3 How the work is cut up, and who carries it

3.1 The first problem is deciding what is in

There is no folder in 3GPP called "satellites". There is a work plan of 9495 rows, and the satellite work is scattered through it under short names people invented one at a time, release after release.

So before anything can be counted, somebody has to write down a rule for what belongs. Every number in this course rests on that rule, which makes it the first thing worth understanding — and the first thing worth distrusting.

3.2 The rule, and the three families taken back out

A work plan row is in this subject when its short name carries NTN as a whole word, or the older spelling nonterr, or when its own title or the title of a row above it says satellite or non-terrestrial.

That rule picks 183 rows out of the 9495 in the whole work plan ^[1], and those rows carry 155 different short names ^[2].

Then three families are taken back out, and each removal is a judgement worth being able to argue with.

• Satellite navigation. 25 rows under 22 short names, AGNSS, AGNSS-GP, ANSS and LCS3-AGNSS among them. Working out where a phone is by listening to navigation satellites is a different subject from connecting through a satellite, however similar the words look.

• Protecting the Earth Exploration Satellite Service. Four rows under NR_mmWave_protect and its parts. That is coexistence with somebody else's satellite — staying out of its way — rather than access through one.

• The SIM Application Toolkit. The short names SAT and USAT1 to USAT5 are the same three letters about a different subject entirely. This removal took nothing with it: no row the rule had picked up carried one of those names in the first place.

The one deliberate exception is worth knowing, because it looks like an error and is not. A row whose short name carries NTN as a word stays in even when it is about navigation, which is why the Release 20 work on staying up when the navigation fix goes away is counted here NR_NTN_GNSS_resilient.

3.3 One short name is not one work item

The 155 short names sit on 183 rows because radio work in 3GPP is split into parts, and each part is its own row.

• The feature row carries the plain short name — NR_NTN_solutions NR_NTN_solutions.

• The core part ends -Core. It is the main body of the rule-writing.

• The performance part ends -Perf. It is what has to be measured and met.

• The conformance part ends -UEConTest. It is the tests that prove a device does what the other three say, and it belongs to RAN5.

The four rows of one feature can run on very different timescales.

The Release 17 NR work's core part was planned to finish in 2022, its performance part in 2023, the feature row itself in 2024 and its conformance part in 2025 NR_NTN_solutions-Core NR_NTN_solutions-Perf NR_NTN_solutions NR_NTN_solutions_plus_CT-UEConTest.

3.4 Homework first, rules after

Two kinds of permission exist and they are not interchangeable.

A study item, opened by a study item description, is permission to do homework. What comes out is a study report, a TR, and nobody has to obey it.

A work item, opened by a work item description, is permission to write or change binding text. What comes out is a specification, a TS, or a change to one that already exists.

The shape shows in the paperwork. 4374 of these documents are pCRs — proposed text for a study report, which is how a report gets written a paragraph at a time ^[6]. 10374 are change requests, the only way a published specification is ever altered ^[7].

And 12603, more than either, are discussion papers: an argument with no text attached ^[8].

The rest of the pile is 2192 draft change requests ^[9], 1712 filed as "other" ^[10], 1407 outgoing liaison statements and 898 incoming ones ^{[11] [12]}, and 478 change request packs ^[13].

A liaison statement is one group writing to another. That the record holds 1407 going out and 898 coming in ^{[11] [12]} is the clearest measure there is of how much of this subject happens between groups rather than inside one.

3.5 Every row points at one paper

The work plan gives each row the number of the document that describes it, and the register carries that paper for every row that names one. It is the shortest honest statement of what a group promised, which makes it the best thing to read when you want to know what a piece of work is actually for.

Two examples sit at the very bottom of the record, and they are the oldest papers in it. RP-101284, from LightSquared, describes the L band item, and the meeting it belongs to is RAN#50, which sat in Istanbul in December 2010 RP-101284 RP-50.

RP-110775, from DBSD and Terrestar Networks, describes the 2 GHz band item and belongs to RAN#52, in Bratislava in mid-2011 RP-110775 RP-52.

Both files were uploaded to the 3GPP server on the same day in 2017, years after those meetings sat. The upload day is when the file went up, never when the document was written, and the meeting code is what dates the older work.

