Summary on Beam Recovery Mechanism 3GPP TSG RAN WG1 Meeting NR AH#3 R1-1716767 Nagoya, Japan, 18th – 21th September 2017 Agenda item: 6.2.2.4 Summary on Beam Recovery Mechanism MediaTek

Introduction Contribution review for WG1 NR Ah-Hoc #3 on beam recovery is summarized on Appendix Based on the summary, issues that are more widely discussed and more convergent are listed the following slides for potential consensus

RS for beam failure detection Agreements: Support L1-RSRP reporting of measurements on SS block for beam management procedures The following configurations for L1-RSRP reporting for beam management are supported SS block only (with mandatory support by UE) CSI-RS only (with mandatory support by UE) SS block + CSI-RS independent L1 RSRP reporting Joint L1-RSRP using QCL-ed SS-block + CSI-RS is optionally supported by UE (with optionally support by UE) Proposal 1: decide RS type(s) for beam failure detection CSI-RS only SSB only CSI-RS + SSB

Quality measure for beam failure Based on contribution review, companies’ views can be categorized into two L1-RSRP VS SINR-based No obvious dominating support for either one Companies’ position can refer to the summary in the appendix or R1-1715012 (Prague) Proposal 2: down-select between the following two alternatives as quality measure for beam failure detection L1-RSRP SINR-based (hypothetical performance)

RS for candidate beam identification Agreements: In addition to periodic CSI-RS, SS-block within the serving cell can be used for new candidate beam identification The following options can be configured for new candidate beam identification CSI-RS only SS block only FFS: CSI-RS + SS block Proposal 3: decide if the following RS type combination should be supported by NR for new candidate beam identification CSI-RS + SSB

Trigger condition for beam recovery request Working assumption Support at least the following triggering condition(s) for beam failure recovery request transmission: Condition 1: when beam failure is detected and candidate beam is identified at least for the case when only CSI-RS is used for new candidate beam identification FFS Condition 2: Beam failure is detected alone at least for the case of no reciprocity In view of above agreement/WA, support and modification to trigger condition 1 is echoed by many companies Proposal 4: Confirm WA on trigger condition 1 for beam recovery request transmission with following revision “Support at least the following triggering condition(s) for beam failure recovery request transmission: Condition 1: when beam failure is detected and candidate beam is identified at least for the case when only CSI-RS is used for new candidate beam identification”

Beam recovery request resource Working assumption: For beam failure recovery request transmission on PRACH, support using the resource that is CDM with other PRACH resources. Note that CDM means the same sequence design with PRACH preambles. Note that the preambles for PRACH for beam failure recover request transmission are chosen from those for content-free PRACH operation in Rel-15 Note: this feature is not intended to have any impact on design related to other PRACH resources Further consider whether TDM with other PRACH is needed CDM is considered feasible by most companies Proposal 5: Confirm WA For beam failure recovery request transmission on PRACH, support using the resource that is CDM with other PRACH resources Decide to support TDM with other PRACH or not

Dedicated PRACH resource allocation Proposal 6: For non-contention PRACH-based beam recovery request channel, dedicated PRACH resources for beam failure recovery are used irrespective of RS type used for candidate beam identification For SSB only or CSI-RS only, a direct association of a SSB or a CSI-RS resource with a subset of dedicated PRACH resources for beam failure recovery is needed Note: this doesn’t preclude the case that more than one SSB/CSI-RS resources are associated with the same dedicated PRACH resource based on NW configuration For CSI-RS + SSB (if supported), decide on the following A direct association a direct association of a SSB and a CSI-RS resource with a subset of dedicated PRACH resources, respectively, for beam failure recovery is needed A QCL assumption of an CSI-RS resource and a SSB is needed, where the SSB is associated with a subset of dedicated PRACH resources for beam failure recovery

