture is played by 5GMS AS. (This may be deployed as a trusted AS within the 5G System or deployed externally.)
The Event Consumer AF is instantiated in the 5GMS Application Provider as a consumer of 5G Media Streaming events from the Data Collection AF.
Reference point R7 is not instantiated in the 5GMS architecture. Configuration of 5GMS-related data reporting in the Media Session Handler by the 5GMS-Aware Application is managed through the existing media session handling client API at reference point M6.

4.2.2.5 QoE Measurement Collection (QMC) functionality p. 21

TS 38.300 defines the QoE Measurement Collection (QMC) feature which enables collection of application layer measurements from the UE. QMC is supported for the following service types in NR cells:

QoE Measurement Collection for DASH streaming services in TS 26.247;
QoE Measurement Collection for MTSI services in TS 26.114;
QoE Measurement Collection for VR services in TS 26.118.

The QMC feature also supports collection of QoE measurements for any of the supported service types carried by the MBS communication service defined in TS 23.247, namely:

MBS broadcast;
MBS multicast.

More details of QMC control and configuration can be found in TS 28.405. A potential solution would be to reuse and expand the QMC functionality to support the reporting of energy consumption information by the UE.

4.2.3 Other Standards Development Organisations p. 21

4.2.3.1 ITU-T p. 21

Within the International Telecommunication Union, the T-sector includes Study Group 5 "Environment and Circular Economy" (SG5). Part of its mandate is to define and develop "methodologies for evaluating ICT effects on climate change and publishing guidelines for using ICTs in an eco-friendly way. Under its environmental mandate SG5 is also responsible for studying design methodologies to reduce ICT's and e-waste's adverse environmental effects, for example, through recycling of ICT facilities and equipment."

Among its activities, ITU-T Study Group 5 is developing technical reports, supplements and recommendations for the environmental requirements of 5G.

Recommendation ITU-T L.1310 [29] contains the definition of energy efficiency metrics, test procedures, methodologies and measurement profiles required to assess the energy efficiency of telecommunication equipment. Energy efficiency metrics and measurement methods are defined for telecommunication network equipment and small networking equipment. These metrics allow for the comparison of equipment within the same class, e.g., equipment using the same technologies.
ITU-T L.1310 Supplement 36 [30] analyses the energy efficiency issues for 5G systems. The focus of this supplement is on methods and metrics used to measure energy efficiency in 5G systems with multi-radio equipment.

Further, the L.1400 series of reports and recommendations present methodologies and guidelines for the assessment of the greenhouse gas emissions and energy consumption of the ICT sector. For example:

Recommendation ITU-T L.1450 [31] presents a methodology for the assessment of the impact of telecommunications systems. It was used in an assessment of the electricity usage and greenhouse gas emissions of the ICT sector [32].

4.2.3.2 ITU-R p. 22

The remit of ITU-R Study Group 6 (SG6) is programme production and interchange. Its Working Party 6A (WP 6A) has published ITU-R Report BT.2385 "Reducing the environmental impact of terrestrial broadcasting systems" [33]. Working Party 6C (WP 6C) has a rapporteur group which has produced the following documents:

ITU-R Opinion 104, "Advice for sustainability strategies incorporating carbon offsetting policies" [34]
ITU-R Report BT.2521, "Practical examples of actions to realize energy aware broadcasting" [35]. This report is based on a webinar held in March 2022.
ITU-R Report BT.2540, "Display energy reduction through image signal processing" [36]. This document describes techniques for producing, transmitting and using metadata which enables display devices to use less energy.

4.2.3.3 MPEG p. 22

The ISO/IEC JTC 1/SC 29 committee Coding of audio, picture, multimedia and hypermedia information has published the ISO/IEC 23001-11:2023 (Green MPEG) standard [37]. The various components of the standard define methods for the reduction of the power consumption of decoders and of displays. A further component defines a method for the selection of energy-efficient media. A final method allows for quality recovery after low-power encoding. The standard is currently in revision, and it is to be extended to enable the carriage of metadata to more efficiently reduce the power requirements of display devices receiving the content with the metadata.

