5G/NR - Multi-TRP Transmission Procedures   

 

 

 

Multi-TRP Transmission Procedures

A strong data beam is not enough if UE loses the control channel that schedules it. Multi-TRP gives the network more than one transmission and reception point to serve the same UE. The procedure depends on who sends the scheduling command and how the data is distributed between those points.

This is a focused child note of Enhanced Massive MIMO. The main scope is the Rel-16 baseline and Rel-17 evolution: PDSCH scheduling, TCI indication, reliability schemes, control-channel diversity and uplink repetition. Later-release fields in the current ASN.1 are identified separately and do not expand that scope.

Executive Summary

Keep the number of scheduling commands separate from the number of transmission occasions. Two TRPs do not necessarily mean two DCIs, two transport blocks or two cells.

Question

Rel-16 baseline / Rel-17 development

Who schedules the PDSCH?

Single-DCI: one command covers the multi-TRP transmission. Multi-DCI: each PDSCH has its own scheduling command.

How does UE identify the receive assumptions?

TCI states identify reference signals and QCL properties. A DCI TCI codepoint can map to two states.

What separates the control paths?

The coresetPoolIndex-r16 configuration distinguishes two CORESET pools for multi-DCI operation; it is not a physical site identifier.

What improves reliability?

Spatial, frequency or time diversity can protect data. Rel-17 also improves control-channel and uplink diversity.

What does Rel-17 add?

PDCCH repetition / SFN operation, PUSCH and PUCCH repetition toward two TRPs, and inter-cell multi-DCI PDSCH operation.

What is outside this note?

Rel-18 multi-TRP unified TCI extensions and later simultaneous multi-panel evolution are not part of the Rel-16/17 baseline.

What does a TRP represent ?

A TRP is a radio transmission and reception point, not automatically a separate gNB or a separate serving cell. The UE needs a usable description of the radio path. The network also needs enough coordination to schedule and deliver the intended data.

Throughput and reliability are different objectives

Two radio paths can carry different spatial layers to increase data rate, or provide diversity for delivery of the same transport block. Those objectives consume resources differently, so a throughput comparison must include the extra frequency and time resources.

For example, repeating a transport block through two directions can help when one direction is blocked. It does not guarantee twice the throughput. If both directions suffer the same interference or obstruction, their failures can remain strongly correlated. UE receive capability, reference-signal quality and scheduler decisions determine whether the second path is useful.

Multi-TRP, CoMP and dual connectivity

Multi-TRP describes the radio procedure used to serve UE from multiple points. CoMP is a broader coordination concept, while dual connectivity describes a different connectivity and protocol arrangement. A drawing with two towers is insufficient to distinguish them.

The CoMP note explains the coordination background. Non-coherent multi-TRP transmission does not require the TRPs to form one phase-coherent distributed precoder. It still requires coordination of resources, timing and UE assumptions. Coherent joint transmission adds tighter channel and phase requirements and should not be inferred from the presence of two TCI states.

  • Identify the scheduled channels, transport blocks and control paths before assigning a multi-TRP procedure to a deployment diagram.

Rel-16: Single-DCI and multi-DCI scheduling

The first distinction is the scheduling command. In single-DCI operation one command describes the scheduled multi-TRP reception. In multi-DCI operation UE decodes separate scheduling commands, with each command carrying the information for its corresponding PDSCH.

Single-DCI: one command with two TCI states

RRC configures candidate TCI states and the applicable PDSCH settings. MAC activation establishes the usable TCI mapping. The TCI field in the scheduling DCI can then select a codepoint associated with two TCI states instead of one.

UE must interpret that selection together with the antenna-port field, DM-RS CDM groups, time-domain assignment and configured repetition scheme. Two TCI states alone do not tell UE whether to split frequency resources, use separate layers or receive repeated occasions. The QCL note explains the underlying receive assumptions, and PDSCH TCI MAC CE explains activation background.

The schematic below separates the control decision from the data paths. It is a logical arrangement, not a requirement for particular physical sites or a particular gNB implementation. Each data arrow represents the portion or occasion assigned to that path.

Single-DCI and multi-DCI control pathsSingle-DCIMulti-DCITRP ATRP BUETRP A / pool 0TRP B / pool 1UEOne DCI+ data path AData path BDCI A+ PDSCH ADCI B+ PDSCH BOne scheduling decision covers both pathsEach PDSCH has its own scheduling command

Figure 1. Logical control and data paths for the two scheduling modes.

