3GPP TSG RAN WG1 Meeting 90bis
`Prague, CZ, 9th – 13th October 2017
`Agenda Item: 7.3.4.1
`Source: MediaTek Inc.
`Title: Remaining Details on Bandwidth Part Operation in NR
`Document for: Discussion
`
` R1-1718327
`
`o
`
`1 Introduction
`Until 3GPP RAN1 AH#3 (September 2017), agreements on bandwidth part (BWP) operation can be summarized as
`follows.
`• Usage scenarios of BWP operation includes the following
`o Enabling reduced UE bandwidth capability within a wideband carrier
`o Enabling reduced UE power energy consumption by bandwidth adaptation
`• Relationship between CA & BWP
`o For each UE-specific serving cell, one or more DL BWPs and one or more UL BWPs can be
`configured by dedicated RRC for a UE
` FFS association of DL BWP and UL BWP
` FFS definition of an active cell in relation to DL BWP and UL BWP, whether or not there are
`cross-cell/cross-BWP interactions
`In Rel-15, for a UE, there is at most one active DL BWP and at most one active UL BWP at a given
`time for a serving cell
`• Initial active BWP
`o There is an initial active DL/UL bandwidth part pair to be valid for a UE until the UE is explicitly
`(re)configured with bandwidth part(s) during or after RRC connection is established
` The initial active DL/UL bandwidth part is confined within the UE minimum bandwidth for the
`given frequency band
` FFS: details of initial active DL/UL bandwidth part are discussed in initial access agenda
`• BWP configuration
`o A bandwidth part consists of a group of contiguous PRBs
` The bandwidth size ranges from the SS block bandwidth to the maximal bandwidth capability
`supported by a UE in a component carrier
` The bandwidth part may or may not contain SS block
` Reserved resources can be configured within the bandwidth part
`o For a connected-mode UE, one or multiple bandwidth part configurations for each component carrier
`can be semi-statically signaled to a UE and configuration parameters include
` Numerology (i.e. CP type, subcarrier spacing)
` Frequency location (the offset between BWP and a reference point is implicitly or explicitly
`indicated to UE) based on common PRB index for a give numerology
` Bandwidth size (in terms of PRBs)
` CORESET (required for each BWP configuration in case of single active DL bandwidth part
`for a given time instant)
`o Separate sets of bandwidth part configurations for DL & UL per component carrier
` For TDD, a UE is not expected to retune the center frequency of channel BW between DL and
`UL if different active DL and UL BWPs are configured for the UE
`o Search space type in CORESET
` At least one of configured DL BWPs includes one CORESET with common search space in
`Pcell
` Each configured DL BWP includes at least one CORESET with UE-specific search space for
`the case of single active BWP at a given time per component carrier
`• Active BWP operation
`o A UE is only assumed to receive/transmit within active DL/UL bandwidth part(s) using the associated
`numerology
` At least PDSCH and/or PDCCH for DL and PUCCH and/or PUSCH for UL
`
` Ex. 1019
`APPLE INC. / Page 1 of 13
`
`

