Currently energy consumption and efficiency can be monitored and considered through O&M and network operation, but not as a service performance criterion, as for example bit rate, latency or availability. Guidance from SA to all working groups states: There is an important type of traffic where energy efficiency policy, for example a maximum amount of energy to be utilized could be applied without conflict with this guidance. Best effort traffic is a type of traffic that is provided as a service to customers everything else being equal. Of course, security, privacy and complexity principles will not be sacrificed, but there is no conflict between a service policy that constrains performance (e.g. latency, throughput, even availability) on the basis of energy consumption and a best effort service, since there are no guarantees in the case of best effort traffic. We can say that best effort traffic is not associated with QoS policy service performance level criteria. Today the 5G system works to support services efficiently, though does not take into account energy consumption at the service level. The use case explores a particular opportunity to identify this information and use it to make more efficient use of all network resources without sacrificing service quality. In particular, information gathered through O&M, and in the future possibly from the network (see 5.1.5 which identifies a gap and opportunity), can be leveraged to make it possible to employ energy consumption information as part of service delivery. In the following use case, the possibility of using energy consumption as a new service criterion for this less constrained type of mobile telecommunication service is explored. A large-scale logistics company L has deployed a large number of communicating components. These are integrated into vehicles, palettes, facilities, etc. Essentially, IoT terminals enable remote tracking and monitoring functions. The information gathered is relevant, but not constrained with respect to latency. In fact, eventual delivery (e.g. after hours or even a full day) of communication is entirely acceptable for L. The MNO M offers a 'green service' which limits the rate of energy consumed for communication over a particular time interval (e.g. per day) and this service is appropriate for L, whose overall corporate goals are also served by 'green service', as they strive to operate with energy efficiency.

L deploys many UEs with associated 'green service' subscriptions from M. These subscriptions policies include the following criteria:

• Best Effort Service (service that is not associated with QoS policy service performance level criteria)
• Energy Constraints applied to service delivery

1. The fleet of trucks belonging to L leave the logistic center located in the middle of the uninhabited region hundreds of kilometers northeast of the major city Erehwon. There are many devices located in this fleet. The trucks and their contents comprise a physically dense group of UEs, all communicating periodically with the network. This 'massive IoT' group leaves the coverage of the logistics center. The network coverage over the road through the uninhabited region is very sparse.
2. As the trucks proceed into extreme low coverage, the energy consumed to communicate with the IoT devices increases. This energy consumption increase is monitored by the 5G network and can be aggregated, e.g. at the slice level.
3. The 'green service' policy for the service provided to L includes a maximum energy consumption rate. At a certain point the IoT communication of the fleet exceeds this maximum energy consumption rate.
4. The policy indicates that latency can be traded off with energy consumption for service to L; the communication service is delay tolerant in this condition. As the energy consumption rate has exceeded the maximum, the latency is increased to enforce this policy. In effect, L's fleet receives very limited service, with high latency, even for a limited period of time, no service at all.

The use case description does not define how operator M offers the 'green service'. One possibility is that the maximum energy consumption policy applies to all services for the subscription of a device deployed by L with operator M. This simple policy may not be appropriate if the UEs deployed by L use different kinds of services at different times. In this case, the policy would apply to specific services (service flows, etc.) A requirement at the service flow level is not pursued in this use case. A further option is that specific network slices apply a 'green service' policy to all services communicating by means of that slice. The use case does not describe how energy consumption is determined. There is related work in SA5 and RAN3 to determine energy consumption. If energy consumption cannot be determined at the granularity, e.g. of a specific service or network slice or even the aggregate energy consumption of a subscriber, it is still possible to identify the total energy consumption of different elements in the 5G network. It is therefore possible, at least in principle, to divide the total energy by the number of served sessions, subscribers, etc. 'Average consumption' of a node or cell or network slice, etc. is a course unit of measurement, and does not reflect the true energy consumption at the finer granularity, though it still can be a useful metric. Though an averaging approach could be useful to count the total amount of energy used to attribute to each subscriber, this approach is not enough to measure the rate of energy consumption as described in this use case. For this, there would have to be finer granularity energy reporting than 'per node' or 'per cell.' Though this is not yet supported in the 5G network.

The IoT devices in the fleet belonging to L are able to communicate with varying latency, depending on the energy consumption required to serve the devices. When the UEs are in poor coverage, they communicate seldom, when under good coverage, they can communicate more frequently. The total energy consumption of M's network has reduced while still providing adequate service to customer L. It is important to emphasize that there has been no trade-off between 'energy efficiency' and 'service quality.' Customer L received what was necessary while using less energy precisely because the energy consumption was taken into account in the service delivery.

The 5G network can monitor energy consumption. The existing energy consumption monitoring is done at an O&M level, per network node, per cell and per network slice. The number of UEs per network node, cell and network slice are also known. Please see Annex A for an overview of existing energy efficiency standardization, which includes the determining energy consumption for use in calculating energy efficiency. The 5G network can enforce performance criteria, as described in clause 6.7 of TS 22.261. Most of the enforcement requirements refer to prioritization, but policies that result in other enforcement are possible too, e.g. gating, charging, credit control, restrictions with respect to maximum allowed resources, etc. Gap: there is currently no means for the 5G network to determine the per subscriber or per network slice service flow energy consumption. This information is not included in network data analytic services.

[PR.5.1.6-1] [PR.5.1.6-2] [PR.5.1.6-3]

1. subscriber granularity (considering all services of the 5G network for the subscriber);
2. network slice granularity.

[PR.5.1.6-4] Subject to operator's policy, the 5G network shall support energy consumption monitoring at per network slice and per subscriber granularity.