3GPP TSG RAN WG1 Meeting 91
`Reno, USA, November 27th – December 1st, 2017
`
`R1-1720097
`
`Intel Corporation
`Source:
`Remaining details of UL data transmission procedures in NR
`Title:
`7.3.3.4
`Agenda item:
`Document for: Discussion and Decision
`
`1 Introduction
`In the previous meeting, the common aspects of waveform determination, multi-slot transmission, frequency hopping, and
`others were discussed and some progress was achieved. In this contribution, we present our view on UL data transmission
`procedures including enhancement on grant-based UL transmission and uplink grant-free transmission. In particular, the
`common aspects for grant-based and grant-free transmissions are discussed in section 2, the SR aspects for grant-based
`transmission are discussed in section 3, and the remaining issues of Type 1 and Type 2 grant-free transmissions are
`discussed in section 4.
`
`2 Common UL Transmission Aspects
`2.1 Configuration of Repetitions
`The repetitions and multi-slot transmission were extensively discussed at the last meeting. It was agreed that TB spanning
`multiple slots can be realized by the repetitions of a TB with RV cycling. However, the following aspects need to be further
`discussed for both grant-based and grant-free UL transmissions schemes:
` Support of non-continuous in time repetitions. The support on non-continuous in time repetitions may not be justified for
`grant-based scheduling. The gNB schedules resources in a short window and can select frequency resources based on
`dynamic channel quality estimations. In this case, attempt to achieve time diversity for grant-based transmissions is not
`justified. However, for grant-free UL transmissions, the channel quality based scheduling is not easily possible,
`therefore maximization of channel diversity and interference randomization may be crucial to achieve target
`requirements for both latency critical and latency tolerant services. These aspects are discussed in more details in section
`4.1 of this contribution.
` Support of mini-slot repetitions. As it was argued above, at least for grant-free transmission, the repetitions of slots/mini-
`slots of the same TB may not be consecutive in time. For an open-loop operation as is typical for grant-free UL
`transmissions, use of repetitions of a smaller data channel duration provides the opportunity at the gNB for an early
`decoding success event in case of favourable channel conditions. Having this in mind, restricting transmission to single
`mini-slot per slot violates the primary targets of supporting UL grant-free transmissions. However, if gNB decides to
`have a continuous transmission shorter than a slot, it can always configure a proper mini-slot duration without
`repetitions. For example, instead of configuration of two aggregated 2-symbol mini-slots, gNB can always configure
`single 4-symbol mini-slot.
`Proposal 1
` For UL transmission with grant, the repetitions are contiguous in valid slots
` For UL transmission without grant, mini-slot repetitions within a slot are supported
`The number of repetitions itself should be configurable, as it was agreed. For grant-based UL and Type 2 grant-free, it is
`still FFS whether only RRC or RRC + L1 signalling is used. In our view, generalizing the signalling to RRC + L1 is a more
`optimal and “forward looking” approach. It would also be an important differentiating feature of Type 2 comparing to Type
`1 when the number of repetitions could change dynamically based on changing propagation conditions. As for the concrete
`design of the signalling, it is preferred that a set of values for repetition number is configured by RRC and the index of the
`particular value is conveyed in DCI.
`Proposal 2
`
`1
`
`APPLE 1008
`
`