Multi-DCI: CORESET pools and HARQ feedback

For multi-DCI operation, CORESETs can be associated with pool index 0 or 1. UE uses that association in the reception procedures. A missing coresetPoolIndex is treated as pool 0 under the applicable Rel-16 rule.

The pool index does not identify a physical tower globally. It distinguishes configured control-resource groups within the procedure. Each PDSCH has its own scheduling information, including its HARQ process information. Do not infer two independent MAC entities or a separate HARQ namespace merely from two pool values.

HARQ-ACK can be handled using joint or separate feedback according to configuration and the applicable codebook rules. Therefore, two TRPs do not automatically require two independent PUCCH feedback transmissions. Check ackNackFeedbackMode-r16, the HARQ-ACK codebook and feedback timing together. The HARQ note supplies the retransmission background.

  • TCI mapping, DM-RS interpretation and scheduling configuration jointly define reception; the number of TRPs alone does not.

Rel-16: How is the PDSCH distributed ?

The same pair of TRPs can use different resource arrangements. Spatial separation, frequency separation and time separation impose different receiver and latency requirements. The distinction between one transmission occasion and repeated occasions of the same transport block is especially important.

Spatial, frequency and time arrangements

Read the two-state TCI indication together with the configured scheme. The table separates the actual reception behavior from the informal scheme numbers often used in presentations, because those numbers are easier to confuse than the parameter names.

Arrangement

What UE receives

What to avoid assuming

Spatial / two DM-RS CDM groups

A single PDSCH can associate different DM-RS port groups with the two TCI states.

This is not automatically two independently decodable copies of a transport block.

fdmSchemeA

One PDSCH transmission occasion of the TB; two non-overlapping frequency allocations use the two TCI states.

Each frequency part is not necessarily a full copy of the TB.

fdmSchemeB

Two PDSCH transmission occasions of the same TB, on non-overlapping frequency resources.

The two allocations do not double delivered application data.

tdmSchemeA

Two non-overlapping PDSCH occasions of the same TB within one slot.

The second occasion still consumes symbols and needs the configured timing.

Slot-based repetition

Occasions of the same TB across the configured consecutive slots; TCI mapping follows the configured pattern.

Repeated occasions need not all use the same redundancy version or TCI state.

Figure 2 illustrates the resource separation. The diagram deliberately omits exact symbol counts, PRB counts and RV values; these come from the configured allocation and scheduling information, not from the names A and B.

Frequency and time separation of two TRP pathsTCI state ATCI state BFDM: same time, different PRBsTDM: separate symbol intervalsRepetition across slotsABABABFrequency splitWithin one slot: time progresses rightSlot nSlot n+1FDM A and FDM B share this geometry, but differ in transmission-occasion / TB mapping.

Figure 2. Conceptual resource arrangements; these are not exact resource-grid allocations.

Repetition, combining and control reliability

A repeated transport block can use different redundancy versions, so repetition does not necessarily mean identical transmitted bits. UE processes the associated observations according to the configured PDSCH and HARQ rules, including the mapping of TCI states across occasions.

The scheduling PDCCH can still limit reliability in a single-DCI arrangement. Providing two data paths does not protect an unreceived command. Multi-DCI offers separate control paths, but the second command schedules its own PDSCH; it does not automatically substitute for a lost command from the other pool. Rel-17 addresses control diversity more directly through repeated or SFN PDCCH transmission.

  • Separate resource geometry, transport-block mapping and control-channel protection when comparing multi-TRP reliability schemes.

Rel-17: What changes beyond the PDSCH baseline ?

Rel-17 extends diversity beyond the downlink data path. The additions protect control reception, support uplink transmission toward two TRPs, and allow an additional PCI in the inter-cell case. Each feature still depends on UE capability and network configuration.

PDCCH diversity and uplink repetition

PDCCH repetition provides two transmissions carrying the same DCI through linked search spaces associated with different CORESETs. This differs from multi-DCI scheduling, where the commands independently schedule their corresponding data transmissions. SFN-based PDCCH provides another control-diversity arrangement.

For SFN PDCCH, the UE uses one search space/CORESET with different TCI states for the two transmitting paths. For uplink PUSCH repetition, the network can indicate transmission of the same contents toward two TRPs using DCI or semi-static configured-grant configuration. The associated spatial relations determine the intended transmit directions. PUCCH repetition can similarly use different spatial relations toward the two TRPs.