4.2.3.4 DVB p. 22

DVB has carried out a study mission to assess the potential for developing energy-efficient video transmission systems. This work has resulted in the creation of a new CM-EE (Energy Efficiency) working group in its Commercial Module. It has also published a report on the topic available as Blue Book S100 "Study Mission report on Energy Aware service Delivery and Consumption" [38].

4.2.3.5 ATSC p. 22

ATSC's "Planning Team 9 - Sustainability in Media and Data Delivery Services (PT9) will study the benefits of broadcast data delivery as relates to sustainable energy usage in a world increasingly dependent on data delivery. The team will consider linear and file-based media delivery as well as linear and file-based data delivery. PT9 will report the results of this work to the Board. If technical work in ATSC is recommended, PT9 will further document rationale for the work and ideally also document possible architectural approaches and requirements, such as interoperability with existing networks, which would accommodate the identified use cases. PT9 does not draft standards or recommended practices; it may draft New Project Proposals and/or Planning Team Reports. PT9 reports to the ATSC Board of Directors and participation is open to all ATSC members." [39].

4.2.3.6 ETSI p. 23

4.2.3.6.1 Summary of energy efficiency standards drafted by the ETSI Environmental Engineering (EE) Working Group p. 23

Table 4.2.3.6.1-1 below shows a summary of energy efficiency standards developed by the ETSI Working Group on Environmental Engineering (EE). The list is non-exhaustive.

Table 4.2.3.6.1-1: List of ETSI Environmental Engineering (EE) specifications

Standard	Summary
ETSI ES 202 706-1 [4]	Defines the measurement method for the evaluation of base station power consumption and energy consumption with static load. The methodology described in this specification is to measure base station static power consumption and RF output power. Within the document it is referred to as "static" measurements. The results based on "static" measurements provide power and energy consumption figures for a Base Station under static load.
ETSI ES 203 700 [41]	Defines power feeding solutions for 5G, converged wireless and wireline access equipment and network, taking into consideration their enhanced requirements on service availability and reliability, the new deployment scenarios, together with the environmental impact of the proposed solutions. The minimum requirements of different solutions including power feeding structures, components, backup, safety requirements, environmental conditions are also defined.
ETSI ES 203 539 [42]	Defines energy efficiency metrics and measurement methods for NFV components including VNFs and NFVI. The energy efficiency of VNF is evaluated according to hardware energy consumption, resource consumption and utilization related with VNF. The energy efficiency of NFVI is evaluated as resource provision capability which is expressed as service capacity of reference VNFs running on it with amount of energy consumption.
ETSI EN 303 470 [43]	Specifies a metric for the assessment of energy efficiency of computer servers. Formalizes the tools, conditions and calculations used to generate a single figure of merit of a single computer server representing its relative efficiency and power impact. The metric is targeted for use as a tool in the selection process of servers to be provisioned.
ETSI EN 303 471 [44]	Specifies the method and metrics to determine the energy efficiency of operational Network Function Virtualisation (NFV) applications and their associated infrastructure. It specifies the method and metrics to determine the energy efficiency of operational Network Function Virtualisation (NFV) applications and their associated infrastructure when that infrastructure is implemented outside the boundaries of the access fixed, cable and mobile networks which they serve.
ETSI ES 203 475 [45]	Specifies terminology, principles and concepts for Energy efficiency and energy management. It aims to establish a common understanding of measurement methodology used to determine the energy efficiency of a good, service and network. It presents a framework for other ETSI standards and other Standard Development Organization documents about Energy Efficiency.
ETSI ES 203 136 [46]	Defines the energy consumption metrics and measurement methods for packet routing and Ethernet switching equipment. It defines the methodology and the test conditions to measure the power consumption of a router or switch. It is applicable to core, edge and access routers. Home gateways are out of scope.
ETSI EN 303 215 [47]	Defines power consumption metrics, methodology and test conditions to measure the power consumption of broadband fixed telecommunication network equipment. It does not cover all possible configuration of equipment, but only homogenous configurations. The types of broadband access technologies covered are: DSLAM DSL, MSAN, GPON OLT and point-to-point OLT equipment.
ETSI ES 202 706 [4]	Defines methods for evaluation of power consumption and energy efficiency of base station in static and dynamic mode. The methodology described is to measure base station static power consumption and dynamic energy efficiency, which are referred to as static and dynamic measurements respectively. The results based on "static" measurements of the Base Station power consumption provide a power consumption figure for the Base Station under static load. The results based on "dynamic" measurements of the Base Station provide energy efficiency information for a Base Station with dynamic load.
ETSI ES 203 184 [48]	Defines the metric, methodology and the test conditions to evaluate the Equipment Energy Efficiency Ratio (EEER) of Transport equipment, including all the transmission equipment connected to the network by means of wired medium (i.e. copper or fibre), typically running at the network OSI Layer 1. The present document also covers the equipment running at the network OSI Layer 2 (e.g. MPLS-TP) that are not included in the ETSI standard on "Measurement Methods for Energy Efficiency of Router and Switch Equipment" (the same approach is followed by ATIS standard on Transport equipment. Examples of typical wired Transport equipment covered by the present document are switches or crosses connects (SDH, OTN) and add/drop multiplexers (DWDM). The present document covers also simpler systems as multiplexers/demultiplexers (DWDM), optical amplifiers, transponders.
ETSI EN 301 575 [49]	Defines energy consumption measurement methods for Broadband CPE telecommunication equipment. Also defines a methodology and test conditions to measure the power consumption of end-user broadband equipment.
ETSI ES 203 215 [50]	Defines energy consumption limits and measurement methods for fixed broadband telecommunication network equipment. Also defines power consumption limits, a methodology and test conditions to measure the power consumption of broadband fixed telecommunication networks equipment. The types of broadband access technologies covered are: DSLAM DSL, MSAN, GPON OLT, Point to Point OLT equipment.