`

` The number of repetitions K for UL transmission with grant and Type 2 UL transmission without grant is indicated by
`DCI grant/activation from a set of values that are configured via RRC
`2.2 Frequency Hopping
`Previously during the NR work/study item the frequency hopping was agreed for DFT-s-OFDM waveform which does not
`support distributed transmission. Further, the following agreements were made regarding frequency hopping:
`Agreements:
` Support PUSCH frequency-hopping for DFT-s-OFDM and CP-OFDM waveform with RA Type 1.
`o At least support intra-slot FH for Msg.3.
` FFS: details including hopping pattern/configurations, signaling designs, etc.
` FFS: whether applicable to all PUSCH durations within a slot
` FFS: whether to support repetition of Msg.3
` Support UE-specific RRC configuration of the following:
`o Mode 1: intra-slot FH only
` FFS: whether applicable to all PUSCH durations within a slot
` Note: Mode 1 is applicable to single slot and repetition case
`o Mode 2: inter-slot only
` Note: Mode 2 is applicable to repetition case
`o FH across mini-slots for repetitions
` FFS: whether it can be enabled by which mode and details, including alignment with slot boundary, pattern
`etc. Target to have a common FH design between slot and mini-slot.
`o FFS: details including the number of configurations, hopping pattern/configurations, signaling designs, etc.
` Support RAR/UL grant indication for PUSCH frequency-hopping
`o FFS: details including how to indicate enable/disable and pattern/mode of FH.
`Agreements:
` The notion of VRB is included in the specifications.
` A non-transparent VRB-to-PRB mapping (i.e. PRB_i=VRB_j where j=f(i)) is supported
`o At least for resource allocation type 1
`o Discuss further whether to support it also for resource allocation type 0
` At least a block-interleaver is used for VRB-to-PRB mapping
`o FFS the details
` A single bit in the DCI indicates localized or distributed VRB-to-PRB mapping.
`
`
`First, the general hopping modes, types, and mechanisms need to be designed and further a reasonable option can be
`selected for Msg3 transmission. As it was agreed, both intra-slot and inter-slot hopping are supported. First the common
`aspects between the two hopping modes are discussed in this section. Then the specific aspects for two different modes are
`considered.
`Note, that VRB-to-PRB mapping is out of scope of present contribution. It is understood that any agreed frequency hopping
`may either be describe by either explicit revaluation of PRB indexes or by applying VRB-to-PRB concept.
`2.2.1 Hopping Patterns
`A baseline assumption for a hopping rule is to provide non-colliding equation for UEs served by the same gNB. LTE
`supported two basic types of hopping: fixed offset (half-band or quarter-band) and pseudo-random pattern-based (cell-
`specific). Both types had non-colliding property for UEs served by the same eNB. However, in NR it is not straightforward
`to reuse LTE hopping types because of the feature of multiple bandwidth parts which are UE-specific.
`Specifically, the 1/2 and 1/4 bandwidth hopping offsets may lead to collisions in case of different overlapped bandwidth
`parts as illustrated in Figure 1.
`
`2
`
`

`

`frequency
`
`Collision of UE1 
`and UE3
`
`time
`
`UE2
`
`UE1
`
`UE3
`
`UE2
`
`UE1 UE3
`
`Collision of UE4 
`and UE5
`
`UE4
`
`1/4 BWP Hopping offsets
`
`UE5
`
`X PRB hopping 
`offset
`
`UE4
`
`UE5
`
`BWP‐1
`
`X PRB hopping 
`offset
`
`BWP‐2
`
`
`Figure 1. Illustration of possible collisions in case of different BWP operation.
`Such potential collisions may be resolved by flexible configuration of hopping offsets both in magnitude and direction. The
`hopping offset may either be calculated as a fraction of hopping bandwidth or as an explicit number of PRBs.
`In LTE, another type of hopping based on pseudo-random equation, sub-band partitioning, and mirroring was specified.
`This option provided good inter-cell interference randomization since the pseudo-random equation is initialized in cell
`specific manner. However, such equation based hopping faces multiple issues in cases of different bandwidth parts,
`configuration of reserved resources, etc. Moreover, with flexible configuration of frequency hopping offsets, sufficient
`inter-cell interference randomization is also enabled. Therefore, it is preferred that single hopping pattern type is adopted in
`NR.
`Proposal 3
` NR supports a single frequency hopping type with explicitly configured hopping offsets
`Currently the maximum number of PRBs in a NR carrier is 275. Therefore, in order to signal any offset within this
`bandwidth 9 bit are needed. It is definitely a large overhead to be placed in DCI. Hence, splitting the configurability
`between semi-static signalling and dynamic DCI is highly desirable.
`A mechanism used in LTE to share resource allocation bits and hopping bits is preferred since it is fair to assume UEs
`requiring frequency hopping are transmitting in a narrow bandwidth. Moreover, the relatively wideband transmission
`already achieves good frequency diversity within the allocated bandwidth.
`The number of hopping bits may also depend on UE frequency hopping bandwidth. However, currently the hopping
`bandwidth may change dynamically with switching of active bandwidth part. Therefore, it seems dependency of RA field
`with regard to BWP size is necessary. Similar to LTE, the number of hopping bits may be adopted as proposed in Table 1:
`Table 1. Relation of maximum hopping bandwidth and the number of hopping bits.
`BW PRB range
`Number of hopping bits
`< X1 (50)
`1
`X1 (50) ≤ and < X2 (150)
`2
`X2 (150) ≤ and ≤ X3 (275)
`2 or 3
`
`
`Note, that for the wide bandwidth operation the number of hopping bits may need to be increased to 3 in order to introduce
`more hopping offsets that are desirable for avoidance of hopping collisions. The hopping field itself encode different
`hopping offsets as in Table 2 below.
`Table 2. Mapping of hopping bits the offsets.
`Number of hopping bits
`Hopping offsets
`1
`A0, A1
`
`3
`
`

