3GPP TSG RAN WG1 #121 R1-2504417
St Julian’s, Malta, May 19th – 23th, 2025
Agenda item: 9.13.1
Source: Qualcomm Incorporated
Title: Time and Frequency Interleaving for 5G Broadcast
Document for: Discussion and Decision
Time Interleaver
Per PMCH configuration of time-interleaving parameters
In the last meeting, the following agreement was made, with down-selection to be performed, with additional necessary conditions to support Option 2, marked in red:
Agreement
The agreement in the previous meeting is updated as follows:
Regarding configuration for time interleaving transmission, the following options can be considered:
- Option 1: Per PMCH configuration, e.g., via pmch-Config
- Option 2: Per MBMS session, e.g., via MBMS-SessionInfo or via MAC CE
- Alt1: If per MBMS session configuration is done, a T_proc of 3 ms between the end of the MSI and the first MTCH in the MCH scheduling period is needed for UE de-rate matching.
- Alt2: If per MBMS session configuration is done, other options (than Alt1 above) are not precluded, e.g. that allow the UE to know the M, N values of the first MTCH scheduled after the MSI
For the sake of simplicity (e.g., to avoid requiring a T_proc), and keeping in mind that all other physical layer parameters are specified “per PMCH”, we propose to configure time-interleaving parameters “per PMCH”. This way, all the MBMS sessions within the PMCH will have the same values of time interleaving parameters .
This will also simplify the design of the “new periodicities” for the MSI, to reduce resource wastage.
Proposal 1: Regarding configuration for time interleaving transmission, the following Option is agreed:
Option 1: Per PMCH configuration, via pmch-Config
Observation 1: Per PMCH configuration of time-interleaving parameters simplifies the design of new MSI periodicities, which are required to minimize resource wastage on account of mismatch between -length units and the legacy MSI periodicities
This is because all MBMS sessions within a MCH scheduling period would have the same value of
New MSI periodicities to minimize resource wastage
In RAN1#120bis, the following agreement was made, regarding introducing new MSI periodicities to minimize resource wastage:
Agreement
In case of time interleaving, in order to reduce the resource wastage due to the potential mismatch between MSI periodicity and the (MxN) pattern:
Adopt an extended set of configurable periodicities for MCH scheduling period for R19 PMCHs that includes the intermediate values among the periodicities defined for the legacy configurations.
May include the case of scheduling different M/N within the same MSI periodicity
FFS the values of the new periodicities.
RAN1 to specify UE behavior, if needed, in case of residual mismatch even after new periodicities are adopted, taking into account the amount of subframe available at the end of the periodicity
Option 1: Drop the time-interleaved TBs which do not have all the N RVs transmitted within the MSI periodicity
Option 2: Drop the time-interleaved TBs which could not be successfully decoded.
It is up to network to decide whether to schedule or not schedule MTCH transmission in the residual mismatch part of the MSI periodicity. UE knows this from MSI (legacy behaviour)
Other Options not precluded
We note that the supported values for are {4,8,16,32} and those for are {1,2,4,8,16}. Further, would imply that a corresponding value of doesn’t exist—i.e., there is no time interleaving. Consequently, the set of (unique) values of are , which will serve as a guide to determine the new MSI periodicities that best accommodate (multiples of) these -length units within MSI periods, while minimizing resource wastage. As stated in the previous subsection, assuming a “per PMCH” configuration, there will only be a single value of within a MCH scheduling period.
Observation 2: For the supported values of when time interleaving is enabled, the set of unique value of the basic time-interleaving unit for scheduling, , are
With a per PMCH configuration of , each MCH scheduling period will comprise a single value of
In Table 1 below, we determine—for each value of —the resource wastage that can occur with different legacy MSI periodicities (considering the overhead from the CAS, the MSI subframe itself), and wherever necessary, propose new periodicities that would reduce this resource wastage.