`

`•
`
`o
`
`o UE expects at least one DL bandwidth part and one UL bandwidth part being active among the set of
`configured bandwidth parts for a given time instant
` Primary focus is to complete the single DL/UL active bandwidth part case
` If time is available later after completing the single active bandwidth part case, multiple active
`bandwidth parts with different numerologies for a UE should be considered
`In case of single active DL BWP for a given time instant in a component carrier
` A UE can assume that PDSCH and corresponding PDCCH (PDCCH carrying scheduling
`assignment for the PDSCH) are transmitted within the same BWP if PDSCH transmission
`starts no later than K symbols after the end of the PDCCH transmission.
` In case of PDSCH transmission starting more than K symbols after the end of the
`corresponding PDCCH, PDCCH and PDSCH may be transmitted in different BWPs
`• BWP activation/deactivation
`o Activation by dedicated RRC signaling
`o Activation/deactivation by scheduling DCI with explicit indication
`o Activation/deactivation by a timer for a UE to switch its active DL bandwidth part to a default DL
`bandwidth part
` The default DL bandwidth part can be the initial active DL bandwidth part defined above
`o Specify necessary mechanism to enable UE RF retuning for active bandwidth part switching
`
`However, there are still some remaining issues as follows and this paper provides our views on the highlighted ones.
`• Relationship between CA & BWP
`– When an Scell is activated, which DL BWP and which UL BWP are active?
`– When an Scell is deactivated, what happen to active DL/UL BWP?
`Initial active BWP → To be addressed by initial access agenda
`– Whether the configuration of initial active DL BWP is the same as that for RMSI and its
`corresponding CORESET
`– Where to signal the configuration of initial active UL BWP
`• BWP configuration
`– Association or pairing of DL BWP and UL BWP?
`– Cross-slot scheduling in BWP?
`– Common search space support
`– PUCCH resource configuration across BWP
`– Max number of BWP configurations per carrier
`• Active DL/UL BWP switching
`– Timer-based active DL BWP switching
`• Whether default DL BWP is configurable or not
`• New timer or reuse DRX timer
`• Triggering conditions of the timer
`– Other remaining details of DCI-based DL/UL BWP switching?
`– Transition time of active BWP switching
`• Active BWP operation
`– AGC & synchronization tracking within active DL BWP, especially for that containing no SS-block
`– Whether DCI size is dependent on the bandwidth of the active DL/UL BWP
`– Whether support cross-BWP retransmission or not
`– Whether support overlap BWP configuration from network and UE side or not
`– BWP operation in DRX mode
`• RRM/CSI measurement & SRS transmission
`– Whether to support CSI measurement outside active BWP
`– Reuse BWP configuration or separate configuration for RRM measurement
`
`2 Relationship between CA & BWP
`Until RAN1 AH#3, related agreements are shown as follows.
`
`Agreements: (RAN1 #88)
`• Resource allocation for data transmission for a UE not capable of supporting the carrier bandwidth can be
`derived based on a two-step frequency-domain assignment process
`
` Ex. 1019
`APPLE INC. / Page 2 of 13
`
`

`

`o 1st step: indication of a bandwidth part
`o 2nd step: indication of the PRBs within the bandwidth part
`o FFS definitions of bandwidth part
`o FFS signaling details
`• FFS the case of a UE capable of supporting the carrier bandwidth
`Agreements: (RAN1 2017 AH#3)
`– For each UE-specific serving cell, one or more DL BWPs and one or more UL BWPs can be configured by
`dedicated RRC for a UE
`o FFS association of DL BWP and UL BWP
`o FFS definition of an active cell in relation to DL BWP and UL BWP, whether or not there are cross-
`cell/cross-BWP interactions
`Agreements: (RAN1 2017 AH#3)
`– In Rel-15, for a UE, there is at most one active DL BWP and at most one active UL BWP at a given time for a
`serving cell
`
`There was one FFS on definition of an active cell in relation to DL BWP and UL BWP in the agreements. From our
`views, there is no need to discuss the definition of an active cell as long as a clear mechanism to indicate active BWP
`when an Scell is activated and what happen to active BWP when an Scell is deactived.
`Q1: When an Scell is activated, which DL BWP and/or which UL BWP are active?
`There could be two options to indicate which DL BWP and which UL BWP are active.
`– Option #1: Indication in RRC signalling for Scell configuration/reconfiguration (shown in Figure 1)
`o RRC signalling for Scell configuration/reconfiguration is used for indicating which DL BWP and/or
`which UL BWP are initially activated when the Scell is activated
`It allows to decouple the discussion of CA and BWP operations no matter which signalling (either
`o
`DCI or MAC CE) is used for Scell activation
`– Option #2: Indication in Scell activation signalling
`o Scell activation signaling is used for indicating which DL BWP and/or which UL BWP are initially
`activated when the Scell is activated
`Joint design for CA activation and BWP activation needs to be considered and it may relate to cross-
`o
`carrier scheduling if DCI is used for Scell activation
`
`CA & BWP RRC-layer Config.
`
`BWP Switch by DCI
`
`BWP Config. For PCell
`BWP #1
`BWP #2
`
`BWP #1
`
`BWP #2
`
`BWP #1
`
`PCell
`
`Initial BWP
`
`PBCH
`
`SCell Activation Signaling
`
`BWP Switch by DCI
`
`BWP Config. For SCell
`BWP #1
`BWP #2
`
`BWP #1
`
`BWP #2
`
`BWP #1
`
`Frequency
`
`SCell
`
`CONNECTED Mode Operation
`Initial Access
`Single Activated Cell Multiple Activated Cells
`
`Time
`
`SS-block
`
`
`
`Figure 1. Illustration of indication in RRC signalling for Scell configuration/reconfiguration (Option #1)
`Due to limited time, our preference is Option #1 because it allows to decouple the discussion of CA and BWP
`operations. Further optimization on CA operation can be considered in future releases.
`
` Ex. 1019
`APPLE INC. / Page 3 of 13
`
`

