Phones keep old SRS hopping pattern as chipmakers beat Ericsson and Huawei
MediaTek, Qualcomm and 11 others won a 13-vs-9 split over faster channel sounding, prioritizing compatibility with Release-17 devices.
Source meetings for this article (8)
Meetings in this article’s source material. Inclusion does not mean that every meeting approved every decision.
- SA#11110 Mar 2026 – 13 Mar 2026 · Fukuoka · 3GPP ↗
- RAN#1119 Mar 2026 – 12 Mar 2026 · Fukuoka · 3GPP ↗
- RAN1#1249 Feb 2026 – 13 Feb 2026 · Gothenburg Metropolitan Area · 3GPP ↗
- RAN4#1189 Feb 2026 – 13 Feb 2026 · Gothenburg Metropolitan Area · 3GPP ↗
- SA5#1659 Feb 2026 – 13 Feb 2026 · India · 3GPP ↗
- RAN3#1319 Feb 2026 – 13 Feb 2026 · Gothenburg Metropolitan Area · 3GPP ↗
- SA2#1739 Feb 2026 – 13 Feb 2026 · India · 3GPP ↗
- RAN2#1339 Feb 2026 – 13 Feb 2026 · Gothenburg Metropolitan Area · 3GPP ↗
3GPP QuarterlyThe next issue by email, at most eight a year.
Ericsson, Huawei, Xiaomi and NTT DOCOMO wanted a new frequency-hopping pattern for the Sounding Reference Signal (SRS) that lets a base station measure the uplink channel — Ericsson cited simulations showing 'non-negligible gains in throughput.' MediaTek, Qualcomm, Nokia, Samsung, Apple and eight more companies rejected it, arguing a new pattern would make it harder to multiplex with older Release-17 phones on the same cell. The 13-vs-9 split was not resolved this quarter, so networks keep the legacy pseudorandom pattern for now.
One dispute did close: for SRS repetition symbols hopping within a single frequency chunk, 3GPP locked in exact starting-position patterns (for example {0,0,1,1} and {1,1,0,0} when two positions are used) and ruled that the number of repeated symbols must always be a whole multiple of the number of starting positions — CATT, Sony and Huawei pushed this through over Samsung and InterDigital, who called it redundant. Chipset makers now have a fixed recipe instead of an open-ended design question. Whether phones must keep the signal's phase steady across those hops was left to each manufacturer to decide, not mandated in the spec.
For phones just connecting to the network, companies including Samsung, Huawei, Apple, Qualcomm and MediaTek proposed timing early channel measurements 3 milliseconds after the phone's connection acknowledgment, and letting the network dynamically set the delay for antenna-switching SRS, CSI-RS and tracking signals via a MAC command rather than a fixed broadcast setting. Neither proposal was formally adopted, but RAN1 did confirm concrete limits: early channel sounding will support up to 32 antenna ports and up to 2 tracking-signal bursts, and this early signal will be locked to the same beam the phone used to first access the cell.
A separate MIMO feedback fight: Ofinno proposed a 1-bit flag so phones could report when their current beam is already the strongest, avoiding a negative-number problem in the reporting format. Google, Spreadtrum, ZTE, Xiaomi, Samsung, Nokia, Ericsson, Huawei and NTT DOCOMO — nine companies against one — rejected it, saying phones should simply avoid triggering a report in that case rather than adding a new signaling bit. Meanwhile OPPO, Xiaomi and Samsung lost a separate 3-vs-7 vote over when to reset beam-event counters, with Ericsson, Huawei, Nokia and others winning support for resetting counters whenever the network reconfigures measurement resources.
Watch for the cross-slot SRS timing fight to resume: Samsung, ZTE and Ericsson want a 1-bit flag marking which of two slots an SRS resource falls in, Xiaomi wants to stretch an existing offset parameter, and Sony wants the position implied by resource order — none were down-selected this quarter, and the choice will decide how much signaling overhead networks pay every time a phone reports channel quality across a slot boundary.