4.2.3.6.2 Definition of Mobile Network Energy Efficiency p. 24

ITU-T L.1310 [29] defines energy efficiency as the relationship between the specific functional unit for a piece of equipment (i.e., the useful work of telecommunications) and the energy consumption of that equipment. For example, when transmission time and frequency bandwidth are fixed, a telecommunication system that can transport more data (in bits) with less energy (in Joules) is considered to be more energy-efficient. For this reason, metrics that can evaluate the performance of a piece of equipment against its energy consumption are to be defined.

From Release 15 onwards, the definition of Energy Efficiency is clarified in 3GPP. The definition does not come directly from 3GPP itself, but rather is adopted from the ETSI Working Group on Environmental Engineering, in ETSI ES 203 228 [66] which aims to define the topology and level of analysis to assess the energy efficiency of mobile networks. In particular, [66] defines metrics for mobile network energy efficiency and methods for assessing (and measuring) energy efficiency in operational networks.

Per ETSI ES 203 228 [66], Energy Efficiency (EE) of a Mobile Network is defined as the relation between the useful output and power consumption, where power consumption is defined as the power consumed by a device to achieve an intended application performance.

Mobile Network data Energy Efficiency EE_NE is the ratio between the performance indicator Data Volume (DV_MN) and the Energy Consumption (ECMN) when assessed during the same time frame (T) as defined in clause 7.1 of ITU-T recommendation L.1331 [67]. This is also shown by the formula:

EE_NE = DV_MN / EC_MN

where DV is the Data Volume, expressed in bits, transported across a network element. The Data Volume measurements are collected via OAM. EC is the Energy Consumption, expressed in Joules, of the same network element. The MN suffix stands for Mobile Network.

This formula is reproduced in several 3GPP Technical Specifications and Technical Reports dealing with energy efficiency (EE).

Clause 8.2 of ITU-T L.1331 [67] illustrates how to measure/collect the information about data volume (for capacity), coverage area (for coverage) as well as energy consumption over a measurement period called T, spanning one week, one month, or longer periods.

In addition, Annex T of TS 23.501 provides examples of how the Energy Information Function (EIF) described in clause 4.2.2.3 calculates the energy consumption for the required granularities.