`

`Q2: When an Scell is deactivated, what happens to active DL BWP and/or active UL BWP?
`
`From our views, all active BWP(s) with a Scell should be deactivated autonomously without any explicit indication
`when the Scell is deactivated.
`
`Proposal #1: For an Scell, RRC signaling for Scell configuration/reconfiguration is used for indicating which DL
`BWP and/or which UL BWP are initially activated when the Scell is activated.
`
`Proposal #2: For an Scell, active DL BWP and/or active UL BWP are deactivated autonomously when the Scell is
`deactivated.
`
`3 BWP Configuration
`
`3.1 Association between DL BWP and UL BWP
`In RAN1 AH#3, the following is agreed and there is one FFS on the association between DL BWP and UL BWP.
`
`Agreements: (RAN1 2017 AH#3)
`– For each UE-specific serving cell, one or more DL BWPs and one or more UL BWPs can be configured by
`dedicated RRC for a UE
`o FFS association of DL BWP and UL BWP
`o FFS definition of an active cell in relation to DL BWP and UL BWP, whether or not there are cross-
`cell/cross-BWP interactions
`In RAN1 AH#2, the following is agreed so separate sets of BWP configurations for DL & UL per serving cell are
`supported in R15.
`
`Agreement: (RAN1 2017 AH#2)
`• For FDD, separate sets of bandwidth part (BWP) configurations for DL & UL per component carrier
`o The numerology of DL BWP configuration is applied to at least PDCCH, PDSCH & corresponding
`DMRS
`o The numerology of UL BWP configuration is applied to at least PUCCH, PUSCH & corresponding
`DMRS
`• For TDD, separate sets of BWP configurations for DL & UL per component carrier
`o The numerology of DL BWP configuration is applied to at least PDCCH, PDSCH & corresponding
`DMRS
`o The numerology of UL BWP configuration is applied to at least PUCCH, PUSCH & corresponding
`DMRS
`o For UE, if different active DL and UL BWPs are configured, UE is not expected to retune the center
`frequency of channel BW between DL and UL
`From our views, there is no use case to support the association between DL BWP and UL BWP within a FDD serving
`cell. The association between DL carrier and UL carrier within a serving cell can be done by carrier association.
`However, there is one use case for TDD system, i.e. UE is not expected to retune the center frequency of channel BW
`between DL and UL. To achieve it, certain association between DL BWP and UL BWP may be needed. One way to
`associate them is to group DL BWP configurations with same center frequency as one set of DL BWPs and group UL
`BWP configurations with same center frequency as one set of UL BWPs. Then, the set of DL BWPs can be associated
`with the set of UL BWPs sharing the same center frequency. However, the question is whether to autonomously switch
`active UL BWP when UE’s active DL BWP is switched to another one with different center frequency. Therefore, we
`have the following proposal.
`
`Observation #1: For FDD system, the association between DL carrier and UL carrier within a serving cell can be done
`by carrier association, which is signaled in RMSI.
`
`Proposal #3: For an FDD serving cell, no association between DL BWP and UL BWP is supported in R15.
`– FFS TDD case
`
` Ex. 1019
`APPLE INC. / Page 4 of 13
`
`