Beam recovery request transmission Agreements: RAN1 agrees that the certain number of beam failure recovery request transmissions is NW configurable by using some parameters Parameters used by the NW could be: Number of transmissions Solely based on timer Combination of above FFS: whether beam failure recovery procedure is influenced by the RLF event Proposal 7: For non-contention PRACH-based beam recovery request transmission, reuse the same mechanisms of initial access preamble transmission designs with potentially different parameter values PRACH preamble sequence Retransmission number constraint Power ramping behavior (e.g. step size, UE Tx beam selection)

Beam recovery request transmission Proposal 8: For gNB response monitoring of non-contention PRACH-based beam recovery request transmission, reuse the same mechanisms of initial access RA response reception designs with potential different parameter values gNB response observation window Common search space for monitoring response Proposal 9: Criteria to decide failure of beam recovery: A timer and a max. number of beam recovery request transmissions Timer expiry or reaching the max. number declare unsuccessful recovery from beam failure

Beam recovery request transmission Agreements: Support the following channel(s) for beam failure recovery request transmission: ⁞ FFS Contention-based PRACH resources as supplement to contention-free beam failure recovery resources From traditional RACH resource pool 4-step RACH procedure is used Note: contention-based PRACH resources is used e.g., if a new candidate beam does not have resources for contention-free PRACH-like transmission Proposal 10: Additionally support contention-based PRACH resources as supplemental beam failure recovery resources

Overall Procedure Proposal 11: decide on the following 2-step BFR procedure Step 1 UE transmits beam failure recovery preamble Step 2 (beam failure recovery response concludes beam failure recovery) 2-step BFR procedure extension Step 2 includes an uplink grant containing gNB request for beam quality report Step 3 UE reports beam quality and concludes the beam failure recovery at lower layer Note: buffer status report can be additionally included, if possible

Beam Failure Recovery Request and Scheduling Request Proposal 12: decide on the following BFR/SR Multiplexing If gNB receives the UE-specific preamble on a non-serving beam of this UE gNB should treat is as beam failure request Otherwise if gNB receives the UE-specific preamble on a serving beam of this UE gNB should treat it as scheduling request

gNB-triggered beam failure recovery Proposal 13: NR supports gNB-triggered beam failure recovery gNB detects beam failure by implementation e.g., based on the measurement of PUCCH/PUSCH DMRS quality or quality of the SRS corresponding to serving beams FFS details of beam recovery procedure, for example: New beam indication in a fallback PDCCH UE monitors a fallback PDCCH with beam sweeping in long term periodicity The fallback PDCCH can be on a carrier different than the carrier on which the indicated beam is transmitted

appendix

Beam Recovery (1/) Trigger condition RS BR channel Monitor response Beam failure detection is based on same L1 IS/OOS as RLM SSB + CSI-RS for new beam detection is not supported in Rel-15 Same RS as RLM for beam failure detection Support direct association between CSI-RS and BRACH Confirm CDM’ed BRACH resource with normal RACH resource Delay PUCCH-based BFRQ Tx until Rel1-6 After BRF, UE continues RLM and UE is allowed to keep search new beam before RLF Max # of BRR Tx is NW configured UE considers BFR unsuccessful when reaching the max # Single search space with a large aggregation level for response HW Configurable evaluation time duration for beam failure Confirm at least trigger condition 1 for Rel-15 Confirm CDM sharing between BRR and PRACH resources Support BRR Tx on multiple beams, irrespective of beam correspondence state (before an observation window) A 2-step procedure similar to contention-free RACH PUCCH for beam reporting is reused for BRR Tx, with candidate beam ID carried To carry partial beam failure info., e.g., failed beam ID When no candidate beam carried, search space is QCL’ed with configured beam set Samsung L1-RSRP based metric for beam failure detection Beam failure based on L1-RSRP can be average RSRP or counting # of worse RSRP in a time window Trigger condition 1 Subset of BPL failure is reported to NW Configurable between either CSI-RS or SSB No contention-based PRACH PUCCH is used to report subset of BPL failure Both max. # of BFRQ transmission and a timer to limit the number of BRR Tx and declare BFRQ failure

