Report of the SA4-e (AH) MBS SWG post 134 (Dec 18 and Jan 29)
Two ad hoc sessions were held:
- Session 1: December 18, 2025, 15:30-17:38 CET (28 participants)
- Session 2: January 29, 2026, 15:30-17:35 CET (29 participants)
The meetings focused primarily on the FS_Energy_Ph2_MED study (Media Energy Consumption Exposure and Evaluation Framework Phase 2), with additional discussions on Release 19 corrections, Advanced Media Delivery Phase 2, and a new QUIC-based streaming study.
Contribution: BBC (Richard Bradbury)
Technical Content:
- Corrects discrepancies between SA4 stage 2 specification (TS 26.502) and CT3 stage 3 specification found during reference implementation
- Adds two missing events to align with CT3:
- "User data ingest session starting" (indicates start of active period)
- "User data delivery has started" (confirms broadcast MBS session start)
- Simplifies table by factoring out repetitive information
- Corrects sequence diagram to ensure correct identifier is passed back
- Editorial fixes for clarity
Impact: Corrective changes only; aligns SA4 spec with CT3 with no impact on CT3 work. Similar corrections planned for Releases 18 and 19.
Decision: Agreed
Contribution: BBC (Richard Bradbury)
Technical Content:
Key Gaps Identified:
- Current EIF (Energy Information Function) from Release 19 only covers user plane network functions (gNB, UPF)
- Does not cover energy consumed by Application Servers (AS) or User Equipment (UE)
- AS are not network functions and do not provide energy information to EIF
Proposed Abstract Data Model:
- Hierarchical model for 5G media streaming allowing energy apportionment at multiple levels:
- Application server instance
- Content hosting configuration
- Distribution configuration
- Streaming session
- Service data flow
- Generic energy report structure with baseline parameters:
- Timestamps
- Energy consumed (joules)
- CO2 emissions per joule
- Data volumes (uplink/downlink)
- Session IDs
- Endpoints
- Extensible for specific use cases
Recommendations:
1. Update TR to clarify EIF scope (Proposal 1 & 2: Agreed)
2. Require candidate solutions to include procedures and parameters for collecting/aggregating energy data from AS, NFs, and UEs (Proposals 3, 4, 5: Endorsed in principle)
- Not gating for existing solutions
- Template/examples to be developed with Julien
- CRs to be prepared for next meeting
Discussion Highlights:
- Franck questioned use of data volume vs. direct energy reporting; Richard explained flexibility for different architectural solutions
- Prakash questioned value of AS energy exposure; Richard affirmed importance given gap in Release 19
- Rufael emphasized factual accuracy when referencing TS 23.501 and avoiding overlap with other groups
- Thomas noted TR should not self-reference; context to be clarified in CR
Decision: Agreed (with follow-up CRs expected)
Contribution: Orange, BBC (Julien Lemotheux, Richard Bradbury)
Technical Content:
- Candidate solution for 5GMS focusing on selection of media streaming service location driven by content steering server based on energy characteristics
- Added new procedures for solution #5 in clause 7.6
- Missing steps for streaming session management now included and renumbered
- Improved text and procedures based on feedback
Minor Corrections Identified:
- Remove "downlink" from Step 5 in diagram 7.6.3.2 to make procedure generic
- Typo in 7.12.2 to be addressed
Decision: Revised to S4aI250208; Endorsed (with minor corrections to be incorporated before merging into Mega CR)
Contribution: Orange (Julien Lemotheux)
Technical Content:
- Candidate solution for Key Issue 6: client-driven management of media delivery service energy optimization
- Title updated to "Client Driven Selection of Media Entry Point in the Generalized Media Delivery System Based on Energy Characteristics"
- Replaced terms like "SIM variant" and "delivery pass" with "media entry point"
- Allows client to reconfigure media handler during session to select different entry point based on energy characteristics
- Playback resumes from previous position (similar to Netflix/Amazon session resumption)
Discussion Highlights:
- Richard noted switching entry points mid-session is disruptive (black screen/spinner) and needs better description
- Shilin questioned step numbering discontinuity and why media entry points passed to media-aware application rather than directly to media stream handler
- Julien explained application needs entry point information to make selection
