6 The new work — what the plenary let start

6.1 Why starting work needs permission

Nobody in 3GPP may simply begin writing binding text. Before the text there is a description of the work — the paper the plenary has in front of it, from the group that wants to do the work, naming the release it is aimed at.

Approving that description is a plenary's decision, and it is the single biggest thing SA#112 did by volume: 80 documents went through agenda item 6 ^[1], more than any other item of the meeting, while item 7 handled 31 documents of finished text ^[2].

The split is worth keeping straight. Item 6 is permission to write. Item 7 is what got written.

6.2 The numbers

The paper arriving:

• 53 new work item descriptions ^[3]
• 10 new study item descriptions ^[4]
• 9 revised study item descriptions ^[5]

The decisions coming out:

• 30 new work and study items approved ^[6]
• 11 existing ones updated ^[7]
• 26 specifications and reports approved ^[8]
• 6 more noted for information ^[9]

Fifty-three new work item descriptions arriving against thirty approvals of work and study items together is not a rejection rate. Many of the fifty-three were revised into other numbers during the meeting — sixty documents of every kind ended with the status "revised" ^[10] — so a single work item can appear twice in the arriving pile and once in the approved one.

6.3 What the plenary let start

Named examples, all approved:

• Post-quantum cryptography migration SP-260663 — a work item on moving to cryptography built to survive quantum computers.

• Withdrawing the 256-bit algorithm specifications SP-260664 — a work item whose whole purpose is to take Rel-18 and Rel-19 text back out.

• Extra capabilities for emergency communications SP-260654 — a study item, on the same subject as the one incoming letter that got a reply.

Four more sit in this course's register, each carrying the status word the meeting's own document list gives it, and all four read "approved":

• SP-260690 — a study on 3GPP capability exposure, from SA WG6.

• SP-260685 — application enablement for satellite access, phase 4, from SA WG6.

• SP-260675 — service based management architecture, phase 4, from SA WG5.

• SP-260656 — a study on charging for ambient power-enabled Internet of Things, from SA WG5.

Read those seven together and the meeting's centre of gravity is visible without any interpretation: security migration, satellites, management, and devices with no battery.

6.4 Finished text

26 specifications and reports were approved, with 6 more noted for information.

The named one here is SP-260692, the report of the study on integrating satellite components into the 5G architecture, phase 4. It was approved.

"Noted for information" is the weaker outcome of the two. It means the plenary has the document in front of it and is not yet signing it off — the same distinction as The kinds of document, and the ten words that decide them draws between approved and noted.

One annex of the meeting's report is empty and worth a sentence. The report keeps a table for approved work item exception sheets, and at this meeting that table holds 0 of them ^[11] — it reads "No entries found." An empty annex is a real finding — it says the plenary granted none, not that nobody kept the record.

6.5 Work items and study items are not the same thing

The two arrive as different document types — 53 new work item descriptions against 10 new study item descriptions and 9 revised ones — and they end in different places.

A study produces a technical report, and that report comes back to a plenary to be approved in its own right. SP-260692 is one: the report of the satellite integration study, phase 4, approved at this meeting. So is SP-260604, though that one summarises a whole release rather than one study.

A work item produces changes to binding text, and those arrive as change requests in packs — The change requests — 689 of them, in packs — written after the permission is given, which means at a later meeting than the one that gave it.

That gap is why a plenary's list of approvals is a forecast rather than a report. 30 new work and study items were approved here, and the text they authorise will show up in the change-request counts of the meetings after this one.

6.6 The long argument: a codec and its training data

The work item that took the most arguing was an AI-based speech codec for voice over satellite, at very low bit rates.

It arrived as SP-260417 and was revised. The version that survived is SP-260696, and the report's account of the outcome is one sentence: "This new WID was approved." ^[12]

What made it hard was not the codec. It was the data needed to train the model, and the meeting handled that by not deciding it: "TSG SA Chair to inform PCG on this issue of 3GPP involvement and legal requirements related to the dataset needed to train the AI model specific to 3GPP work." ^[13]

6.7 The other long argument: how 3GPP writes

Agenda item 9 is a top-level item devoted to one study, on modernising how 3GPP writes and handles its specifications. It drew eleven documents ^[14].

The study's own paperwork was noted rather than approved — its status report SP-260406 and its draft technical report SP-260407.

Two positions were put. Qualcomm argued for markdown and git SP-260544. Ericsson and Nokia demonstrated 3GPP Forge, a git-based home for specifications and change requests SP-260600. Both papers were noted.

The chair closed it with a guidance document SP-260679, and the report records: "This was endorsed." ^[15] The guidance itself says two things:

• "- Continue with trial of separated ASN.1 in FS_6GSpecs" ^[16]

• "Interested companies are encouraged to demonstrate the use of Forge for 3GPP specifications (e.g. 38.331)." ^[17]

A trial continues, and companies are encouraged to demonstrate. Nothing was adopted. That is a real outcome and it is the one to report — the alternative, "3GPP moved to git", is not in this record.

6.8 What did not go through

Three things were held or pulled, and each shows a different way a plenary declines.

• Held for comparison. Every working group's terms of reference was postponed SP-260470, with the reason stated: "It was decided to ask SA WG4, SA WG5 and SA WG6 to also submit ToRs to the next TSG SA meeting and SA WG1, SA WG2 and SA WG3 may update or re-submit the same ToR proposals." ^[18]

• Held with no reason recorded. The change to the change-request process for 3GPP Forge was postponed SP-260411, and the report's whole account of it is "This was postponed." ^[19]

• Pulled by its authors. The paper from ten public bodies asking that 6G be designed to be secure from the start was noted SP-260584, and its revision was withdrawn during the meeting SP-260643.

There is also an agenda item that drew nothing at all: item 8, the study on AI/ML consistency alignment ^[20]. An item with its own place on the agenda and no contributions is a subject that is live enough to have a slot and quiet enough to have no paper.

Set the two postponements side by side and you can see the difference between a plenary declining well and declining badly. The terms of reference were held with a stated plan, a named set of groups and a next meeting to come back to. The Forge change-request process was held with three words and no recorded reason at all — and anybody who has to know why will have to wait for the chair's notes or ask somebody who was in the room.

6.9 Where this leads

Permission to write is one half of a plenary. The other half is the writing itself, arriving as corrections in bulk: The change requests — 689 of them, in packs.