The decisions behind this
No consensus on new SRS hopping pattern for combined features
A key issue was whether to allow a new frequency-hopping pattern when two SRS enhancement features are used simultaneously: the Rel-17 'RPFS start RB index hopping' and the new 'intra-repetition hopping'. The legacy pseudorandom pattern helps multiplex users; a new deterministic pattern could sound the entire bandwidth faster.
Two alternatives from a prior meeting (RAN1#123) were debated. Alt A, supported by 13 companies including MediaTek, Nokia, and Qualcomm, proposes no new pattern, arguing it eases multiplexing with older Rel-17 devices. Alt B, supported by companies including Ericsson, Huawei, and Xiaomi, would introduce a new pattern for faster full-band sounding, with sub-options (Alt 1, 2, 3) for the pattern details. Ericsson cited simulation results showing 'non-negligible gains in throughput' for the new pattern. The moderator proposed supporting Alt A, but no consensus was reached and the decision was deferred.
The RAN1#124 meeting record records a partial agreement on this point: "Support Alt B from the agreement in RAN1 #123 meeting as follows:"
| Support Alt A (no new pattern, backward-compatible) | Support Alt B (allow a new pattern for faster sounding) |
|---|---|
| MediaTek · Nokia · Qualcomm · Spreadtrum · CATT · Apple · OPPO · Rakuten · vivo · InterDigital · HONOR · Tejas Networks Ltd · SS13 companies | Ericsson · Huawei · Xiaomi · NTT DOCOMO · Sharp · ZTE · ETRI · NEC · ChinaTelecom · Fujitsu10 companies |
The lack of a decision leaves a key design choice open. If Alt A is chosen, network behavior remains backward-compatible with Rel-17, simplifying deployment. If Alt B is chosen, networks could potentially gather channel state information more quickly at the cost of more complex scheduling to avoid collisions with legacy devices.
Split views on supporting a second cross-slot SRS scenario
The document discusses 'cross-slot SRS,' where an SRS resource set can span two consecutive slots (a downlink or special 'S' slot followed by an uplink 'U' slot). This helps schedule SRS more flexibly in TDD networks. Two scenarios were defined earlier: Scenario 1 (resources span the slot boundary) and Scenario 2 (some resources are entirely in the S slot, others entirely in the U slot).
A proposal to support Scenario 2 was put forward. Proponents like Samsung, Nokia, and Huawei argued it avoids complex configuration, especially for antenna switching. Opponents like OPPO, Xiaomi, and Lenovo argued it is unnecessary as the same functionality can be achieved by configuring two separate SRS resource sets. The meeting did not reach a decision.
| Support Scenario 2 | Do not support Scenario 2 (use two sets instead) |
|---|---|
| Samsung · Nokia · Huawei · ZTE · Ericsson · MediaTek · InterDigital · Rakuten · ETRI · Panasonic · NTT DOCOMO · Tejas · CATT · ChinaTelecom · Sony · Fujitsu · HONOR · KDDI · TCL · SS20 companies | OPPO · Xiaomi · Lenovo · Qualcomm · Spreadtrum5 companies |
Supporting Scenario 2 would give network operators more flexibility to bundle SRS resources across the two slots into a single, triggerable set, which could simplify management. Not supporting it would maintain a simpler specification but require more configuration steps for operators.
- DeferredEarly CSI acquisition: No consensus on mandatory joint triggering in 5G-AdvancedR1-2600647
- DeferredNew CSI subband sizes proposed for low-density 48-port CSI-RSR1-2600647
- DeferredProposal: Dynamic slot offset signaling for early measurements in MSG4R1-2600647
- DeferredProposed timeline for early channel measurements uses 3ms offsetR1-2600647
- DeferredKey debate on SRS frequency hopping patterns remains unresolvedR1-2600322
- DeferredSRS intra-repetition hopping patterns finalized for 5G MIMO Phase 6R1-2600321