4.2.4 Industry fora p. 25

4.2.4.1 Greening of Streaming p. 25

Greening of Streaming is a member association investigating energy efficiency in the context of media streaming applications [51]. One of the challenges the group is aiming to address is that of accurately measuring the energy expenditure of streaming services, given that currently the available data is sparse and not very precise. It further intends to define best practices.

4.2.4.2 DIMPACT p. 25

"DIMPACT is a collaborative initiative between leading media, entertainment and technology companies and world-class researchers" [52]. The group is convened by Carnstone Partners Ltd, and research and technical expertise is provided by researchers from the University of Bristol. It has currently over 20 members. The group has developed a tool to measure the emissions of serving digital media and entertainment products. This tool is available as a web application and is able to estimate emissions originating from video streaming, online banner advertising, digital publishing, and audio streaming. The DIMPACT website makes available several publications explaining their methodology [53] and defining principles for streaming and digital media carbon footprinting [54].

4.2.4.3 Ultra HD Forum p. 25

The Sustainability Working Group of the Ultra HD Forum is investigating energy efficiency opportunities throughout content distribution, from content encoding through distribution and display. One result has been a regular series of public demonstrations directed to sustainability at major broadcast conferences (specifically IBC and NAB) where the Ultra HD Forum regularly has a booth [55]. Some of these demonstrate the degree of influence the consumer can have on the energy consumption by the display.

4.2.5 Greenhouse Gas Protocol p. 25

The Greenhouse Gas Protocol [9] is a joint initiative of the World Resources Institute and the World Business Council for Sustainable Development (WBCSD) that "provides standards, guidance, tools and training for business and government to measure and manage climate-warming emissions." The first edition of their reporting standard was published in 2001, establishing a reporting framework for businesses. Relative to a given company, the concept of "scopes" is introduced, which delineate direct and indirect emission source, and are used for accounting and reporting purposes. The reporting principally involves the six greenhouse gasses that are defined in the Kyoto protocol: carbon dioxide (CO₂), methane (CH₄), nitrous oxide (N₂O), hydrofluorocarbons (HFCs), perfluorocarbons (PFCs) and sulphurhexafluoride (SF₆). Scopes 1 (sources owned or controlled by a company giving rise to direct greenhouse gas) and 2 (the electricity purchased and consumed by a company gives rise to greenhouse gas emissions) are defined in [10], and Scope 3 (All other indirect emissions) is defined in [11].

4.2.5.1 Scope 1 p. 25

Sources owned or controlled by a company give rise to direct greenhouse gas emissions. The activities undertaken by a company that give rise to scope 1 emissions include the generation of electricity, heat, or steam; physical or chemical processing; transportation of materials, products, and waste; fugitive emissions.

4.2.5.2 Scope 2 p. 25

The electricity purchased and consumed by a company gives rise to greenhouse gas emissions. Scope 2 emissions occur at the facility where the electricity is generated, rather than where the electricity is consumed. For the reporting company, these emissions are therefore counted as one form of indirect emissions. As purchased electricity is for many companies one of the largest sources of greenhouse gas emissions, it also offers a significant potential for reductions, either by investing in energy efficient technologies, by energy conservation, or by switching to less greenhouse gas intensive sources of electricity.

4.2.5.3 Scope 3 p. 26

All other indirect emissions can be reported under scope 3. The emissions reported in this category are the consequence of the activities of a company, but they come from sources not owned or controlled by this company. These indirect emissions arise elsewhere in the corporate value chain of a given company, both upstream and downstream. Upstream emissions are indirect emissions relating to purchased or acquired goods and services. Downstream emissions relate to sold goods and services. Scope 3 indirect emissions are categorised into 15 distinct categories, of which the upstream categories are:

Purchased goods and services: Extraction, production, and transportation of goods and services purchased or acquired by the reporting company in the reporting year, not otherwise included in categories 2 to 8.
Capital goods: Extraction, production and transportation of capital goods, purchased or acquired by the reporting company in the reporting year.
Fuel- and energy-related activities (not included in scope 1 or scope 2): Extraction, production, and transportation of fuels and energy purchased or acquired by the reporting company in the reporting year.
Upstream transportation and distribution: Transportation and distribution of purchased products or services in the reporting year, including inbound and outbound logistics; transportation between a company's own facilities.
Waste generated in operations: Disposal and treatment of waste generated by the reporting company's operations in the reporting year.
Business travel: Transportation of employees for business-related activities in the reporting year in vehicles not owned or operated by the reporting company.
Employee commuting: Transportation of employees between their homes and their worksites during the reporting year, in vehicles not owned or operated by the reporting company.
Upstream leased assets: Operation of assets leased by the reporting company in the reporting year.