`

`A0, A1, A2, A3
`2
`A0, A1, A2, A3, A4, A5, A6, A7
`3
`Values of A0-A7 are configured UE-specifically by RRC. As it was mentioned, it may be either a configured fraction of the
`hopping bandwidth or an explicit PRB offset. Seems the fraction of hopping bandwidth is more a UE-specific attribute in
`NR and it may be hard to align collisions operating only by bandwidth fractions, therefore the explicit PRB offset may be a
`more suitable choice. Each value of the offset may be any in range of [0… hopping bandwidth-1].
`Proposal 4
` Explicit frequency hopping flag is included into DCI format scheduling UL transmission
`
`If the frequency hopping flag is toggled, the following number of hopping bits are taken from the resource allocation
`Type 1 indication field:
`o 1 bit: if hopping bandwidth less than X1 PRB
`o 2 bit: if hopping bandwidth larger or equal than X1 and less than X2
`o 2 or 3 bits: if hopping bandwidth larger or equal than X2 and less or equal than X3
`o The set of {X1, X2, and X3} values is {50, 150, 275}
` The hopping bits encode an index of a hopping offset explicitly configured by UE-specific RRC signalling
`2.2.2
`Intra-slot hopping
`The intra-slot hopping can be realized without repetitions. In LTE, the intra-subframe hopping was realized by hopping
`between the two 7-symbol (6 symbol in case of ECP) slots in the subframe. The following aspects need to be taken into
`account for designing intra-slot hopping:
` Slot partitioning for hopping. Since PUSCH part of the slot may vary significantly, a rule to split the PUSCH for
`hopping should be defined. The similar problem is discussed for long PUCCH intra-slot hopping. Therefore, the design
`for long PUCCH intra-slot hopping partitioning can be reused for PUSCH partitioning.
` DM-RS location. The slot may contain only one DM-RS symbol in the beginning or contain additional DM-RS in later
`part of the slot depending on configuration. The intra-slot hopping should be possible only if there is additional DM-RS
`in the second part of the slot.
`Proposal 5
` For intra-slot frequency hopping
`o Frequency hopping boundary is calculated based on PUSCH duration using the same rule defined for PUCCH with
`N symbol duration
`2.2.3
`Inter-slot and inter-mini-slot hopping
`Similar to LTE, the hopping rule for inter-slot and intra-slot can be common. For inter-slot, the hopping can be organized
`between the configured repetitions. The repetitions could be configured in terms of slots or mini-slots as discussed in
`section 2.1 above.
`For mini-slots, the hopping rate may be restricted to two different positions in order to align collision patterns in case of
`different mini-slot durations and minimize the overhead of UE transient periods. In case of mini-slot aggregation, inter-
`mini-slot frequency hopping can be applied. In particular, in case when N mini-slots is applied for the mini-slot aggregation
`for data transmission, the hopping boundary can be defined as floor(N/2) or ceil(N/2), i.e., hopping occurs around the
`middle of aggregated mini-slots. Further, when inter-mini-slot frequency hopping is enabled, intra-mini-slot frequency
`hopping can be disabled.
`In case of slot aggregation, inter-slot frequency hopping can be employed. In particular, in case when N slots are used for
`slot aggregation for data transmission, the hopping boundary can be defined as floor(N/2) or ceil(N/2), i.e., hopping occurs
`around the middle of aggregated slots. Further, when inter-slot frequency hopping is enabled, intra -slot frequency hopping
`can be disabled.
`Proposal 6
`
`4
`
`