Table 1: Determination of new MSI periodicities
Proposal 2: To reduce resource wastage, specify the following set of new MSI periodicities: {rf7, rf14, rf27, rf53, rf106, rf211}.
and reading bits from the circular buffer
With regards to the value of , the following agreement was made in RAN1#120.
Agreement
The equation of k0 for time interleaving is modified to avoid puncturing of systematic bits.
Further study the exact equation for k0 for the n-th subframe of a transmission belonging to a same TB , ), including
Option 1: for each CB of the TB, , where (as per TS 36.212), and is the redundancy version of the n-th subframe.
Option 2: For each CB of the TB, follow the principle of NR TBoMS to determine
Option 1 and Option 2 assumes all CBs of the TB have the same RV index for all CBs of the TB, indexed by
FFS whether different CBs of the TB mapped to each subframe may have different RV indices (e.g., indexed by different per CB)
We note that Option 2 in the agreement above would require maintaining multiple values of (corresponding to the CBs of the TB) for circular buffer management. This is difficult in terms of implementation and maintaining a single pointer across the entire TB is strongly preferred. There is expected to be negligible performance difference, it at all, between the two approaches. To this end, we make the following proposal.
Proposal 3: For time-interleaved PMCH, specify the starting pointer for each CB of the TB as , where and is the redundancy version of the n-th subframe.
and are defined as follows:
For ,
For
For
LBRM for time-interleaved PMCH
In RAN1#120bis, the following agreement was made, regarding LBRM for time-interleaved PMCH.
Agreement
LBRM parameters (e.g. ) are consistent between the broadcast network and the (multiple) UEs that are receiving the broadcast transmission.
These multiple UEs may belong to different UE categories, thus having different values of , .
FFS: how the consistent NIR and NCB is calculated
The agreement lends itself to the following primary observation.
Observation 3: To maintain consistency between the broadcast transmission and the (multiple, potentially different category) UEs, the LBRM parameter(s) must be configured to the UE via control signaling, e.g., via PMCH-Config for each PMCH.
On top of the fact that the LBRM parameters for time-interleaved PMCH have to be signaled (as opposed to implicitly determined based on UE category), the following additional observation needs to be kept in mind, regarding differences from the legacy (unicast) equation for determining the value of , which essentially defines when and how LBRM kicks in.
Observation 4: To determine the soft buffer size for a time-interleaved PMCH transport block in each subframe (i.e., the value , the following differences with respect to unicast need to be noted
The term is replaced by , where denotes the number of interleaved TBs
The above observation leaves us with two degrees of freedom, with respect to the network configuration of the LBRM parameters, to determine —the terms (which denotes how many component carriers (CCs) are used for time-interleaved PMCH) and (which denotes the total number of soft channel bits corresponding to different UE categories). To provide the broadcast network maximum flexibility in selecting the values of and , we propose to signal these fields in control signaling, from among all (practical) candidate values for these parameters, that are supported in the specifications.
Observation 5: Signaling the values of and from an exhaustive set of candidate values supported in the current specifications for DL-SCH, provides the broadcast network maximum flexibility in configuring the LBRM parameters.
Considering the above observations, we make the following proposal.
Proposal 4: For PMCH with time-interleaving, a UE shall determine the value of for LBRM according to the equation below:
is configured in PMCH-Config from:
{1, 3/2, 6/5, 8/5, 2, 12/5, 8/3, 3, 5, 32}
is configured in PMCH-Config from
{1827072, 3654144, 5481216, 7308288, 9744384, 12789504, 14616576, 17052672, 19488768, 24360960, 29233152, 34105344, 36541440, 38977536, 42631680, 47431680, 303562752}
M is the number of TBs for time-interleaving
Codeblock cycling across subframes of a TB
As pointed out in [1, 2], the current design of the time and frequency interleavers result in a localization in frequency (across the subframes of a time-interleaved TB) for each codeblock of the TB. This is because the eventual mapping of coded bits (from the output of codeblock concatenation (Section 5.1.5 of TS 36.212), followed by bit-level scrambling (Section 6.3.1 of TS 36.211)) to resource elements, are identical across the subframes of the TB.