Repetition toward two TRPs should not be read as a blanket requirement for simultaneous transmission from two UE panels. Follow the actual occasion mapping, UE capability and power constraints. See Uplink MIMO and Configured Scheduling for the underlying uplink mechanisms.

Inter-cell operation and the unified-TCI boundary

Rel-17 inter-cell multi-TRP permits a TCI state to refer to an SSB with a PCI different from the serving-cell PCI for multi-DCI PDSCH operation. The additional PCI is part of the configured reference-signal interpretation, not an automatic serving-cell change.

The activated states can be associated with at most one additional PCI at a time in this Rel-17 procedure. In ASN.1, additionalPCI-r17 is an index into an additional-PCI list, not the PCI value itself. UE therefore needs the list and the TCI state together to resolve the reference.

Rel-17 also introduced unified TCI for single-TRP beam indication. Its presence in the same release does not mean that the later multi-TRP unified-TCI framework was already available. The extension of unified TCI to multi-TRP belongs to Rel-18 and is outside this focused note. The Beam Management note provides the broader beam context.

  • Rel-17 improves control and uplink diversity and adds inter-cell operation; do not backdate the Rel-18 multi-TRP unified-TCI extension.

RRC, MAC CE and DCI: How to read a trace

The over-the-air procedure is distributed across three control layers. RRC supplies possible states and resource arrangements. MAC activation establishes usable mappings, and DCI selects the scheduled behavior. A useful trace analysis must reconstruct all three before interpreting the PDSCH.

Configuration-to-transmission checklist

Start before the scheduling slot. A DCI codepoint only has meaning with the mapping active at that time. Checking an unrelated earlier RRC configuration or ignoring a later MAC update can produce the wrong TRP association.

Layer / observation

What to verify

UE capability

Supported multi-TRP modes, simultaneous reception limitations, repetition capabilities and release-specific features.

RRC CORESET configuration

coresetPoolIndex-r16, search spaces and the relevant control-channel TCI configuration.

RRC PDSCH configuration

Candidate TCI states, repetition scheme, time-domain allocation and DM-RS settings.

MAC activation

For Rel-16 PDSCH multi-TRP, inspect the Enhanced TCI States Activation/Deactivation for UE-specific PDSCH MAC CE and its codepoint mapping.

Scheduling DCI

TCI codepoint, antenna ports, resource assignments, MCS, HARQ process, NDI and RV as applicable.

Feedback

Joint/separate feedback configuration, HARQ-ACK codebook, PUCCH resource and feedback timing.

Rel-17 extensions

Linked PDCCH repetition configuration, UL spatial relations and additional PCI mapping when those features are used.

A useful illustrative case is one DCI selecting a two-state codepoint with fdmSchemeB. UE expects two frequency-separated occasions of the same TB, rather than two independent application payloads. If the trace instead shows one DCI in each CORESET pool, decode each PDSCH assignment separately before deciding how the feedback is constructed.

Practical constraints and misleading symptoms

Poor performance can come from a stale reference-signal assumption, insufficient receive capability, correlated interference or an unprotected control path. These problems can look similar in throughput measurements even though their remedies are different.

Check per-path channel estimates and decoding outcomes, then compare control detection and HARQ feedback. Also compare the occupied resources and latency budget against the single-TRP case. More repetitions may improve successful delivery while reducing useful bits per allocated resource. Network coordination delays can limit scheduling flexibility, but single-DCI and multi-DCI are air-interface modes, not specifications of an exact backhaul topology.

  • Reconstruct configuration, activation and scheduling in time order; then compare reliability using equal resource and latency assumptions.

RRC ASN.1

These selected definitions connect the procedure to actual RRC syntax. They were checked against the latest published TS 38.331 v19.4.0, while the explanations remain focused on Rel-16 and Rel-17. Referenced types are not all expanded, so this is not a complete RRCReconfiguration message.

Rel-16 repetition configuration

The repetition configuration selects frequency/time arrangements or slot-based behavior. The ASN.1 choice alone does not provide the allocated PRBs, symbols or transport-block size. Those values still depend on the associated PDSCH configuration and scheduling command.