The downstream categories are:

Downstream transportation and distribution: Transportation and distribution of products sold by the reporting company in the reporting year between the reporting company's operations and the end consumer, if not paid for by the reporting company (in vehicles and facilities not owned by the reporting company).
Processing of sold products: Processing of intermediate products sold in the reporting year by the downstream companies.
Use of sold products: End use of goods and services sold by the reporting company in the reporting year.
End-of-life treatment of sold products: Waste disposal and treatment of products sold by the reporting company (in the reporting year) at the end of their life.
Downstream leased assets: Operation of assets owned by the reporting company and leased by other companies in the reporting year.
Franchises: Operation of franchise in the reporting year.
Investments: Operation of investments (including equity and debt investments and project finance) in the reporting year.

4.2.6 Report on carbon impact of video streaming p. 26

After significant debate over the environmental footprint of video streaming, a group of broadcasters and media companies working with the DIMPACT consortium (see clause 4.2.4.2) and the Carbon Trust published in 2021 a widely agreed estimate of hourly emissions associated with video streaming [80] that considers devices capable of streaming across both mobile and broadband networks. This work resulted from a previously established methodological framework developed by BBC Research & Development and the University of Bristol to measure broadcast and streaming energy.

The report considers consumption of video streaming services on different combinations of representative devices, including home routers, TV peripherals (e.g. Set Top Box) and end user terminal devices including TV, Laptops, Desktop, Tablets and Mobile phones. The overall video traffic is normalised for an hour of viewing at a representative end user device after normalising average time spent on different devices. There are some specific use cases that have been specifically demonstrated for their streaming energy. These include:

Smartphone (Apple iPhone) connected to mobile network.
Laptop computer connected to a home router via Wi-Fi.
50" smart TV set connected to a home router via Wi-Fi.

The methodology normalises overall streaming across mobile and broadband networks and assigns energy consumption per unit data to each network, expressed in kWh/GByte. The methodology also averages the energy consumption in proportion to the viewing time on the different types of end user terminal device considered.

Table 4.2.6-1 summarises the average energy consumption per hour of video streaming across all end user terminal devices considered. For convenience, this figure is also converted to the equivalent grams of carbon dioxide that would be emitted based on a representative energy mix for Europe in 2020.

Table 4.2.6-1: Energy consumption and emissions associated with video streaming process components for Europe in 2020 [80]

Video streaming component stage	Streaming energy consumption (Wh/hour)	Streaming emissions (gCO₂e/hour)	Proportion of total
Data centres	1	< 1	1%
Transmission network	20	6	10%
Home router	71	21	38%
TV peripheral	10	3	5%
Screens	86	25	46%
Total	188	56	100%

Taking an average EU energy mix and a representative device considering the overall content distribution chain, the report estimated that an hour of video streaming consumes 188 Wh of energy or ~56 gCO₂ e per hour of video streaming. In particular:

Data centres (including hosting, encoding and CDNs) account for less than 1 gCO₂e/hour based on energy consumption of approximately 1 Wh/hour, representing roughly 1% of total energy and therefore emissions.
Network transmission (core and access) accounts for 6 gCO₂e/hour and 20 Wh/hour (10% of total energy and therefore emissions).
Home routers account for 21 gCO₂e/hour and 71 Wh/hour (38% of total energy and therefore emissions).
End-user devices account for 28 gCO₂e/hour (25 gCO₂e/hour from viewing devices and 3 gCO₂e/hour from peripherals) based on energy consumption of 96 Wh/hour (86 Wh/hour from screens and 10 Wh/hour from peripherals).