`

`
`
`In case of inter-mini-slot hopping, the number of frequency position changes within a slot is limited to one and the
`hopping boundary is aligned with the case of intra-slot frequency hopping
`2.2.4 Frequency Hopping for Random Access Procedure
`Another outstanding issue is the configuration of frequency hopping and repetitions for Msg3 transmission. The main
`problem with Msg3 is that all related RRC configuration parameters need either be set to a default value or be signalled in
`RMSI. Potentially the following parameters to be discussed:
`- Configuration of hopping mode (intra-slot or inter-slot)
`- Configuration of hopping offsets
`- Configuration of repetitions
`The intra-slot hopping for Msg3 was agreed, and therefore it may be used as a default value for Msg3 transmission. The
`hopping offset for Msg3 should be based on a fraction of the initial BWP since other configurations of hopping bandwidth
`are not available at this point and placing the hopping parameter into RMSI is not justified in terms of overhead. For
`example, the following hopping offsets may be assumed by a UE for Msg3:
`Number of hopping bits
`Default hopping offsets
`1
`reserved, BW/2
`2
`reserved, BW/2, BW/4, -BW/4
`3
`reserved, BW/2, BW/4, -BW/4, BW/8, -BW/8, BW/12, -BW/12
`
`
`On the number of repetitions, as it is also proposed for basic operation, it should be configured by DCI. Therefore, RAR
`grant may directly indicate the number of repetitions to be used for Msg3.
`Proposal 7
` For Msg3 transmission the following parameters are assumed by a UE
`o Intra-slot frequency hopping
`o Default frequency hopping offsets as a 1/2, 1/4, 1/8, 1/12 fraction of initial UL bandwidth part
`3 Grant-based UL Transmission
`The following agreements related to SR-based scheduling were achieved during the previous meeting:
` For each “SR configuration”, the following is indicated via RRC
`o A periodicity and offset which identify the slots/symbols to be used for SR
` FFS the offset for the SR periodicity shorter than one slot for a given SCS
` Non-periodic SR solutions to meet URLLC latency requirements are not precluded
` At least support following as the periodicity of resources for SR
`o FFS other values with taking into account the alignment with 14 symbols
`
`
`
`Subcarrier spacing (kHz)
`15
`30
`60
`120
`
`Supported periodicities [ms]
`2 symbols, 7 symbols, 1, 2, 5, 10, 20, 40, 80
`2 symbols, 7 symbols, 0.5, 1, 2, 5, 10, 20, 40, 80
`2 symbols, 7 symbols (6 symbols for ECP), 0.25, 0.5, 1, 2, 5, 10, 20, 40, 80
`2 symbols, 7 symbols, 0.125, 0.25, 0.5, 1, 2, 5, 10, 20, 40, 80
`
`
`The open issue of additional values for periodicities is discussed first. In our understanding, currently agreed two values of
`sub-slot periodicities may significantly limit UE multiplexing capacity for the case of low latency services. Therefore, more
`sub-slot values are desirable even if there is no alignment within 14 symbol periodicity. Note that for ECP, two additional
`values may be added with no harm: 3 and 4 which fit to 12 symbols.
`Moreover, the alignment within the slot is strictly not required. The periodicity may slightly break when crossing the slot
`boundary in order to achieve the same SR resource position within the slot. In that sense, even finer granularity of SR
`
`
`
`5
`
`