`

`3.2 Cross-slot scheduling in BWP
`Until RAN1 2017 AH#3, agreements related to DL scheduling timing are shown as follows.
`
`Agreements: (RAN86bis)
`• NR supports at least same-slot and cross-slot scheduling for DL.
`o Note: it is already agreed that NR supports same-slot and cross-slot scheduling for UL.
`• For slot-based scheduling, NR specification should support the following
`o DL data reception in slot N and corresponding acknowledgment in slot N+K1
` All UEs should support K1≥1 with exact values for K1 FFS
` Some UEs may support K1=0 (FFS conditions)
`o UL assignment in slot N and corresponding uplink data transmission in slot N+K2
` All UEs should support K2≥1 with exact values for K2 FFS
` Some UEs may support K2=0 (FFS conditions)
`Agreements: (RAN89)
`• All Rel. 15 UE supports minimum value of K0 equal to 0, i.e., DL assignment and the scheduled DL data are in
`the same slot.
`Cross-slot scheduling in the downlink with nonzero K0 presents significant opportunities for further power saving in the
`UE when it operates in low-BW BWP. The typical PDSCH/PUSCH reception procedure is shown in the below Figure
`2a. In existing LTE, PDCCH of subframe #k schedules PDSCH of subframe #k. Since UE does not know if there is its
`own PDSCH until UE finishes PDCCH decoding, UE needs to keep RX RF on and buffer PDSCH region for a while
`even there is no its own PDSCH. In case of PDCCH-only, this is waste of current consumption. If cross-slot
`scheduling concept is introduced, as shown in Figure 2b, current consumption can be reduced. PDCCH of subframe #k
`schedules PDSCH of subframe #(k+1). After receiving PDCCH part, RX RF can be turned off and baseband continues
`to decode PDCCH. If there is desired DCI, UE receives PDSCH in subframe #(k+1). In this way, there is no waste of
`RX RF due to uncertain PDSCH reception.
`
`Figure 2. RX RF on duration comparison with/without cross-slot scheduling
`
`According to Table 1, support cross-slot scheduling (i.e. K0 = 1) in low-BW BWP can save 59.3% and 23.9% UE
`power consumption, respectively, compared to same-slot scheduling with full BW and low BW.
`
`
`
`Table 1. Power consumption comparison for K0 = 0 with full/low BW & K0 = 1 with low BW for no data case
`No Data for UE
`K0 = 0
`K0 = 0
`K0 = 1
`with Full BW
`with Low BW
`with Low BW
`Symbol
`Power
`Symbol
`Power
`Symbol
`Power
`Duration
`Contribution
`Duration
`Contribution
`Duration
`Contribution
`
`Relative
`Power
`
`Power State
`
` Ex. 1019
`APPLE INC. / Page 5 of 13
`
`