Decision: Revised to S4aI250210; Noted (further clarification needed)
Contribution: Samsung (Prakash Kolan)
Technical Content:
- Addresses three SA1 use cases related to energy provisioning:
1. Dynamic service adjustment based on network energy information
2. Service adjustments based on energy configuration
3. Tolerance to QoS degradation
- Proposes adding table summarizing different way forward options for these use cases into TR
- Provides reference points for candidate solutions to map to specific options
Discussion Highlights:
- Richard noted contradiction in requirement 5.3 "tolerance to QoS degradation" - content suggests optimizing without degrading QoS
- Julien suggested simplifying/merging options (e.g., option 2 with option 1; options 3 and 4)
- Prakash agreed to work on simplification
Decision: Endorsed as basis for future work (with simplification to be addressed in revision)
Contribution: Qualcomm (Thomas Stockhammer)
Technical Content:
- Update on interactions with Green Streaming (GSS) organization
- GSS focus: streaming app energy measurement and CDN (complementary to 3GPP telco aspects)
- GSS ongoing projects:
- Remote energy measurement platform
- Yang data model for energy
- White papers
- Proposal: Joint online webinar (90-120 minutes) on energy awareness in mobile streaming
- Involve 3GPP, GSS, and other organizations
- Schedule before next 3GPP meeting
- Goal: information exchange and awareness raising
Discussion Highlights:
- Julien noted timing challenge as study aims to finish in Goa
- Rufael welcomed industry input for documentation in TR
- Richard interested in Yang data model
- Gaëlle noted first meeting useful for mutual understanding even if not immediately impacting TR
Decision: Noted; Thomas to discuss with 5G-MAG and sync with rapporteur
Contribution: InterDigital (Franck Aumont)
Status: Not presented due to time constraints
Decision: Noted without presentation; to be resubmitted
Contribution: Nokia (Daniel Venmani)
Technical Content:
- Candidate solution for relocating media application servers based on energy consumption at different service locations
- Procedure:
1. Launch media session
2. Collect QoE measurements from UE
3. Energy Information AF requests energy consumption reports from media AS for each service location
4. AF ranks service locations by energy consumption
5. AF informs UE to enable selection of most energy-efficient location
- Main technical focus: enabling AS to report energy consumption per service location (current gap in SA4)
Discussion Highlights:
- Richard questioned necessity of steps 9, 18, and 23 in call flow; suggested removal
- Richard noted expanding EIF to include AS information is outside SA4 remit (SA2 responsibility)
- Rufael emphasized AS energy reporting should be handled by EIAF within SA4, not by expanding EIF scope
- Prakash and Julien noted similarity to Julien's solution; suggested merging and highlighting differences
- Main difference: use of QMC and source of information
Decision: Revised to S4aI250209; Noted (not endorsed due to outstanding comments; further work needed)
Contribution: BBC, Orange (Richard Bradbury)
Technical Content:
- Expanded summary of EIF based on latest Release 19 developments from SA2
- Incorporated corrections from Rufael (Huawei):
- Terminology aligned with TS 23.501 ("energy consumption information" not "energy related information")
- EIF obtains energy data from OM and computes values per PDU session using SMS information
- Describes what SA2 studied and specified in TS 23.501:
- EIF role in collecting, calculating, and exposing energy consumption data at various granularities
- Added stage 3 specification details (TS 29.566, TS 29.122):
- Structure of energy notifications and reports
- Timestamps and optional energy info data (joules consumed during reporting period)
Discussion Highlights:
- Rufael emphasized importance of consistency with SA2 outcomes
- Franck raised concern about information granularity difference:
- 5GC NF can request per QoS flow data
- Media AF cannot get per QoS flow data (potential loss of granularity at data flow level)
- Richard clarified text comes from baseline TR 26.942 Release 19; if discrepancy exists, separate CR needed
Decision: Agreed (to be added to merged common CR)
Contribution: BBC, Orange (Richard Bradbury, Julien Lemotheux - Rapporteur)
Technical Content:
- Guidelines for documenting candidate solutions as annex to TR
- Standardizes structure inspired by previous studies (e.g., TR 26.804 for AMD)
- Proposed skeleton includes:
- Mapping to key issues