`

`periodicities may be achieved. Therefore, it is proposed to also consider 3 and 4 symbols SR periodicity for the normal CP
`case as well, i.e. 4 or 3 SR occasions per slot.
`In the same time, the offset configuration needs to be jointly considered with PUCCH resource configuration. The following
`is considered:
`Table 3. Relation of SR periodicity and offset configuration.
`SR periodicity type
`Offset configuration
`Slot based SR periodicity SR offset relative to SFN = 0 to determine a slot is configured by the same set of values as
`the SR periodicity.
`SR offset within a slot follows the associated PUCCH resource configuration
`SR offset within a slot follows the associated PUCCH resource configuration
`
`Sub-slot based SR
`periodicity
`
`
`Proposal 8
` SR slot offset relative to SFN = 0 to determine a slot is configured by the same set of values as the SR periodicity
` SR offset within a slot follows the associated PUCCH resource configuration
` The following additional values of SR periodicities and offsets are adopted:
`o 3 and 4 SR occasions per slot
`4 Grant-free UL Transmission
`4.1 Configuration of Resources for Repetitions
`At RAN1 NR AH#2, agreements were made regarding the configuration of periodically occurring resources for two
`identified grant-free transmissions types. Support of one resource per periodic occasion was agreed and the support of
`multiple resources is for further study. The basic timeline of the configuration is shown in Figure 2.
`
`SFN = 0
`
`U1
`
`U1
`
`U1
`
`U1
`
`Offset
`
`Periodicity
`
`
`
`Figure 2. Configuration of single resource.
`Note, that according to current agreements, the resources for repetitions are not configured separately and therefore the
`repetitions should follow the periodic configuration. In that case, many useful scenarios are not supported, for example,
`configuration of periodic bundled transmissions e.g. for VoIP or V2X. In order to fix this, support of multiple resources
`should be agreed.
`During offline discussion at the previous meeting, the following options were identified to interpret “resource”:
` Option 1: One of the K repetitions (K>=1) of a TB is mapped to “a resource” at least consisting of time/frequency-
`domain resource
` Option 2: K repetitions (K>=1) of a TB are mapped to “a resource” at least consisting of time/frequency-domain
`resource
`Note that both options can achieve the same thing with some restrictions and rules defined for performing repetitions. The
`configuration should give the gNB a possibility to avoid detection of initial transmission in a bundle in order to maximize
`UE detection performance with lower gNB efforts. In case if repetitions including initial transmissions can start in non-
`overlapping manner within one configuration, it is possible to utilize all repetitions to detect UEs without multiple
`hypotheses.
`
`6
`
`

`

`SFN = 0
`
`U1
`
`U2
`
`U1
`
`U2
`
`U1
`
`U2
`
`U1
`
`U2
`
`Offset
`
`Periodicity
`
`Figure 3. Configuration of multiple resources.
`A further detail to configuration of multiple resources is whether they are consecutively occurring or with a specific time
`pattern within the configured period. The consecutively occurring repetitions provide the least latency to process the whole
`bundle, however there are the multiple aspects that can be solved by configuration of non-consecutive repetitions:
` Some resources may be unavailable or planned to be used by gNB for other purposes. In that case, the repetitions may
`either be dropped on these resources or explicitly configured to avoid transmission on these resources by postponing.
` Moreover, the time patterns can be introduced with configuration of arbitrary or quasi-arbitrary resource occasions
`similar to what was done for Rel.12 D2D that may be beneficial to randomize collisions and interference both in intra-
`cell and inter-cell (see in Figure 4).
`
`
`
`SFN = 0
`
`Repetition pattern
`
`U1
`
`1
`
`0
`
`U2
`
`1
`
`U3
`
`1
`Periodicity
`
`Offset
`
`U1
`
`U2
`
`U3
`
`U1
`
`U2
`
`U3
`
`U1
`
`U2
`
`U3
`
`
`
`Figure 4. Non-consecutive repetitions.
`In summary, the following parameters should be configured to a UE to support such configuration:
` P – periodicity of occasions to start transmission measured in slots or mini-slots;
` O – offset relative to SFN=0 measured in slots or mini-slots;
` RTP (repetition time pattern) – repetition bitmap pattern of ‘n’ bit (e.g. 8). The pattern is repeated within periodicity P.
`Each ‘1’ represents whether particular resource is available for repetition;
` K – number of repetitions including initial transmission. K may be different from the number of ‘1’ in RPT.
`Another issue is that when multiple repetitions K are configured, when the UE can start transmission should be defined. In
`our view, a one-to-one mapping should be defined between resource and repetition index. I.e. the UE should wait for the
`nearest instance of initial resource to start transmission. One can argue, that this may introduce additional alignment latency.
`However, the latency concern can be resolved by using multiple resource configurations shifted in time like illustrated in
`Figure 5. In that case, the UE may select the nearest resource configuration to start transmission. Another approach is to
`configure the periodicity to a small value that is smaller than overall transmission duration. In that case, the gNB would
`need to detect initial transmission.
`
`SFN = 0
`
`U1
`
`U2
`
`U3
`
`U4
`
`U1
`
`U2
`
`U3
`
`U4
`
`U1
`
`U2
`
`U3
`
`U4
`
`U1
`
`U2
`
`U3
`
`U4
`
`U1
`
`U2
`
`U3
`
`U4
`
`U1
`
`U2
`
`U3
`
`U4
`
`U1
`
`U2
`
`U3
`
`U4
`
`U1
`
`U2
`
`U3
`
`U4
`
`Offset‐1
`Offset‐2
`
`Periodicity‐1
`Periodicity‐2
`
`Figure 5. Multiple configurations to support low latency with repetitions.
`
`
`
`7
`
`

