3GPP TR 26.942 CR0022r2: Application Server Energy Information

Change Request Overview

Type: Category B (Addition of Feature)
Release: Rel-20
Work Item: FS_Energy_Ph2_MED
Source: BBC, Orange

This CR introduces a new Candidate Solution for reporting energy consumption of Application Servers in the Data Network as part of end-to-end service chain energy monitoring.

Key Issue Mapping

This Candidate Solution addresses:

• Key Issue #2: Energy-related monitoring and measurement, specifically for Application Servers instantiated in the Data Network
• Key Issue #1: Energy-related information exposure to Media Session Handler for energy-efficient optimization

Main Technical Contributions

Functional Description

The solution recognizes that total energy consumption for application services must account for:
- 5G System energy (5G Core and RAN) via Energy Information Function
- Application Server energy in the Data Network (focus of this solution)
- UE energy consumption

Key principle: Application Servers exist in the Data Network and are not Network Functions, therefore outside the scope of the Energy Information Function and must be accounted for separately.

Energy Reporting Granularities

The solution identifies critical granularities for energy optimization:

1. Per Service Location:
2. Enables steering application sessions to different service locations based on energy costs

3. Supports aggregated view across multiple (Edge) Application Server instances hosting the same service location

4. Per Service Data Flow:

5. Application Server has visibility of Application Data Unit bytes transferred (uplink/downlink)
6. Service Data Flows identified by 3-tuples or 5-tuples
7. Can also be identified by DNS host name (e.g., TLS client "hello" message)

Important Limitation: Application Servers cannot report energy consumption per UE (no visibility of UE identities), but the Energy Information AF can perform reverse IP address lookup in 5G Core for per-UE routing.

AS Energy Report Structure

The solution proposes a comprehensive AS Energy Report with the following baseline parameters:

Core Parameters

• AS instance identifier: Unique identifier across all Data Networks
• Timestamp: Start of energy consumption sampling period
• Sample period: Duration of the sample
• Energy consumed: Total energy during sampling period
• Environmental cost of energy supply: (Optional) CO₂ equivalent emission per unit energy, with breakdown by source if multiple sources

Application Data Flow Parameters

For each active Application Data Flow:
- Service Data Flow description: IP 5-tuple identifying client and Application Server endpoints
- Uplink data volume: Total Application Data Units delivered to AS instance
- Downlink data volume: Total Application Data Units delivered from AS instance
- AS host name: (Optional) Application Server host name for UE connection
- Application session identifier: (Optional) Unique identifier for application session
- Configuration identifier: Unique identifier for configuration in logical Application Server
- Service location identifier: (Optional) Identifier for service location exposed by AS instance

Energy Attribution Model

The solution proposes a proportional attribution model:
- Total energy consumption measured over sampling period
- Bytes transferred per Service Data Flow metered during same period
- Average energy attributed to Service Data Flows proportionally to data volume

Supported Aggregations

The raw information enables multiple aggregation views:

1. Per Service Data Flow: Combine with network energy information from EIF
2. Per UE: Via reverse IP lookup by authorized recipients (e.g., Energy Information AF)
3. Per Data Network/Network Slice: Via client IP address analysis
4. Per Media Streaming Session: Using media delivery session identifier
5. Per Virtual Host: Using configuration identifier and canonical domain name
6. Per Service Location: Using service location identifier
7. Per Content Hosting Configuration/Provisioning Session: Aggregate across all Edge AS instances

Architecture Integration

The solution integrates with the Energy Information AF architecture (from Clause 7.6) via reference point E3, enabling:
- Periodic AS Energy Reports from Application Server to Energy Information AF
- Generic procedure realization (step 13 in generic procedure, step 19 in 5GMS procedure)

System-Specific Mappings

5GMS System

• Content Hosting Configuration identifier maps to Provisioning Session
• Distribution Session identifier maps to canonical host name
• Media delivery session identifier for longitudinal analysis
• Correlation with QoE metrics reports (TS 26.512)

RTC System

• RTC AS Media Function configuration identifier maps to Provisioning Session
• Support for in-band energy-related information extraction from media

Gap Analysis

• AS Energy Report is entirely new for Release 19
• No service-based interface currently defined at reference point E3
• Energy Information AF is also new, requiring new interface specifications

Proposed Normative Work

Stage 2 Requirements

1. Generic AS Energy Report specification:
2. New stage 2 Technical Specification
3. Service operations for subscription and notification via E3

4. Based on generic architecture from Clause 7.6.2.2

5. 5GMS instantiation (TS 26.501):

6. Map generic properties to 5GMS concepts

7. Extensions for in-band energy information from M2d (downlink) and M4u (uplink)

8. RTC instantiation (TS 26.506):

9. Map generic properties to RTC concepts
10. Extensions for in-band energy information from RTC AS Media Function

Stage 3 Requirements

1. Generic data structures:
2. New stage 3 Technical Specification

3. Service-based interface definitions for E3

4. 5GMS extensions (TS 26.512):

5. 5GMS AS Energy Report data structures

6. RTC extensions (TS 26.113):

7. RTC AS Energy Report data structures

Additional References

• TS 26.512: 5G Media Streaming (5GMS); Protocols
• TS 26.113: Real-Time Media Communication; Protocols and APIs

Design Principles

• Raw information provision: No preconceived usage patterns, enabling flexible downstream processing
• Privacy considerations: Care required to avoid exposing identifiable data (client IP, GPSI, SUPI) to unauthorized consumers
• Equal attribution assumption: Energy per byte during sampling period for simplified calculation
• Correlation support: Common filters (Service Data Flow, application session identifier) enable correlation across EIF, AS, and UE data sets