3.5.1 One work item, several papers

A row points at one document, but the four rows of one work item do not have to point at the same one. The Release 17 NR feature row names RP-220208, approved at RAN#95-e, while its core part names RP-221169, approved at RAN#96 RP-220208 RP-221169.

Both are marked "WID revised", which is the second thing to understand about these papers: a work item keeps its identity while its description is replaced underneath it.

The work plan always points at the newest revision, so the document it names is what a group is bound to now — not what it originally promised.

3.6 Nine releases, and the strays at the bottom

The 183 rows are spread over nine releases, and the spread itself tells the story of when this subject actually happened.

Release	Rows of this subject
Rel-8	1 Rel-8
Rel-10	8 Rel-10
Rel-12	2 Rel-12
Rel-15	1 Rel-15
Rel-16	2 Rel-16
Rel-17	23 Rel-17
Rel-18	47 Rel-18
Rel-19	58 Rel-19
Rel-20	41 Rel-20

Every row above the Rel-15 one is a stray, and each is worth naming so that nobody later mistakes it for the start of the subject.

The Rel-8 row is a stage 1 row of ANSS, positioning work whose parent row the navigation exclusion removes and which the rule keeps only because the row's own title does not repeat the word. It carries no document at all ANSS.

The eight Rel-10 rows are two old band items for the ancillary terrestrial component of mobile satellite service in North America, split into their parts. Neither carries a document L_Band_LTE_ATC_MSS S_Band_LTE_ATC_MSS.

The two Rel-12 rows are IMSSat and its stage 1 row — reaching the telephony core over a satellite television return channel — and neither carries a document either IMSSat IMSSat-SA1.

The single Rel-15 row is where the line this course follows actually begins: the study on New Radio to support non-terrestrial networks, which produced TR 38.811 FS_NR_nonterr_nw TR 38.811. Its revised study document was approved at RAN#76, in West Palm Beach, in June 2017 RP-171450 RP-76.

From there the shape is plain. Two rows in Rel-16, 23 in Rel-17, then 47, 58 and 41 in the three releases after it. NR over satellite, release by release and The same idea on the older radio, for things that are not phones follow the two radio lines through that run.

3.7 Eighteen group codes carry the documents

35306 meeting documents name one of the 155 short names ^[15], and they were filed at 550 separate meetings ^[16]. Eighteen group codes appear: five RAN working groups, six SA working groups, four CT working groups, and the three plenaries where work is opened.

Group	Documents	Meetings	First	Last
RAN4	7490 RAN4	34	R4-97-e	R4-120
RAN2	6714 RAN2	44	R2-103	R2-134
RAN1	6492 RAN1	47	R1-91	R1-126
SA2	3588 SA2	50	S2-128	S2-175
RAN5	2455 RAN5	20	R5-96-e	R5-112
CT1	2164 CT1	37	C1-126-e	C1-162
RAN3	1725 RAN3	40	R3-101	R3-133
RAN plenary	1273 RAN	36	RP-76	RP-112
SA5	751 SA5	36	S5-125	S5-167
SA1	711 SA1	29	S1-80	S1-114
SA3	551 SA3	23	S3-105-e	S3-128
SA6	402 SA6	16	S6-59	S6-73
SA4	272 SA4	14	S4-132	S4-136
CT3	202 CT3	28	C3-113-e	C3-146
CT4	180 CT4	30	C4-101-e	C4-136
CT plenary	144 CT	23	CP-90-e	CP-112
SA plenary	119 SA	27	SP-80	SP-112
CT6	73 CT6	16	C6-106-e	C6-127

3.7.1 What the top of that table means

RAN4 carries more than any other group: 7490 documents over 34 of its meetings RAN4. It owns the radio requirements — TS 38.108 for the node in orbit, TS 38.101-5 for the device, TS 38.133 for radio resource management TS 38.108 TS 38.101-5 TS 38.133 — and the band items of this subject name RAN WG4 as their leading group.