`

`Full BW Rx
`
`100%
`
`5.5/14
`
`39.0%
`
`0
`
`0
`
`Low BW Rx
`
`DCI Processing
`Only
`Micro-sleep
`
`TOTAL
`
`50%
`
`33%
`
`5%
`
`
`
`0
`
`0
`
`0
`
`0
`
`8.5/14
`
`14/14
`
`3.0%
`
`42.3%
`
`5.5/14
`
`19.6%
`
`0
`
`8.5/14
`
`14/14
`
`0
`
`3.0%
`
`22.6%
`
`0
`
`1/14
`
`4.5/14
`
`8.5/14
`
`14/14
`
`0
`
`3.6%
`
`10.6%
`
`3.0%
`
`17.2%
`
`
`
`Therefore, if NR supports K0 = 0 and 1 and BWP configuration can further include the configuration of K0 value, it’s
`helpful to further reduce UE power consumption. For high-BW BWP, multiple K0 values can be configured to a UE
`and a DCI field can be introduced to indicate to the UE which K0 value is applied for the DL scheduling. When there is
`no active BWP switching, K0 = 0 can be applied for PDSCH scheduling. When there is active BWP switching, K0 = 1
`can be applied for PDSCH scheduling to allow sufficient transition time. For low-BW BWP, single K0 value (i.e. K0 =
`1) can be configured to a UE for further power saving if latency is not an issue.
`
`Observation #2: Support cross-slot scheduling (i.e. K0 = 1) in low-BW BWP can save 59.3% and 23.9% UE power
`consumption, respectively, compared to same-slot scheduling with full BW and low BW.
`
`Proposal #4: NR supports K0 = 0 and 1 and BWP configuration further includes the configuration of one or
`multiple K0 values.
`– If multiple K0 values are configured to a UE, a DCI field is used to indicate to the UE which value is
`applied for the DL scheduling
`
`3.3 CSS support in BWP
`In RAN1 NR AH#2, it was agreed as follows.
`• At least one of configured DL BWPs includes one CORESET with common search space at least in primary
`component carrier
`• Each configured DL BWP includes at least one CORESET with UE-specific search space for the case of single
`active BWP at a given time
`–
`In case of single active BWP at a given time, if active DL BWP does not include common search space,
`then UE is not required to monitor the common search space
`Though it was agreed that at least one of configured DL BWPs includes one CORESET with common search space
`(CSS) at least in primary component carrier, it doesn’t resolve the issue on how a connected-mode UE monitors DCIs
`for system information (SI) update, RACH response (msg2), pre-emption indication and other group-based commands
`if there is no common search space in an active bandwidth part. Since SI update can be done by UE-specific signalling
`and no group-based commands except pre-emption indication agreed so far, CSS is only needed for RACH response
`(msg2) and pre-emption indication. There could be two alternatives to resolve the issue in bandwidth part operation.
`• Option #1: No periodic gap for RACH response monitoring on Pcell
`o For Pcell, one of configured DL bandwidth parts includes one CORESET with the CSS type for RMSI
`& OSI
`o For Pcell, each configured DL bandwidth part includes one CORESET with the CSS type for RACH
`response & paging control for system information update
`o For each serving cell, each configured DL bandwidth part includes one CORESET with the CSS type
`for pre-emption indication and other group-based commands.
`o Pros: Periodic time gap for DCI monitoring for RMSI, OSI, RACH response & paging control is not
`required
`o Cons: Duplicated signalling overhead for DCI for RACH response & paging control on Pcell
`• Option #2: Periodic gap for RACH response monitoring on Pcell
`o For Pcell, one of configured DL bandwidth parts includes one CORESET with CSS type for RMSI,
`OSI, RACH response & paging control for system information update
`o For each serving cell, each configured DL bandwidth part includes one CORESET with the CSS type
`for pre-emption indication and other group-based commands.
`o Pros: No duplicated signalling overhead for DCI for RACH response & paging control
`o Cons: Periodic time gap for CSS & GCSS monitoring is required
`
` Ex. 1019
`APPLE INC. / Page 6 of 13
`
`

`

`Figure 1 illustrates an example for Option #2. Since there could be many gaps for intra-frequency/inter-frequency
`RRM measurement, beam measurement, reception of SS-block which uses different numerology from PDCCH/PDSCH
`& SS-block based RLM according to RAN4’s study, introducing additional gaps for common search space monitoring
`could result in large UE throughput loss. To minimize the gap number in DL, from our views, Option #1 is preferred.
`
`Frequency
`
`Case #1
`
`BWP #1
`
`Periodic Time gap
`BWP #1
`
`Active BWP for a UE
`BWP #2
`
`Common search space
`BWP #2
`
`BWP #1
`
`Carrier
`
`Carrier
`
`Time
`
`
`
`Case #2
`
`BWP #1
`
`Figure 3. Example illustration of Option #2
`
`Observation #3: There are many use cases (shown as follows) requiring gaps and too many gaps will degrade UE
`throughput performance so RAN1 should strive to minimize the number of gaps.
`– Intra-frequency/inter-frequency RRM measurement
`– Beam measurement
`– Reception of SS-block using different numerology from PDCCH/PDSCH
`– SS-block based RLM
`
`Proposal #5: Support the following for CSS monitoring on Pcell.
`– For Pcell, one of configured DL bandwidth parts includes one CORESET with the CSS type for RMSI &
`OSI
`– For Pcell, each configured DL bandwidth part includes one CORESET with the CSS type at least for
`RACH response
`o FFS paging control for system information update
`– For each serving cell, each configured DL bandwidth part includes one CORESET with the CSS type for
`pre-emption indication and other group-based commands.
`
`3.4 Maximal number of configured BWPs
`In RAN1 NR AH#2 (June 2017), the following use cases of BWP operation was concluded as observation.
`1. Use case #1: Support reduced UE bandwidth within a wideband carrier
`2. Use case #2: Support UE bandwidth adaptation for UE power saving
`3. Use case #3: Support FDM of different numerology (i.e. CP type, subcarrier spacing)
`To support above three use cases, from our views, the maximal number of configured BWP can be 8 with the following
`reasons though use case #3 is deprioritized due to the pre-requisite of multiple active BWPs.
`1. To support use case #1, it requires at least 1 configured BWP but, considering load balancing within a wideband
`carrier, it would be better if 2 BWPs with different frequency locations can be configured to a UE. It means that
`the maximal number of configured BWPs can be 2.
`2. To support use case #2, it requires at least 2 configured BWP but, considering load balancing within a wideband
`carrier, it would be better if 2 pairs of BWPs (2 BWPs for each pair) with different frequency locations can be
`configured to a UE. It means that the maximal number of configured BWPs can be 4.
`3. To support use case #3, it requires at least 2 configured BWP with different numerology but, considering load
`balancing within a wideband carrier & use case #2, it would be better if 2 pairs of BWPs (2 BWPs for each pair)
`with different frequency locations for a given numerology. It means that the maximal number of configured
`BWPs can be 8 if up to two different numerology are considered.
`
`Figure 1 illustrates an example on maximal number of configured BWPs to support different use cases.
`
` Ex. 1019
`APPLE INC. / Page 7 of 13
`
`