- Functional description
- Collaboration scenarios (referencing annexes in TS 26.501/26.506)
- Architecture mapping
- Energy-related information (baseline parameters)
- Procedures (including sequence diagrams)
- Gap analysis (deltas from current specs)
- Proposed normative changes
- Summary
Clarifications:
- Template applies to new candidate solutions added during phase 2, not existing solutions from phase 1
- Goal: help extract normative work and analysis for conclusion section
- Not intended to modify existing solutions already in TR
Discussion Highlights:
- Thomas questioned purpose if not applied to existing solutions; concerned about methodology for evaluation
- Julien explained template facilitates aggregation and analysis for new contributions
- Richard noted both key issue-centric and solution-centric approaches valid; selection for normative work addressed later
Decision: Agreed
Contribution: BBC (Richard Bradbury)
Technical Content:
- Conceptual model to clarify what constitutes media application service
- Hierarchical structure:
- Media delivery session consists of service data flows
- Service data flows consist of application data flows
- Examples: segmented media delivery, RTC sessions
- Accounts for deployment dynamics:
- Peer-to-peer sessions
- Multiple service location endpoints
- Endpoints on different physical/virtual servers (including edge)
- Network considerations:
- Traffic routing depends on UE route selection policy
- Multiple PDU sessions, network slices, access networks
- UE mobility and edge computing
- Energy implications:
- Energy consumption changes dynamically due to these factors
- EIF does not provide complete picture (lacks AS and UE energy visibility)
- Mapping to 5GMS and RTC models detailed
Discussion Highlights:
- Thomas concerned about introducing many new capitalized terms and abstract definitions
- Potential complexity and divergence from existing 5GMS architecture
- Questioned if new concepts require updates to TS 26.501/26.506
- Nervous about implications for future normative work
- Richard clarified:
- Conceptual framework for informational purposes only
- Not new data model or intended for normative work
- Most terms already in existing specifications
- Generic model, not specific to 5GMS or RTC
- Iraj questioned if model covers entire 5GMS operation or just energy consumption
- Richard: conceptual overview for user plane operations, arranging existing concepts
- Prakash supported inclusion; most terminology already referenced in SA2/SA4 specs
- Julien: model important for structuring report and clarifying granularity levels
Resolution:
- Thomas requested clear scoping to TR only, not basis for normative changes in existing specs
- Introductory text explaining rationale and scope to be added
- Richard agreed to add clarifying text
Decision: Endorsed (with clarifications to be added in revision)
Contribution: Orange, BBC (Julien Lemotheux, Richard Bradbury)
Technical Content:
- Updates to candidate solution 5 defining Energy Information Collector and Energy Information Application Function
- Goal: establish baseline procedures for energy information reporting reusable by other candidate solutions
- Key changes:
- Clarified subscription and reporting process with EIF
- Information requests now require subscription detailing needed data (service data notification, application ID)
- Standardized process: after subscription, EIF exposes energy report to requesting component
- Same principle applies to reports from Energy Information AF and Energy Information Collector
- Initial reporting step introduced:
- Configure and receive first report before media session starts
- Ongoing reporting during media streaming session
- Updates as more detailed information (e.g., IP 5-tuple) becomes available
- Aligned with new template for candidate solutions:
- Includes gap analysis section
- Includes proposed normative changes section
- Special numbering to avoid disrupting references in other TR parts
Discussion Highlights:
- Richard: Steps 18 and 19 should be inside loop
- Richard: Step 7 should happen as result of media session, not provision (steps 7-9 need to move up)
- Daniel questioned motivation for EIAF to choose client and for client to move between AS
- Julien acknowledged missing information
Decision: Revised to S4aI260027, then S4aI260028; Noted (further revision needed)
Multiple additional candidate solutions were submitted but noted without presentation due to time constraints:
All to be resubmitted for SA4#135 in Goa.