Results confirmed that devices in the home - including home routers and TV peripherals as well as the end user terminal devices themselves - consume a significant proportion of total energy and account for almost 89% of emissions across the whole streaming value chain. Depending on the choice of user device, the emissions associated with an hour of video streaming can vary from 8 gCO₂e/hour for a smartphone to 16 gCO₂e/hour for a laptop computer or 58 gCO₂e/hour for a 50" smart TV set.

Several limitations of these findings are highlighted in [80], including lack of representative data, attribution of network energy over millions of consumers and the methodological assumptions to simplify the modelling.

4.2.7 Energy estimates for broadcast and streaming energy consumption in the UK p. 28

Ofcom, the UK communications regulator published in 2022 through Carnstone its own energy estimates for broadcast and streaming energy consumption in the UK [82] using the previously established methodological approach developed by BBC Research & Development referred to in clause 4.2.6. The estimates compare the energy footprint of an hour of viewing media on a representative user device across the entire distribution value chain. The analysis was initially carried out for the year 2019 and was later updated for 2021. Even though the scale of energy consumption was similar in magnitude to that of the EU (as described in clause 4.2.6), hourly energy consumption for video streaming was estimated to be lower at 113 Wh per hour of streaming. This was equivalent to ~33 gCO₂e/hour streaming on a representative viewing device for the UK.

Table 4.2.7-1: Energy and Emission footprint of an hour of video streaming in the UK

Video streaming component stage	Streaming energy consumption (Wh/hour)	Streaming emissions (gCO₂e/hour)	Proportion of total
Network transmission	12.0	3.5	10.8%
Customer Premises Equipment	21.5	6.2	18.6%
Peripherals	21.5	6.2	18.6%
Viewing devices	58.0	17.0	51.3%
Total	~113.0	33.0	~100.0%

Like previous studies, the Ofcom report [82] confirmed that a considerable proportion of the total energy consumption - approximately 89% - occurred within the home, concentrated around viewing devices and in-home networks.

Methodological and data-related concerns are similar to those described at the end of clause 4.2.6, necessitating further efforts for improved and robust measurements.

4.2.8 Study on predicted environmental impact of audiovisual media consumption in France p. 28

In 2022, the audiovisual communications regulator for France, Arcom, and its communication networks regulator, Arcep, published a joint study [61] on the predicted environmental impact of audiovisual media consumption in France over the period from 2022 to 2030. The study assesses the environmental impact of consuming audiovisual mass media, taking all the component parts into account: hardware (user devices), networks (fixed broadband and superfast broadband, digital terrestrial, and satellite) and data centres. Its scope includes the main systems used to access audiovisual media: linear and time-shifted television and radio, audio and video streaming (including video-on-demand services), and video sharing platforms. Every type of impact has been assessed (carbon footprint, consumption of mineral and metal resources, final energy consumption) including energy usage, the target for the framework of the report.

In this study, energy usage (termed final energy consumption) is measured in kilowatt-hours (kWh) and refers to the quantity of electricity consumed during the usage phase of the three tiers of the digital value chain (user devices, networks and data centres). It concerns itself only with the usage stage of terminals, networks, and data centres; upstream electricity consumption for the manufacturing phase is not addressed by this indicator.

A comparative assessment of nine audiovisual usage scenarios (on the scale of one hour of audio or video content consumption in France in 2022) is considered in the report:

A1: Listening to live FM radio on a radio set
A2: Listening to live FM radio on a car radio
A3: Listening to live radio via the Internet on a smartphone connected to the fixed network
A4: Listening to music/podcast on a streaming platform (app) on a smartphone connected to the Internet via mobile network
V1: Watching a TV channel in HD on a television via integrated Digital Terrestrial Television (DTT) access
V2: Watching a TV channel in HD on a television connected to the Internet via a TV decoder linked to an ISP box (managed IPTV)
V3: Watching catch-up TV in HD on a smart TV connected to the Internet via a TV decoder linked to an ISP box
V4: Watching Subscription Video-on-Demand (SVoD) in HD on a smart TV connected to the Internet via fixed network
V5: Watching online videos on a video sharing platform in HD on a smartphone connected to the Internet via mobile network

To estimate energy consumption of devices, four different devices have been evaluated under laboratory test conditions (two smartphones, one PC and one smart TV set).

