MRSS dominates the 3GPP RAN #113 plenary

Let's just agree to disagree, shall we?

3GPP TSG RAN #113 · Madrid · 14–17 September 2026

Welcome to the first of three special reports covering the September 3GPP plenaries, held earlier this month in Madrid.

Back in June I ended the RAN #112 analysis by saying the migration question was “loaded and primed for September”, after the three candidate options were first proposed in Fukuoka in March. Plenary chair for Madrid meeting, Samsung’s Dr Younsun Kim, opened by ruling that the discussion “shall focus on down selection”. Well, they all certainly tried, but not everyone was happy and a final decision was postponed until TSG WAN #114 in December.

When you read the SA #113 analysis (coming tomorrow) then you will know the consequence of SA inheriting a decision it had not planned for. You have to remember the size of the challenge, with so many stakeholders, positions, views and constraints. I keep saying this: it’s amazing that we continue to get regular Releases at all. The Chair tried four times in one session to find suitable words that the plenary delegates (all 381 of them) can live with. This is consensus standardisation in action. The unfortunate issue is that the months are ticking by and the 6G deadline looms ever larger.

There are already a number of great analysis pieces out there that cover the RAN proceedings. What follows is my own summary of the RAN #113 proceedings, based on the Chair’s draft report and the submissions referenced. I also draw on the RAN plenaries I have covered over the past 18 months and the ongoing coverage in the 6G Futures weekly newsletter. Oversights, omissions and errors are mine, and I will gladly correct them. And as usual, I try not to taint the summaries with my own views. Well, I try...

The MRSS debate

We should start with a recap of the options. As defined at RAN #111, every option is an addition to multi-RAT spectrum sharing (MRSS), which is the agreed baseline for running 5G NR and 6G Radio in the same carrier.

• Option 1: 6G RAN node as anchor, aggregated with NR cells (6G-anchored dual connectivity)

• Option 2: 5G RAN node as anchor, aggregated with 6G cells (the NSA-like arrangement)

• Option 3: No aggregation in the RAN; 5G RAN and 6G RAN each connected to their own core

The headline of one key discussion document, signed by 12 operators and two vendors, was “only Option 1 is supported as an additional migration option.” They offered a practical argument, as a 6G standalone launch will lean on one or two low or mid bands for coverage; therefore if MRSS proves impractical across more than a couple of FR1 bands, or does not work well in FR2, dual connectivity to existing 5G carriers is how an operator matches 5G throughput on day one. Option 2 was dismissed as a repeat of 5G non-standalone, which as we know didn’t work out too well, and option 3 was dismissed as reactive aggregation with no radio awareness. But then Deutsche Telekom withdrew its signature saying that none of the options are needed. That put DT in the same place as T-Mobile USA, who does not want 6G to become “further advanced 5G”.

The Chair ran a check of operator preferences and tallied the results as follows (noting that a few operators voted for multiple options!):

• MRSS only: 8 operators

• Option 1: 18 operators

• Option 2: 3 operators

• Option 3: 6 operators

After an offline session the Chair tried again, which is where it got incredibly complicated. I’ve tried to document the full process in the longer archival version of this, but here’s the end result. The Chair declined to formally agree those proposals but asked the working group chairs to take them on board, while reminding companies that they dump a great deal of work on those chairs and then expect magic. His own forecast was rather bleak: “in three months we will have exactly the same situation, company views will not change much.”

NTT Docomo warned that with every delay 3GPP loses the industry’s trust; Telecom Italia thought September or December made little difference; and Vodafone suggested that 5G Non-Standalone “was not a bad solution”, the problem was the delay in the options (!). Vodafone also made its support for Option 1 conditional on a lightly evolved 5G core and said that if the 6G core looks different the question would need revisiting, which is essentially the same dependency NGMN flagged in its letter to all three TSGs (its promised paper update did not arrive), and it is why the migration decision and SA WG2’s November architecture deadline are really one decision.

So what was actually agreed?

With every attempt at new wording meeting objections, the Chair fell back on the moderator’s original text, the same proposal that had drawn “a larger number” of concerns earlier in the session. It was duly endorsed and has three parts split as follows:

On MRSS: “Efficient sharing of at least NR and 6GR DL control channel resources shall be supported in 6GR”, striving for the downlink control overhead RAN1 tabled for the shared-resource cases. This is the first hard plenary requirement on MRSS efficiency. There is some urgency to this (at least amongst the supporting companies): if NR and 6G Radio each configure their own control resource set in a 10 MHz FDD carrier, the additional overhead is over 14 per cent (increasing to 23 per cent in the worst case). The operators asked for “dynamic” sharing and actual got “efficient” sharing intot he agreed text, leaving some wriggle room for RAN1 discussions. The operators’ second and third proposals (shared reference signals, multiplexed uplink control and sounding) were not taken up.

On migration: the decision on an additional option is postponed to RAN #114, which will also “discuss and decide on split bearer” for Options 1 and 2.

On Option 3: a “simplified variant” may be considered. How simple you ask? Well, no RAN impact is expected or specified; no core network aggregation and no traffic split; two single registrations, to the 5G core over NR and the 6G core over 6G Radio; network-controlled mobility only; and one target use case, concurrent voice over NR on the 5G leg with 6G data on the 6G leg. The final revision added the line “SA will make final decision on this option”, and the approved reply to SA WG2 asks for a feasibility answer by RAN #114 and a TSG SA decision no later than SA #114 (i.e. by the next plenary).