Beam Recovery (2/) Trigger condition RS BR channel Monitor response DCM No special handling for “a subset of BPL failure” Beam failure detection based on BLER quality Confirm trigger condition 1 If both conditions are supported, it’s up to NW config. No PUCCH with beam sweeping Configured with either PUCCH or PRACH for BFRQ Support contention-based PRACH Reuse 4-step RACH in initial access, but modify to allow beam report in msg3 Support decoupled beam recovery and RLF, i.e., no indication to higher layer based on results of beam recovery Non-coupled cases can also be supported Declare beam recovery failure if no new beam identified Declare beam recovery failure if UE cannot receive gNB response within a NW configurable timer LGE Use PRACH for beam failure Use PUCCH for subset of serving beam failure New candidate beam condition can be: Quality above a configured/predefined threshold Better quality than serving SSB/CSI-RS Support SSB+CSI-RS for new beam detection Support BRACH CDM’ed to PRACH BRACH resource can be directly associated with CSI-RS SSB BRACH is used if configured CSI-RS beam cannot satisfy predefined condition QCM Handle partial BPL loss in BM framework as much as possible Support a UE initiated message in L1/L2 for requesting further BM steps (for partial BLP loss) Increased # of cyclic shifts for BR PRACH in FDM’ed resources Support configurability between contention-based and non-contention based PRACH Support beam-swept PUCCH with multi-bits formats NW configurable response window One UE request, and one NW response before end of a response window NW configurable max. # of BFRQ A NW configured timer to oversee beam recovery procedure

Beam Recovery (3/) Trigger condition RS Beam recovery request TX NW response ZTE Both RSRP and SINR are supported, their time window parameters are configurable for beam monitor and candidate beam detection For candidate beam identification, support an additional offset value Support trigger condition 1 SSB for beam failure detection in addition to CSI-RS Use of PUCCH, non-contention / contention-based PRACH is NW configurable PUCCH is associated with periodic CRI Non-contention PRACH and contention-based PRACH are associated with SSB by reusing mapping between SSB and PRACH PUCCH with beam sweeping for in-sync UL Tx, Reuse original PRACH sequence and format If no response, RLF indication to higher layer when Max. TX number reached Timer expiry To support the configurability, use “process” for configuration. Each process corresponds to a set of criteria. Intel Use PUCCH for UE with RX beamforming, use PRACH for UE with only omni-directional ant. Or no PUCCH aligned numerology/NW Rx beam for PUCCH/PRACH recovery resources when they are configured at same time instances Support configurable numerology for PUCCH/PRACH recovery resources Without beam correspondence, PUCCH carries CRI explicitly Support a subset of NW beams to receive PUCCH for one UE when no NB beam correspondence Support a configurable timer to oversee BR procedure (restriction based on # of transmission is complex if PUCCH and PRACH are used sequentially) Nokia SSB + CSI-RS for candidate beam detection Map PUCCH to SS/CSI-RS resource by config. Use SSB-specific SR signal for BR, which can be on PRACH or PUCCH Support contention-based PRACH as fallback, when dedicated resource not usable (signal-level based) UE can be configured with dynamic switch between contention-free/contention-based RACH for BR Link-specific resource for BR L2 triggers/initiates BR procedure Retransmission with either contention-free signal or contention-based signal is defined in L2

Beam Recovery (4/) Trigger condition RS Beam recovery request TX NW response MTK L1-RSRP for beam failure detection Configurable parameters for threshold and evaluation method (involve offset between serving and candidate beam) Confirm trigger condition 1 but extend it for SSB. Prioritize non-contention PRACH based use cases Reuse 4-step RACH as in initial access as supplement Either PUCCH or non-contention based PRACH, no simultaneous configuration RAR and reTx behavior reuse RACH procedure for non-contention PRACH A timer to oversee BR procedure, and # of max. Tx to constrain PRACH Tx AT&T Beam failure recovery can be run on 1 or multiple BPLs corresp. to a CORESET group CSI-RS + SS-blocks for new beam detection Unified indexing framework for CSI-RS and SSB If configured with CSI-RS, CSI-RS is tried before SSB is tried PUCCH is configured per CORESET group PUCCH corresponding to functioning CORESET groups are used to carried info on failure CORESET groups Contention-based PRACH is supported Pure L1 procedure BFR fail/succeed triggers an IS/OOS to RLF procedure RLM/RLF timers and thresholds configurable based on service require. InterDigital Support (direct) association between CSI-RS and non-contention based PRACH Support (direct) association between SSB and non-contention based PRACH Implicit beam indication by resource selection for non-contention PRACH Support TDM’ed BRACH with normal RACH resources PUCCH for BRR Tx method is used on top of non-contention PRACH as complementary RLF is triggered after a max. # of BR request Tx, or a timer expiry Introduce a fallback scheme (and corresp. time window) search space during beam recovery

