Content for TR 26.942 Word version: 19.0.0
7.10 Solution #9: UE energy metrics abstraction
7.10.1 Key Issue mapping
7.10.2 Functional description
7.10.2.1 Introduction
7.10.2.2 Energy-related information relevant to the UE
7.10.2.3 Estimating the energy usage of media applications running on the UE
7.10.2.4 UE energy information index abstraction
7.10.3 Procedures
7.10.4 Summary
...
...
7.10 Solution #9: UE energy metrics abstraction p. 63
7.10.1 Key Issue mapping p. 63
This Candidate Solution addresses Key Issue #1.
7.10.2 Functional description p. 63
7.10.2.1 Introduction p. 63
Clause 6.15a.5 of TS 22.261 defines information exposure, which clearly states that information related to energy consumption and efficiency is not only necessary for network internal optimization, but also will benefit the service adjustment for third parties such as Application Service Providers. This Candidate Solution to Key Issue #1 determines the method of obtaining this additional energy information with the objective of measuring the effects of service adjustments on a single UE at a time. A summary is provided of energy information relevant to the UE that can be exposed by the UE to external entities in an abstract way, without exposing any internal information relating to the UE.
7.10.2.2 Energy-related information relevant to the UE p. 63
Table 7.10.2.2-1 lists several device properties which are UE energy information. Alongside the entries defined by the deviceCapabilities object in Table 8.4.2.2-1 of TS 26.565, these may be useful in calculating a consolidated UE energy metric.
| Property | Description |
|---|---|
| UE battery life | Level of the battery of the UE (e.g. in %). Based on an estimate of how long the battery will power the device on a single charge. In simple terms, it indicates how long the device can run before needing a recharge which, depending on usage, can be a matter of hours or an entire day. |
| UE energy consumption (EC) | UE charging/discharging rate (e.g. in Watt-hours, negative or positive). Calculated as the integral of power consumption over time as defined in TS 28.310. |
| UE energy saving duration preference | Remaining time during which the UE wants to be considered in energy saving (e.g. in minutes). |
| UE battery capacity | The total UE battery capacity (e.g. in mAh). The total amount of electricity generated due to electrochemical reactions in the battery and is expressed in ampere hours. For example, a constant discharge current of 1 C (5 A) can be drawn from a 5 Ah battery for 1 hour [86]. |
| UE source of power supply | e.g. "battery", "plugged in". Indicates whether the UE is currently operating on battery or is plugged in to an external power supply, which may or may not be renewable energy (e.g., solar panel). This could include the ratio of renewable energy over different time granularities (e.g. plugged in 30% renewable over the last 24 hours as described in TS 28.310). |
| UE battery discharge rate | e.g. between 0 and 1000. Calculated as the battery capacity (in mAh) divided by the number of hours it takes to /discharge the battery. For example, a battery capacity of 500 Ah that is theoretically discharged to its cut-off voltage in 20 hours will have a discharge rate of 500 mAh/20 h = 25 mA [86]. |
| UE measurement duration | Energy Measurement time interval. |
| UE carbon intensity | In g CO2-e/Wh. A measure of the global greenhouse gases emitted per unit of generated electricity, measured in grams of CO2 equivalents per watt-hour (for conversion to carbon emissions as defined in TS 22.261 and TS 23.700-66). |
7.10.2.3 Estimating the energy usage of media applications running on the UE p. 64
The Accubattery application [87] is an example, where we can understand that it allows the user to visualise the battery usage of each individual application installed on an Android UE, as illustrated in Figure 7.10.2.3-1 below.
![Copy of original 3GPP image for 3GPP TS 26.942, Fig. 7.10.2.3-1: Example of UE application showing per-application battery consumption (source: [87])](../img/tinv-26-942-7.10.2.3-1.gif)
Figure 7.10.2.3-1: Example of UE application showing per-application battery consumption (source: [87])
(⇒ copy of original 3GPP image)
(⇒ copy of original 3GPP image)
Similar techniques could be used to determine the battery consumption of individual media applications as input to the calculation of the UE energy-related information index.
7.10.2.4 UE energy information index abstraction p. 65
UE energy information is private UE information, and it is possible that UE manufacturers/chipset vendors do not want to share this raw information with third parties, network operators or with the Application Service Provider. Hence, this Candidate Solution proposes a method which allows to deduce an abstract UE energy information based on internal UE energy related characteristics such as its battery capacity, battery discharge rate, etc. This calculation and implementation of the UE energy index is left to individual UE manufacturers and/or chipset vendors. But this index value is abstracted such that it does not reveal any internal properties of the UE such as its battery capacity, battery discharge rate, etc, when this information needs to be shared with external parties such as network services providers, content providers, etc.
As per the definition in clause 3.1 of the present document, UE Energy Information refers to energy-related information collected by the UE. Some examples are shown in Table 7.10.2.2-1. This abstract value of the UE energy-related information is an integer index value corresponding to a particular UE battery capacity, UE supply and UE energy consumption rate, etc. The index is, by design, a relative value which is not comparable with that calculated by another UE with similar energy capacity and energy supply, whether from the same or a different manufacturer.
For example, two UEs from the same manufacturer have 4000 mAh battery capacity. But the way that each of these UEs consume their energy (drain their batteries) is completely different for some or all of the following reasons, amongst others:
- Different battery drain rate due to the age of each of the UEs. The older the UE's battery drains faster than that of the newer UE.
- Different battery drain rate at different ambient temperature (e.g. 40°C versus -5°C) as a result of different electrical efficiency in the UE.
- A premium flagship model might consume energy at a different rate than a basic model, even if both are using the same application under the same conditions.
- Different user behaviours that result in different energy consumption.
7.10.3 Procedures p. 65
This Candidate Solution proposes a new metric; procedures for reporting this metric from the UE to an external entity are described in solution #4 in clause 7.5.
7.10.4 Summary p. 65
This Candidate Solution describes an abstract UE energy information index as described under clause 7.10.2.4. This index is an integer value that does not reveal information about the UE that can be fingerprinted. In addition, the energy index is a relative value which is unique to an individual UE and is not comparable with values calculated for any other UE, even if both UEs are identical (e.g. from the same manufacturer, the same model with the same battery capacity and similar characteristics, etc.) for the reasons mentioned in clause 7.10.2.4.
- This Candidate Solution describes an abstract UE energy information index, which represents the energy consumed by the UE of a particular 3GPP service during a media session via the 5G system.
- This Candidate Solution does not propose any standardised methodology for the calculation of the proposed energy index value. Hence, the proposed methodology is limited by the lack of a standardised way to implement the calculation of the UE energy-related information index value.
- It is intentionally left to the discretion of the UE manufacturer/chipset vendors to determine the most appropriate and most accurate method to calculate the index value depending upon individual UE characteristics.
- The only requirement is that more energy-intensive activities on the UE are reflected in a higher index value (which is a higher integer value).
- The proposed Candidate Solution has several limitations and cannot be implemented. Some examples include: how to calculate the energy consumption information of the individual hardware/software components (such as decoders/encoders) that are involved within the UE during a media streaming session, how to calculate the energy consumption information of the renderers that are used in the media session, etc.