`

`Numerology #1
`
`Numerology #2
`
`BWP #1
`
`BWP #1
`
`BWP #2
`
`BWP #1
`
`BWP #2
`
`BWP #3
`
`BWP #4
`
`BWP #2
`
`BWP #3
`
`BWP #4
`
`BWP #5
`
`BWP #6
`
`BWP #7
`
`BWP #8
`
`Wideband Carrier @ gNB
`
`
`Figure 4. Maximal number of configured BWPs, considering different use cases with load balancing
`
`Use Case #1
`
`Use Case #1+2
`
`Use Case #1+2+3
`
`Proposal #6: The maximal number of configured BWPs is 4 or 8 in NR R15.
`
`4 Active DL/UL BWP Switching
`
`4.1 Timer-based active DL BWP switching
`Until RAN1 2017 AH#3, agreements related to active DL/UL BWP switching are shown as follows.
`
`•
`
`•
`
`Agreements: (RAN1 2017 AH#2)
`• Activation/deactivation of DL and UL bandwidth parts can be
`•
`by means of dedicated RRC signaling
`• Possibility to activate in the bandwidth part configuration
`by means of DCI (explicitly and/or implicitly) or MAC CE [one to be selected]
`•
`by means of DCI could mean
`• Explicit: Indication in DCI (FFS: scheduling assignment/grant or a separate
`DCI) triggers activation/deactivation
`•
`Separate DCI means DCI not carrying scheduling assignment/grant
`Implicit: Presence of DCI (scheduling assignment/grant) in itself triggers
`activation/deactivation
`•
`This does not imply that all these alternatives are to be supported.
`• FFS: by means of timer
`• FFS: according to configured time pattern
`Agreements: (RAN1#90)
`•
`Support activation/deactivation of DL and UL bandwidth part by explicit indication at least in (FFS:
`scheduling) DCI
`– FFS: In addition, MAC CE based approach is supported
`Support activation/deactivation of DL bandwidth part by means of timer for a UE to switch its active DL
`bandwidth part to a default DL bandwidth part
`– The default DL bandwidth part can be the initial active DL bandwidth part defined above
`– FFS: The default DL bandwidth part can be reconfigured by the network
`– FFS: detailed mechanism of timer-based solution (e.g. introducing a new timer or reusing DRX timer)
`– FFS: other conditions to switch to default DL bandwidth part
`Agreements: (RAN1 2017 AH#3)
`– NR supports the case that a single scheduling DCI can switch the UE’s active BWP from one to another (of the
`same link direction) within a given serving cell
`o FFS whether & how for active BWP switching only without scheduling (including the case of UL
`scheduling without UL-SCH)
`
`•
`
`The discussion of this section focuses on the following two issues.
`
` Ex. 1019
`APPLE INC. / Page 8 of 13
`
`