RAN2 with 6714 RAN2 and RAN1 with 6492 RAN1 come next, and they carry the protocol and physical-layer work respectively. The leadership lines name RAN2 as the leading group of the Release 17, 18 and 19 NR items, and RAN1 as the leading group of the early studies NR_NTN_solutions NR_NTN_enh NR_NTN_Ph3 FS_NR_nonterr_nw.

SA2 is the largest group outside RAN, with 3588 documents — and it spreads them over 50 meetings, more separate meetings than any group here SA2. A group that appears at fifty meetings has been carrying a thread for a long time rather than doing one burst of work.

3.7.2 The tail, and what each group in it does

RAN5 does the conformance tests: 2455 documents RAN5, and every short name ending -UEConTest names it. CT1 writes the protocol the device speaks: 2164 CT1. RAN3 owns the network interfaces and carries 1725 RAN3 TS 38.413 TS 38.423.

Then the system trades, each with one job: SA5 with 751 for charging and management SA5, SA1 with 711 for the requirements SA1, SA3 with 551 for security SA3, SA6 with 402 for the application layer SA6, and SA4 with 272 for speech coding SA4.

The three smallest are CT working groups: CT3 with 202 CT3, CT4 with 180 CT4 and CT6 with 73 CT6.

3.7.3 How to read a meeting code

Every code in that table is a group's own short code and a meeting number.

The register carries the code beside the group's name — R4 for RAN4, S2 for SA2, CP for the CT plenary RAN4 SA2 CT — and it carries the meetings themselves under the same spelling, so R4-116 is the meeting whose own title reads RAN4#116 R4-116.

Several codes in the table carry a trailing -e, among them R4-97-e, C1-126-e and S3-105-e. That is part of the meeting's name as the record spells it, and nothing here says what the letter stands for.

The first and last meeting on each row are the ones the register holds for that group's documents of this subject.

They are not the first and last meeting that group ever held, and reading them that way is mistake M2 of this school's own writing rules: a floor in a store written up as the start of the work.

3.7.4 The plenaries are where work is opened, not done

The RAN plenary carries 1273 documents RAN, the CT plenary 144 CT and the SA plenary 119 SA. Those are small numbers for a reason: a plenary approves the paper that starts a piece of work and the one that closes it, and everything in between happens in a working group.

3.8 The busiest single meeting

RAN4#116 had 555 of these documents on its table, more than any other meeting in this record ^[17]. It sat in Bengaluru from 2025-08-25 to 2025-08-29, and 633 people were registered for it R4-116.

That is not an outlier so much as the top of a range.

RAN4#115 in Malta handled 450 of them, with 531 registered R4-115, and the RAN4 meetings after it run in the same few hundreds: 405 at RAN4#117 in Dallas R4-117, 391 at RAN4#118 in Gothenburg R4-118 and 319 at RAN4#119 R4-119.

The register holds an entry only for the meetings it needed — the busiest, and the ones that opened a work item. For the other 482 of the 550 there is no entry, so no days, no place and no attendance.

3.9 Who writes the documents

2664 different names appear as the writer of one of these documents ^[18]. Eight of them appear more often than the rest.

Filer	Documents
Huawei, HiSilicon	2287 [19]
Ericsson	2184 [20]
CATT	2076 [21]
Qualcomm Incorporated	1337 [22]
Samsung	1285 [23]
Nokia, Nokia Shanghai Bell	1043 [24]
OPPO	848 [25]
MediaTek Inc.	846 [26]

Two cautions ride with that table, and both come from the method line of the entries themselves.

These are the names as each document spelled them. Nothing was resolved to a parent company, so one firm writing under two spellings appears twice and its two counts were never added together.

And the count is of documents filed, which is a measure of effort, not of influence. This record holds no vote of any kind, so nothing here says whose papers were agreed and whose were noted and forgotten.

3.10 Where this leads

The machinery above is the frame for everything after it. The releases in the table are followed radio line by radio line in NR over satellite, release by release and The same idea on the older radio, for things that are not phones, and the reasons the groups gave for opening each row are Why anyone wanted it, and who said so.

To follow it yourself, the work plan is the one file that holds the rows, the short names, the planned days and the number of the paper that opened each one. Everything in this chapter was read out of it, or out of the document list beside it.