The original Option 3 that Boost, AST SpaceMobile, Skylo and EchoStar actually proposed was a dual-stack route whose selling point was 4G-to-6G and NTN migration with service continuity, whereas what went across to SA is a voice-continuity scenario. The satellite operators got their option kept alive, however, it is now in a form that doesn’t really satisfy their original intentions.

There was slightly better news on the 6G-6G aggregation thread, which had been parked since the March plenaries. RAN2 is tasked to study aggregation for non-collocated sites over non-ideal backhaul of up to 10 ms latency, assuming dual independent schedulers, “with the goal of enabling a multi-vendor interoperable solution”. Verizon and Docomo called multi-vendor a requirement; Deutsche Telekom sees no need for dual connectivity in 6G at all; ZTE doubts 6G requirements can be met at 10 ms; and Huawei warned that a multi-vendor condition may rule features out. It is, as the Chair observed, the migration argument all over again.

6G reference devices

Instead of multiple device categories in the LTE sense, 6G will have “reference device types”, each defined by a minimum mandatory set of features. Every 6G UE is associated with exactly one and may report additional capabilities on top. Each type carries at least a minimum number of transmit and receive chains and a specified minimum and maximum UE bandwidth for each direction.

The draft proposal offered two types, with Massive IoT and eMBB as example use cases. The fixed wireless access (FWA) device type proved somewhat problematic. Verizon pointed out that an FWA device type had been agreed in March and had now “suddenly become a service”. When the Chair asked who wanted a third reference type for FWA, 27 companies responded positively. Also, AT&T wants XR added, vivo wants RedCap covered, and the automotive industry (Bosch, BMW, Volkswagen, Toyota, Hyundai, BYD and others) and the railways both submitted papers asking not to be forgotten.

The conclusions were endorsed with a note that decisions on device types are taken “solely” at plenary level, with RAN4 allowed to continue only on eMBB minimum requirements. I think we heard the exact same argument from operators a year ago.

No consensus yet on modulation

Modulation was due a decision this quarter and ended up with a partial answer. After what the Chair called a repeat of the RAN1 discussion, the conclusion reads: 1024QAM is supported in the uplink for FWA usage, and “there is no consensus on the support of 4096QAM in 6G Rel-21”. Qualcomm was one of the stakeholders who didn’t like the original wording of “not supported”, and this way the door is left open for the cable-aligned companies to claim that says 4K QAM is already in use in the industry. MediaTek says that 4K (i.e. 4096QAM) does not outperform 1K (1024QAM) and Huawei simply doesn’t support it.

Constellation shaping, which June’s plenary said RAN1 must “decide or stop”, got nowhere really, apparently due to misaligned simulation assumptions. Not quite “the dog ate my homework”... Better news elsewhere, as RAN3 settled June’s one-word argument that 6G will support the higher layer split, CU-DU and CP-UP, “with 5G HLS as the baseline”. Five European operators had it endorsed that a 6G band’s over-the-air device requirements start from those of the equivalent 5G band. SA’s new “H” reference point naming was noted (and no, I am still none the wiser why it is H) and SA WG3 reported interim agreements on radio-layer security.

Attempts to reopen the completed requirements report for UE-side data collection, a “unified” AI/ML framework and relaxed extended-coverage text were all discarded. Vodafone withdrew a proposed study on antenna control interfaces and the Chair twice asked why technical proposals on sensing and NTN had been sent to the plenary at all, which “is for project management”.

6G is brought to you by 3GPP

There was one very clear decision taken in Madrid. 3GPP will submit a single radio interface technology to ITU-R for IMT-2030, based on 6G Radio alone. There will be no set of technologies bundling NR, as was done with LTE for IMT-2020. Its name will be “6G”, with the footnote “Developed by 3GPP as 6G, Release 21 and beyond”.

The schedule was argued (obviously) before target dates were settled as follows, in a liaison approved for SA and the Project Coordination Group:

• February 2027: High-level presentation at the ITU-R Working Party 5D workshop

• June 2027: Initial submission template

• June 2028: Partial self-evaluation

• February 2029: Final self-evaluation

• Approval December 2026; target December 2028: RAN-led self-evaluation study item

Against the Rel-21 timeline agreed in June, the final self-evaluation target is one month before the March 2029 ASN.1 freeze. No room for error.

---

What to look for at RAN #114

RAN #114 meets in Boston on 7 to 11 December 2026, the same week as SA #114.

• Migration decision, third attempt: whether any additional option beyond MRSS is supported, and if so which one.

• Split bearer: to be “discussed and decided” for Options 1 and 2.

• Simplified Option 3: SA WG2 feasibility feedback due by RAN #114; TSG SA’s final decision due that week.

• 6G core architecture: SA WG2’s deadline is the same November meeting.

• Reference device types: parameters and minimum feature sets for the three types, and whether a fourth (XR, vehicular, a RedCap successor) is added.

• MRSS: RAN1’s response to the control channel requirement and whether the operators return with their uplink and reference-signal proposals.

• ITU-R: approval of the self-evaluation study item.

• 6G-6G aggregation: first RAN2 findings on the 10 ms non-collocated case.

• 4096QAM and constellation shaping: whether either comes back before the Rel-21 package is assembled.

The Rel-21 package will be approved in March 2027 and RAN1’s normative work starts the following quarter. December is therefore the last plenary at which a migration option can be chosen (and allowing for a full quarter to then scope out the work).

Will stakeholder positions change by December? If nothing is agreed, the outcome is MRSS only, which is exactly what Deutsche Telekom, T-Mobile USA, Telstra and Jio asked for. The “none” camp does not need to win the argument, it needs only to keep objecting. Time for some negotiations between the camps? Consensus, as the Chair said more than once, is what 3GPP is based on.

Next up: the SA #113 report.