`

`Q1: New timer or reuse DRX timer
`Q2: Triggering/resetting conditions of the timer
`For Q1, our preference is new timer due to the following reasons.
`• The considerations of DRX inactivity timer configuration and the timer configuration for timer-based active DL
`BWP switching are very different
`o DRX inactivity timer usually requires larger value to reduce the latency of potential DL data packet
`transmission
`o The timer for timer-based active DL BWP switching requires smaller value to maximize UE power
`saving gain
`• Enforcing two different schemes to share the same timer may defeat their own design purpose and will
`complicate the designs
`• From UE perspective, the introduced complexity of one new timer is negligible
`For Q2, the triggering & resetting conditions of the timer for timer-based active DL BWP switching are shown as
`follows. An example is illustrated in Figure 5.
`• Timer-triggering condition: The timer is triggered when UE receives a DCI to switch its active DL BWP from
`the default BWP to another
`• Timer-resetting condition: The timer is reset when a UE receives a DCI to schedule PDSCH(s) in the BWP other
`than the default BWP
`
`BWP Switch by DCI
`Slot
`
`With data scheduling
`No data scheduling
`
`BWPs For a Cell
`Default
`BWP #2
`BWP
`
`D. BWP
`
`BWP #2
`
`BWP #2
`
`D. BWP
`
`Timer is Triggered
`
`Timer is Reset
`
`Timer Expires
`
`
`
`Figure 5. Example illustration of timer-based active DL BWP switching
`
`Observation #4: Technical reasons for a new timer at least include the following.
`– The considerations of DRX inactivity timer configuration and the timer configuration for timer-based active DL
`BWP switching are very different
`– Enforcing two different schemes to share the same timer may defeat their own design purpose and will
`complicate the designs
`– From UE perspective, the introduced complexity of one new timer is negligible
`
`Proposal #7: Support a new timer for timer-based active DL BWP switching.
`
`Proposal #8: Support the timer-triggering & timer-resetting conditions for the new timer as follows.
`– A UE triggers the timer when it successfully decodes a DCI to switch its active DL BWP from the default
`DL BWP to another
`– A UE reset the timer when it successfully decodes a DCI to schedule PDSCH(s) in the BWP other than the
`default DL BWP
`
` Ex. 1019
`APPLE INC. / Page 9 of 13
`
`