Contribution: Samsung, Nokia (Prakash Kolan)
Decision: Noted without presentation; to be resubmitted
Contribution: Qualcomm (Thomas Stockhammer)
Decision: Noted without presentation; to be resubmitted
Contribution: Dolby (Jason Cloud)
Technical Content:
- Discussion paper on using CMMF (Common Media Format) for multi-access media delivery within 5G system
- Application-layer based approach
- Implementation:
- Multi-access capability in media player's access client
- Multiple HTTP clients initialized, each bound to different network interface
- Enables concurrent downloads from various access networks
- Example: client connected to both Wi-Fi and 5G can toggle interfaces or handle network degradation
- Integration with 5G system:
- Mirrors multi-service location use case
- Minimal architecture changes required
- Mainly notes in call flows and potential client setup details in stage 3
Discussion Highlights:
- Richard questioned feasibility on 3GPP UE (application awareness of ATSSS, interface abstraction)
- Prakash clarified:
- For ATSSS transparent mode, Richard's concern valid
- Non-transparent mode: application aware of delivery
- Application multipath supported in spec
- Jason confirmed Android clients expose interfaces to application, allowing binding
- Thomas expressed confusion:
- Application would not see two HTTP clients or access networks as described
- Functionality handled under the hood, outside 5G media streaming scope
- Changes should be addressed in IETF protocols
- Does not align with Release 20 multi-access discussions
- Very different from ATSSS
- Richard suggested issue might be implementation detail rather than specification concern
Decision: Noted for now
Contribution: Xiaomi (Emmanouil Potetsianakis)
Technical Content:
- Work plan for FS_QStream_MED study
- Focus: segmented media delivery for on-demand and live video services
- Objectives:
1. Identify relevant applications, protocols, and technologies
2. Develop evaluation framework
3. Conduct evaluations
4. Integrate findings
5. Coordinate with other groups
- Output documents:
- TR 26.934: Test platform for media delivery technologies (joint with QUIC study of RTC)
- TR 26.835: Evaluation of QUIC-based streaming protocols
- Timeline:
- Goa meeting: Finalize work plan and specification skeletons
- Early phase: Application selection, metrics definition
- Subsequent meetings: Test framework design, technology definition
- Later stages: Simulation setup, platform development, communication with external groups, result analysis
Discussion Highlights:
- Thomas requested descriptions for each objective for easier reading
- Thomas questioned why simulation setup/platform development not starting earlier (Goa)
- Core issue should be coordinated with other study items
- Should begin sooner than April
- Emmanouil clarified:
- Test platform TR and theoretical study of framework start in Goa
- Actual implementations begin after Goa, aiming for tangible results by April
- Coordination with RTC group expected for relevant objectives
- Gaëlle asked for clarification on phases of objective 4
- Emmanouil explained:
- Phase 1: Framework setup for segmented media
- Phase 2: Define relevant network scenarios
- Phase 3: Run experiments and evaluations
- Split due to joint framework with RTC for different applications
Decision: Agreed as baseline (comments to be incorporated for SA4#135 submission)
Special powers granted to MBS SWG to send reply LS to S4-251660 were not exercised.
Valerie (InterDigital) raised concerns about timeline for energy feasibility study:
- Importance of structuring candidate solutions
- Ensuring all proponents have fair chance for adoption in TR
- Suggested further discussion with Julien offline
Julien suggested organizing offline meeting approximately one week before next scheduled meeting to process and discuss contributions, given large number of documents.
Session 1 (Dec 18, 2025):
- Duration: 2h 21m 45s
- Participants: 28
- Average attendance: 1h 49m 9s
Session 2 (Jan 29, 2026):
- Duration: 2h 53m 54s
- Participants: 29
- Average attendance: 1h 40m 14s