For the smart TV and the PC, a measurement module (digital watt meter) is inserted between the device and the mains power outlet. This module measures energy consumption in Alternating Current (AC). The watt meter is connected to a computer to record the energy consumption measurements.
For smartphones, measurements are taken using software probes to record energy and data consumption.

Energy is measured in units of milliwatt-hours per second (mWh/s) or milliamp-hours (mAh). The measurements are sampled for a period of one minute. Several iterations are performed (a minimum of three samples) to ensure relevance and to limit artifacts related to the measurement itself. Testing conditions are noted for traceability of the measurements.

Two measurement modes are possible:

Systematic content change between iterations: This measurement mode has the advantage of eliminating the effects of content caching strategies in the terminal device or delivery network but has the disadvantage of introducing variability. However, this measurement mode is more representative of real-world user behaviour.
Iterations are conducted on a continuously played video: This measurement mode has the advantage of controlling for the variability of content but has the disadvantage of potentially underestimating consumption due to caching technologies.

The systematic content change solution is favoured in the scenario V5 (video sharing platforms). On the other hand, the continuous video strategy is used when it is useful to control for the content's impact and to study certain parameters (such as video codec).

Given the diversity of hardware studied, it was decided that the user journey would not be automated.

The data measured under laboratory test conditions are very specific. They are conducted on a single device (two for smartphones), which performs a single precise usage. This allows, for example, consumption during content playback to be differentiated from browsing a content catalogue. However, these measurements are not necessarily representative of the entire equipment landscape. Thus, comprehensive and representative data from the literature on a more diverse equipment pool were preferred over certain laboratory measurements for the quantification of audio and video usage at the national level in France.

In the context of the present document:

The method to estimate the energy consumption of the mobile network described in [61] is not reusable because it uses a theoretical calculation based on the total amount of energy consumed by the mobile network, the mobile network usage duration per subscriber and a formula allocating energy consumption per subscriber per year and per data volume.
The method to estimate the energy consumption of data centres described in [61] is not reusable either because it is based on external estimates.
The method to estimate the energy consumption of a UE described in [61] could be used as a basis for evaluating the energy usage/savings of multimedia standards features and proposals on UEs.

4.3 Challenges in accurately estimating energy consumption p. 30

Even though several regulators, academics and media organisations have examined and published energy estimates, significant challenges remain in accurate and robust estimates of energy consumption as noted in [73] and [79].

The primary challenge so far has been around obtaining accurate power data at the hardware level for mobile and fixed networks, and for end user devices. Understanding how the relationship between data and energy plays out with data throughput and peaks is another point of concern. Accurate energy consumption data by both server hardware and network components remains difficult to achieve in practice.

The other big challenge is the attribution of energy for specific data throughput on the existing value chain and the allocation of this energy between stakeholders benefitting from network connectivity. The issue here is that hardware elements service multiple users at the same time, and in addition the relationship between energy and data throughput is not linear due to an often significant base load incurred by having the equipment switched on in the first place. Legal and regulatory pressures to achieve net zero emissions make allocation a key business strategy requirement. Agreements between network operators and their consumers in this regard are contentious due to responsibilities and costs of achieving net zero. System boundaries for energy measurements are also quite important in this discussion and where to draw the line can often influence the onus for responsibility.

Further, while individual elements of a transmission system could in principle be measured and instrumented, for reporting applications it would be advantageous to be able to determine the energy used to deliver a given media item from source to sink, i.e. as the data passes through a variety of network elements.

The implications of these challenges are that only relatively coarse-grained assessments of energy use can currently be made. These assessments are typically seen as sufficient for policy making, but are insufficient for understanding the energy bottlenecks in existing systems, or for understanding how to design more efficient systems. Further, current practices are insufficient for reporting duties as standardised by the Greenhouse Gas Protocol, and as required by the European Union and the State of California (Scope 3 reporting - see clause 4.2.5). The collection of fine-grained energy consumption data in (near-)real time from media applications running on user devices and Application Servers would contribute towards solving these issues. The present document can help to define a futureproof measurement framework to assist in this aim.