`

`Proposal 9
` For UL transmission without grant
`o Transmission of a TB can only start in occasions configured by the offset and the period (offset and period were
`agreed at RAN1 NR AH#2)
`o Repetitions are mapped with respect to the initial transmission according to a repetition time pattern (RTP)
`represented as a bitmap repeated within the period
`During the emails discussion after the last meeting, a working assumption was made that one of the following three
`sequences are configured:
` {0, 2, 3, 1}
` {0, 3}
` {0}
`
`As it was also discussed in the related email discussion, the motivation for configuration of {0, 3} is not clear. It neither
`performs better than {0, 2, 3, 1} nor provides capability of blind combining of neighbouring TTIs without testing RV
`hypothesis. Therefore, it is proposed that sequence {0, 3} is excluded from the working assumption.
`
`When RV cycling is applied, it should be done with respect to some resource in the resource configuration in order to avoid
`RV blind detection at gNB. It is natural to link the first resource for start of the RV cycling sequence with the first resource
`in K repetitions. In that case, the gNB always has information to assume a particular RV without blind detection.
`Proposal 10
` Confirm the working assumption with an update that sequence {0, 3, 0, 3} is excluded from the possible configurations:
`o For UL transmission without UL grant, for a TB transmission with K repetitions
` The repetitions follow an RV sequence and it is configured by UE-specific RRC signalling to be one of the
`following:
` Sequence 1: {0, 2, 3, 1}
` Sequence 2: {0, 0, 0, 0}
`4.2 HARQ Retransmissions
`Multiple HARQ processes
`According to RAN2 decision, SPS will have one process, i.e. one resource configuration. Therefore, the Type 2 grant-free
`(which is equal to RAN2 SPS) has only one resource configuration for now. Taking this into account, in order to support
`multiple HARQ processes for Type 2, the single configuration should accommodate multiple HARQ processes. Therefore, a
`hybrid approach which accommodates both multiple processes within one configuration and configuration-specific HARQ
`process numbering should be targeted.
`Accordingly, “CURRENT_TTI” component of LTE like equation can be generalized such that it corresponds to a
`transmission opportunity composed of either an individual resource or a set of resources identified by initial transmission of
`a TB that is followed by its repetitions (the initial and K repetitions being referred to as a single transmission opportunity).
`Then, a hierarchical relationship can be defined as follows:
`- Step 1. Identify the set of one or more HARQ process IDs, defined by starting HARQ process index, for a given
`resource configuration following Option 2.
`- Step 2. Using Option 1, the HARQ process IDs for each of the one or more transmission opportunities within a
`resource configuration are identified if and when multiple processes are configured per resource configuration.
`For the above two-step HARQ process ID determination approach, the HARQ processes need to be partitioned semi-
`statically across different resource configurations and the Option 2 equation can be further generalized to accommodate
`resource configurations with different number of HARQ processes.
`One example of such a generalized equation is:
`HARQ Process ID = {[floor(CURRENT_TTI/semiPersistSchedInterval(i))] modulo numberOfConfSPS-Processes(i) +
`harqProcessOffset(i)} modulo totalNumberOfConfSps-Processes;
`
`8
`
`