Beam Recovery (5/) Trigger condition RS BR channel Monitor response CATT No additional definition for beam failure No SSB+CSI-RS for new beam identification No PUCCH for BFRQ No simultaneous configuration of PUCCH and non-contention PRACH 4-step RACH are supported # of BFRQ Tx is determined by a time window, not by direct NW configuration Discuss when to start monitoring UE-identified beam for response Discuss when to continue or stop monitoring failed beam, after BFRQ Tx vivo Specify beam failure event details, e.g., threshold, timer L1-RSRP as quality measure Allow BRQ Tx when an entire UE panel is blocked, even beam failure is not detected yet Not supporting condition 2 Both SSB and CSI-RS can be configured for candidate beam at the same time When no beam correspondence, beam sweeping behavior as initial access is followed PUCCH at LF can be used Both PUCCH and PRACH can be used when configured Not supporting contention-based PRACH Unsuccessful recovery includes that UE finds no candidate beam Two options for handling unsuccessful recovery Single beam failure recovery timer Two separate timer for candidate beam detection and BFRQ TX. Expiry of either one triggers failure No higher layer indication when successful indication Upon unsuccessful recovery, UE no longer monitors serving BPLs but wait for RLF only OPPO L1-RSRP for beam failure detection Support UE-triggered DL beam measurement reporting to maintain backup/candidate beams at gNB side, upon a subset of BPL loss Same RS type for beam failure and candidate beam detection No support for SSB + CSI-RS No PUCCH beam sweeping Support CDM’ed BRACH resource with normal RACH resource For BFRQ, support following Configured with PRACH Configured with both contention-based PRACH + PUCCH, but up to UE for selection one at a time.

Beam Recovery (6/) Trigger condition RS BR channel Monitor response Sharp Beam failure detection based on BLER Support condition 1 but removing the constraint on using CSI-RS for candidate beam SS-block for candidate beam detection UE is configured with either PUCCH or non-contention PRACH Support contention-based PRACH as fallback Contention-based RACH is used when UL out-of-sync, or # BFRQ reaches a max. # NI A trigger condition consists of a measurement metric related threshold and time window, threshold and time window is configurable Multiple trigger condition sets are configured (trigger event) Support UE and gNB failure triggering CSI-RS nested in SSB beam (aligned with ATT) Intermediate states allow improved handling of failure by transitioning to other beams within SSB Beam handoff and recovery possible with low-latency and possibly handled exclusively within L1 Lenovo, MM Support only condition 1, and extends its applicability to SSB MAC-CE to activate / deactivate the use of BR non-contention PRACH (RSRP for BF detection) Change config option to Candidate beam detection SSB only SSB+CSI-RS Support only FDM and TDM Contention-based PRACH is used if non-contention PRACH is not configured Tx BRR in first available PUCCH or PRACH For non-contention PRACH, use time/freq. to differentiate candidate beams, and sequence to differentiate UEs Response is monitoring in a configured window, either for PRACH or PUCCH

Beam Recovery (7/) Trigger condition RS BR channel Monitor response SONY gNB response may trigger aperiodic reporting Support multiple beam reporting opportunities to split report into groups for latency reduction NW can indicate to ignore subsequent report opportunities if a beam is found in previous report NEC NW may predict potential candidate beam based on e.g., Tx direction and UE position NW may use a different beam (e.g., refined beam) than indicated by UE for response delivery. The response can be followed by a beam training