Daily Dose of 6G

The control channel inherits 5G’s skeleton

Agreed

A 6G control channel element is six resource element groups, and the aggregation levels are 1, 2, 4, 8 and 16 — the same set as 5G. Control resources can sit contiguously or non-contiguously in frequency, in units of six resource blocks. Other granularities are still being studied.

The quieter decision matters more: the 5G and 6G resource block grids align on a shared carrier. Without that, the “efficient sharing of NR and 6GR downlink control resources” the plenary made binding in September would be close to unimplementable. A preliminary MRSS performance assessment went up with it.

Also tabled

Uplink control gets a dedicated channel plus the option of riding on the uplink data channel. The target is two or three formats against 5G’s five — the clearest sign yet of the “fewer options” principle reaching the physical layer.

Reference signals get bigger, and a choice between two ways to shrink them

Agreed

The maximum is 48 orthogonal DMRS ports using a length-2 time-domain orthogonal cover code — twice 5G Rel-18’s 24. Single-symbol DMRS is capped at 24, the exact figure still under study, and the same limits apply to CP-OFDM uplink data. The sequence stays a Gold sequence, and devices can be told about co-scheduled users so they can suppress the interference.

Still open

Maximum layers per device will be 8, 12 or 16. Two overhead-reduction methods are on the table — sparse orthogonal DMRS and superimposed pilots — and the stated goal is to pick at most one. Simulation results at the next two meetings decide that, the layer count and codeword-to-layer mapping together.

Four calls that went to Madrid

RAN1 also priced the 7 GHz plan on existing grids. Reusing 5G mid-band sites at around 7 GHz instead of 3.5 GHz, in urban macro cells serving indoor users, costs about 3 to 13 dB on initial-access signals and channels, worst on Msg3 and Msg5 — the uplink messages that complete a connection.

RAN2 starts deleting duplicate 5G mechanisms

With TR 38.760-2 endorsed as the working baseline, RAN2 applied the leaner-standard principle in four places:

• DRX and wake-up signal: one schedule governs both. No separate monitoring cycles.
• Measurement: aim for a single measurement object and reference-signal configuration, at least for mobility, and study merging reporting events.
• Random access: unify 4-step and 2-step around common building blocks — at least the response format and the monitoring identifier. A 2-step device can fall back to 4-step without an explicit network indication; conditions deferred to the work item.
• RRC: replace the Need M/N/R/S codes with a single release-on-absence principle, and design one re-establishment procedure for all 6G cases.

Agreed

Mobility stays network-controlled: device-based connected-mode mobility without network notification is no longer studied and survives only in the inactive state, and 5G-style subsequent mobility is not a 6G baseline. For early synchronisation, two-step network-controlled early downlink sync is the baseline; contention-based and device-triggered variants were deprioritised.

Still open

Whether one sequence number can replace separate PDCP and RLC counters was pushed to October. Cross-site aggregation over backhaul of up to 10 ms was reported to the plenary with the impacts of both carrier aggregation and dual connectivity, and no choice between them.

The three architectures are really three answers about AI

Roughly two dozen proposals were merged into the three candidates the plenary later referred to. What separates them is how AI enters the network: an optional AI domain beside the connectivity domain; AI served by ordinary 6G network functions; or an architecture organised around AI agents. The summary was noted, not agreed.

Still open

The mobility anchor is a placeholder called xAMF, which could resolve to an enhanced 5G AMF, a new 6G AMF, or NAS routing through a dedicated Signalling Routing Function. Slicing has two directions: keep 5G’s structure, or make the device unaware of slices entirely. Key issues 1, 17, 18 and 19 are the ones most exposed to whichever way the architecture goes.