`

`Where i – index of the resource configuration and multiple parameters such as semiPersistSchedInterval(i),
`numberOfConfSPS-Processes(i), harqProcessOffset(i), are configured per each resource configuration.
`Proposal 11
` The following rule for calculation of HARQ process ID is adopted for UL grant-free transmission
`o HARQ Process ID = {[floor(CURRENT_TTI/semiPersistSchedInterval(i))] modulo numberOfConfSPS-Processes(i)
`+ harqProcessOffset(i)} modulo totalNumberOfConfSps-Processes;
`
`i – index of the resource configuration
`
`semiPersistSchedInterval(i) – period of grant-free resources for a given resource configuration
` harqProcessOffset(i) – HARQ process ID for a given resource configuration
`
`totalNumberOfConfSps-Processes – total number of HARQ processes for grant-free
`
`5 Conclusions
`In this contribution, we discussed design aspects of UL transmission. Based on the analysis, we have the following
`proposals. First, the proposals on common aspects to all transmission types are discussed:
`Proposal 1
` For UL transmission with grant, the repetitions are contiguous in valid slots
` For UL transmission without grant, mini-slot repetitions within a slot are supported
`Proposal 2
` The number of repetitions K for UL transmission with grant and Type 2 UL transmission without grant is indicated by
`DCI grant/activation from a set of values that are configured via RRC
`Proposal 3
` NR supports a single frequency hopping type with explicitly configured hopping offsets
`Proposal 4
` Explicit frequency hopping flag is included into DCI format scheduling UL transmission
`
`If the frequency hopping flag is toggled, the following number of hopping bits are taken from the resource allocation
`Type 1 indication field:
`o 1 bit: if hopping bandwidth less than X1 PRB
`o 2 bit: if hopping bandwidth larger or equal than X1 and less than X2
`o 2 or 3 bits: if hopping bandwidth larger or equal than X2 and less or equal than X3
`o The set of {X1, X2, and X3} values is {50, 150, 275}
` The hopping bits encode an index of a hopping offset explicitly configured by UE-specific RRC signalling
`Proposal 5
` For intra-slot frequency hopping
`o Frequency hopping boundary is calculated based on PUSCH duration using the same rule defined for PUCCH with
`N symbol duration
`Proposal 6
`
`In case of inter-mini-slot hopping, the number of frequency position changes within a slot is limited to one and the
`hopping boundary is aligned with the case of intra-slot frequency hopping
`Proposal 7
` For Msg3 transmission the following parameters are assumed by a UE
`
`9
`
`

`

`o Intra-slot frequency hopping
`o Default frequency hopping offsets as a 1/2, 1/4, 1/8, 1/12 fraction of initial UL bandwidth part
`Proposal 8
` SR slot offset relative to SFN = 0 to determine a slot is configured by the same set of values as the SR periodicity
` SR offset within a slot follows the associated PUCCH resource configuration
` The following additional values of SR periodicities and offsets are adopted:
`o 3 and 4 SR occasions per slot
`Proposal 9
` For UL transmission without grant
`o Transmission of a TB can only start in occasions configured by the offset and the period (offset and period were
`agreed at RAN1 NR AH#2)
`o Repetitions are mapped with respect to the initial transmission according to a repetition time pattern (RTP)
`represented as a bitmap repeated within the period
`Proposal 10
` Confirm the working assumption with an update that sequence {0, 3, 0, 3} is excluded from the possible configurations:
`o For UL transmission without UL grant, for a TB transmission with K repetitions
` The repetitions follow an RV sequence and it is configured by UE-specific RRC signalling to be one of the
`following:
` Sequence 1: {0, 2, 3, 1}
` Sequence 2: {0, 0, 0, 0}
`Proposal 11
` The following rule for calculation of HARQ process ID is adopted for UL grant-free transmission
`o HARQ Process ID = {[floor(CURRENT_TTI/semiPersistSchedInterval(i))] modulo numberOfConfSPS-Processes(i)
`+ harqProcessOffset(i)} modulo totalNumberOfConfSps-Processes;
`
`i – index of the resource configuration
`
`semiPersistSchedInterval(i) – period of grant-free resources for a given resource configuration
` harqProcessOffset(i) – HARQ process ID for a given resource configuration
`
`totalNumberOfConfSps-Processes – total number of HARQ processes for grant-free
`
`
`
`10
`
`