128-antenna phones get a smaller timing extension than some vendors wanted

Companies disagree over how much extra processing time a phone should get when it has to crunch Channel State Information from 128 antenna ports. vivo, Google, Xiaomi, Apple, Qualcomm and MediaTek argued the existing timing rules, built for smaller antenna counts, would leave phones short on time; Samsung, ZTE, Huawei/HiSilicon and Ericsson wanted to keep the legacy deadline unchanged to avoid adding complexity. Nobody got exactly what they asked for.

3GPP settled on a compromise for phones using the lower-power 'Capability 2' processing timeline: the CSI reference deadline (nCSI_ref) gets scaled by a factor of 2, not the factor of 4 (ceil(128/32)) that OPPO and NEC's opponents had floated. The same meeting also picked the simpler one-part CSI report format for wideband feedback on PUCCH formats 3 and 4, over a two-part alternative Qualcomm, Xiaomi, Apple and eight others backed for its potential overhead savings — Google, CMCC, Ericsson, Samsung and 11 other companies preferred keeping the legacy one-part structure.

For fast Layer 1/2 handovers, 3GPP confirmed phones can use semi-persistent CSI-RS — reference signals the network can switch on and off — for network-scheduled candidate-cell measurements, activated via a MAC CE. But extending that same flexibility to event-triggered reports, where the phone itself decides when to report, split the room 9 to 6: Spreadtrum, Ericsson, CMCC, ZTE, Samsung, Lenovo, vivo and MediaTek said letting the network control the reference signal undercuts the point of phone-driven triggering; Fujitsu, Nokia, Qualcomm, Huawei, NTT DOCOMO and CATT disagreed. Aperiodic CSI-RS for the same feature failed by a similar 10-to-7 margin, with opponents citing signaling overhead between network nodes.

MediaTek used field measurements to challenge whether 6-layer MIMO is worth standardizing for 6-antenna phones at all. Its SSB-based SNR data showed only about 5% of readings exceed 25 dB, while rival simulations claim 6-layer MIMO only beats 4-layer above 30 dB SNR — a condition MediaTek says rarely occurs in real networks. Its own link-level simulations, using a 3% transmitter error vector magnitude and antenna efficiency losses of at least 3 dB, found no gain for 6L6R over 4L6R, and it proposed dropping the requirement work entirely; no company has yet spoken up to defend continuing it.

RAN1#120bis will pick up the timing-reference proposal for candidate-cell CSI-RS measurement in LTM, skipped this quarter for lack of time, plus the still-open framework for early CSI acquisition after handover — including whether reports travel via UCI or a MAC CE. Also unresolved: which formula (InterDigital's Option 2a versus the simpler Option 3) will scale angles in the FR3 channel model used to simulate 7-24 GHz MIMO performance for 5G-Advanced and 6G.