`

`5 Active BWP Operation
`
`5.1 AGC & synchronization tracking
`According to the reply LS from RAN4, AGC is not required for the bandwidth part switch in single-carrier operation
`due to the assumption of the same cell (i.e. same carrier and same gNB). However, time/frequency synchronization
`tracking could be another critical issue because wider bandwidth part (e.g. 275 PRBs) requires finer timing resolution
`for better data channel decoding performance and it can’t be achieved by PSS/SSS which occupy 127 REs only. In
`addition, bandwidth part may not contain PSS/SSS. Therefore, wideband reference signals are beneficial for a UE to
`fine-tune its time/frequency synchronization accuracy & obtain accurate AGC level to allow data scheduling with high
`MCS level immediately after bandwidth part configuration switch. Potential options for the wideband reference signals
`could be as follows. Our preference is Option #1 and FFS the necessity of Option #2.
`
`• Option #1: Periodic wideband time/frequency tracking reference signal (TRS) if it exists
`• Option #2: Aperiodic wideband time/frequency tracking reference signal (TRS) triggered by the bandwidth part
`configuration switch signalling
`
`Observation #5: There may be AGC & time/frequency tracking issues due to the following.
`– Active bandwidth part may not contain SS-block
`– Active bandwidth part is switched from bandwidth part configuration with narrow bandwidth to that with wide
`bandwidth
`
`
`Proposal #9: DL bandwidth part configuration further includes periodic TRS configuration for time/frequency
`tracking & AGC settling.
`– FFS the necessity of aperiodic TRS triggered by DCI-based activation/deactivation signalling
`
`5.2 BWP operation in DRX mode
`There is another issue regarding to active BWP operation during UE’s DRX mode. It involves at least two questions.
`
`Q1: Which DL BWP should be assumed for a UE as the active DL BWP by default during DRX on duration?
`Q2: Whether it’s necessary to support active DL/UL BWP switching for a UE in DRX mode and how?
`For Q1, for better power saving gain, it’s reasonable for a UE to assume the default DL BWP as the active DL BWP by
`default for DCI monitoring during DRX-ON duration. However, it’s also beneficial to allow the network to switch a
`UE’s active DL BWP from the default DL BWP to the one with wider bandwidth to shorten the time for DL data
`reception when there is DL data scheduling. For active UL BWP indication/switching, it can be indicated by UL
`scheduling DCI. It’s also beneficial to support timer-based active DL BWP switching to allow UE’s faster fall-back to
`the default BWP when there is no DL data scheduling. Therefore, for Q2, we think both scheduling DCI-based active
`DL/UL BWP switching and timer-based active DL BWP switching should be supported in DRX mode. Figure 6
`illustrates an example of BWP operation in DRX mode.
`
`Active BWP Switching DCI
`
`DL Data Reception
`
`Inactivity Timer
`
`Time
`Timer for default BWP fall-back
`Inactivity Timer
`
`PDCCH Monitoring
`DRX Cycle
`DRX-OFF
`
`DRX w/o
`BWP
`
`Bandwidth
`
`DRX w
`BWP
`
`DRX-ON
`DRX Cycle
`
`DRX-OFF
`
`DRX-ON
`
`Active BWP Switching DCI
`
`Figure 6. Example illustration of BWP operation in DRX mode
`
`Time
`
`
`
` Ex. 1019
`APPLE INC. / Page 10 of 13
`
`

`

`Proposal #10: In DRX mode, a UE should assume that the default DL BWP is the active DL BWP by default
`during DRX-ON duration.
`
`Proposal #11: Support scheduling DCI-based active DL/UL BWP switching in DRX mode.
`
`Proposal #12: Support timer-based active DL BWP switching in DRX mode.
`
`6 RRM/CSI Measurement & SRS Transmission
`In RAN1 #90, the following agreements is achieved for measurement and SRS transmission.
`Agreements: (RAN1#90)
`• When a UE performs measurement or transmit SRS outside of its active BWP, it is considered as a
`measurement gap
`– FFS: details of measurement gap configuration
`– During the measurement gap, UE is not expected to monitor CORESET
`However, it’s not clear whether to support both RRM measurement and CSI measurement outside active/configured DL
`BWPs or just RRM measurement outside active/configured DL BWPs. Since CSI measurement requires higher UE
`complexity and CSI reporting is mainly for MIMO scheduling, it should be limited in active DL BWP, not all
`configured BWPs. Regarding how to assign MCS for 1st several DL data packets after active DL BWP switching, it can
`be resolved by the following.
`1. The network assigns robust MCS to a UE for 1st several DL data packets based on RRM measurement reporting
`2. The network signals to a UE by active DL BWP switching DCI to trigger aperiodic CSI measurement/reporting
`to speed up CQI convergence
`In addition, allowing periodic CSI measurement outside active DL BWP may introduce new gap in addition to the
`following cases and thus degrade UE throughput performance. However, aperiodic CSI measurement outside active DL
`BWP can be considered.
`• Intra-frequency/inter-frequency RRM measurement
`• Beam measurement
`• Reception of SS-block which uses different numerology from PDCCH/PDSCH
`• SS-block based RLM
`For radio resource management (e.g. BWP switch or BWP reconfiguration), RRM measurement is more suitable. For
`beam management, at least CSI-RS RRM measurement/reporting in active BWP is necessary. However, CSI-RS RRM
`measurement/reporting in other configured BWP is also beneficial for the network to know which BWP has better
`channel condition and able to switch a UE from one BWP to another. Regarding whether to reuse BWP configurations
`for CSI-RS RRM measurement, it’s related to RRC signalling design and should be up to RAN2’s decision. Since not
`all configured BWPs include SS-block, SS-block RRM measurement/reporting can’t be done in all configured BWPs.
`In addition, there may be some SS-bl