Following is based on 38.331 v19.4.0 (Release 19): RepetitionSchemeConfig-r16

RepetitionSchemeConfig-r16 ::= CHOICE {
    fdm-TDM-r16                        SetupRelease { FDM-TDM-r16 },
    slotBased-r16                      SetupRelease { SlotBased-r16 }
}

Following is based on 38.331 v19.4.0 (Release 19): RepetitionSchemeConfig-v1630

RepetitionSchemeConfig-v1630 ::=   SEQUENCE {
    slotBased-v1630                    SetupRelease { SlotBased-v1630 }
}

Following is based on 38.331 v19.4.0 (Release 19): FDM-TDM-r16

FDM-TDM-r16 ::=                SEQUENCE {
    repetitionScheme-r16           ENUMERATED {fdmSchemeA, fdmSchemeB,tdmSchemeA },
    startingSymbolOffsetK-r16      INTEGER (0..7)                                    OPTIONAL  -- Need R
}

Following is based on 38.331 v19.4.0 (Release 19): SlotBased-r16

SlotBased-r16 ::=              SEQUENCE {
    tciMapping-r16                 ENUMERATED {cyclicMapping, sequentialMapping},
    sequenceOffsetForRV-r16        INTEGER (1..3)
}

Following is based on 38.331 v19.4.0 (Release 19): SlotBased-v1630

SlotBased-v1630 ::=            SEQUENCE {
    tciMapping-r16                 ENUMERATED {cyclicMapping, sequentialMapping},
    sequenceOffsetForRV-r16        INTEGER (0)
}

The slot-based mapping can be cyclic or sequential, and the RV offset controls a separate part of the repetition behavior. The v1630 definition adds the zero-offset alternative. Keep that versioned extension when checking a decoded configuration instead of treating the original integer range as the only possible setting.

TCI state and release boundaries

A TCI state combines an identifier with one or two QCL descriptions. Its Rel-17 extension can identify an additional PCI and related joint-TCI information. Later extensions remain visible below so the listing matches the current specification without implying earlier-release support.

Following is based on 38.331 v19.4.0 (Release 19): TCI-State

TCI-State ::=                       SEQUENCE {
    tci-StateId                         TCI-StateId,
    qcl-Type1                           QCL-Info,
    qcl-Type2                           QCL-Info                                                    OPTIONAL,   -- Need R
    ...,
    [[
    additionalPCI-r17                   AdditionalPCIIndex-r17                                      OPTIONAL,   -- Need R
    pathlossReferenceRS-Id-r17          PathlossReferenceRS-Id-r17                                  OPTIONAL,   -- Cond JointTCI1
    ul-powerControl-r17                 Uplink-powerControlId-r17                                   OPTIONAL    -- Cond JointTCI
    ]],
    [[
    tag-Id-ptr-r18                      ENUMERATED {n0,n1}                                          OPTIONAL    -- Cond 2TA
    ]],
    [[
    pathlossOffset-r19                  ENUMERATED {
                                            dB-12, dB-8, dB-4, dB0, dB4, dB8,
                                            dB12, dB16, dB20, dB24, dB28, dB32, dB36,
                                            dB40, dB44, dB48, dB52, dB56, dB60}                     OPTIONAL    -- Cond JointTCI
    ]]
}

Following is based on 38.331 v19.4.0 (Release 19): QCL-Info

QCL-Info ::=                        SEQUENCE {
    cell                                ServCellIndex                                               OPTIONAL,   -- Need R
    bwp-Id                              BWP-Id                                                      OPTIONAL, -- Cond CSI-RS-Indicated
    referenceSignal                     CHOICE {
        csi-rs                              NZP-CSI-RS-ResourceId,
        ssb                                 SSB-Index
    },
    qcl-Type                            ENUMERATED {typeA, typeB, typeC, typeD},
    ...
}

For this note, the principal Rel-17 inter-cell field is additionalPCI-r17. The pathloss-reference and uplink-power-control fields relate to the Rel-17 unified-TCI framework and must be interpreted with their presence conditions. The tag-Id-ptr-r18 and pathlossOffset-r19 fields are later additions, not Rel-16/17 requirements. QCL Type D describes spatial receive parameters; a TCI state is not a literal antenna coordinate or a globally unique TRP identifier.

  • Use the release suffix and field conditions to separate the current ASN.1 definition from the capabilities of a Rel-16 or Rel-17 UE.

Reference

The relevant procedure clauses and information-element definitions were consulted, rather than every clause of these specifications. Rel-16 and Rel-17 documents establish the scope; the current RRC text supplies the verified ASN.1 listings.