Observation 6: For PMCH with time-frequency interleaving, the mapping of the coded bitstream of a TB to physical resources is identical across the subframes to which a time-interleaved TB is mapped
This leads to localization in frequency for each constituent codeblock (within the coded bitstream), across the subframes, potentially affecting BLER-vs-SNR performance adversely.
To mitigate the potential performance loss, codeblock cycling—i.e., using a different ordering (by means of cyclic shifts) of codeblocks, while mapping the bits to each of the subframes—was proposed and evaluated in [1], demonstrating meaningful performance gains over even a TDL-A channel. The approach in [1] employed cyclically shifting the codeblocks by a unit of a single codeblock across successive subframes—i.e., subframe (RV) 0 of the TB would have a cyclic shift (in units of bits equal to the size of a codeblock) of 0 units; subframe (RV) 1 of the TB would have a cyclic shift of 1 unit; subframe (RV) 2 of the TB would have a cyclic shift of 2 units, and so on.
Observation 7: In [1], the mapping of the coded bitstream to physical resources employed a successive cyclic shift of one codeblock worth of bits, when mapping to the successive subframes of the time-interleaved TB.
While the cyclic shift units applied above depend on the subframe index, for general values of and , a successive cyclic shift of one unit across the subframes is strictly suboptimal. This can be seen from the fact that for a TBS with a large (e.g., much greater than ) , employing successive cyclic shifts of one CB-unit across successive subframes of the TB results in each CB being localized to a fraction of the bandwidth.
A straightforward modification to the above approach for general values of and may be used to remedy the above issue. Cyclically shifting the codeblocks in successive subframes by a quantity ensures that each CB spans (approximately) the entire bandwidth in frequency.
Observation 8: For general values of and , for mapping of bits to physical resources, successive cyclic shifts of codeblocks’ worth of bits should be employed, when mapping to the successive subframes of the time-interleaved TB
This maximizes the frequency-domain span of each codeblock, across the subframes.
The performance benefits that CB cycling across subframes can provide, are demonstrated in Fig. 1, with corresponding simulation parameters listed in Table 2.
Table 2: Simulation parameters to evaluate codeblock cyclic across subframes of time-interleaved TBs
Figure 1: Performance of codeblock cycling across subframes of a TB
Observation 9: Cyclically shifting the coded bit sequence for a TB at the subframe , by CBs’ worth of bits provides up to a gain, versus the current time-frequency interleaver design
Using a shift of CBs’ worth of bits for the subframe (as in [1]) incurs up to a penalty relative to the above design
Based on the above observation, we make the following proposal, regarding codeblock cycling across the subframes of a TB (with an illustrative depiction of cyclically shifting the coded bit sequence in Fig. 2).
Figure 2: Illustration of cyclically shifting a coded bit sequence by bits.
Proposal 5: Cyclically shift the bit sequence in Section 6.3.1 of TS 36.211 for the subframe of the time-interleaved TB by bits, where
,
denotes the number of subframes to which the time-interleaved TB is mapped
denotes the number of CBs in the time-interleaved (scaled) TB
Frequency Interleaver
Number of columns
In RAN1#120bis, the following agreement was made, regarding the number of columns of the frequency interleaver.
Agreement
The number of columns, , of the frequency interleaver is , where denotes the number of CBs in the TB (including scaling, if any), denotes the number of OFDM symbols of the PMCH,
Working assumption:
m = , and denotes the operation to compute the greatest common divisor.
As described previously, the value of m = in the above equation minimizes the “column span” of the codeblocks (i.e., the number of columns spanned by the codeblocks, during the column by column writing to the frequency interleaving matrix). Any increase beyond this number would lead to a degradation in frequency diversity, when the elements of the matrix are read out row by row. As a result, we propose to confirm the above working assumption on the value of m.
Proposal 6: Confirm the working assumption on the value of m, in the following agreement from RAN1#120bis
Agreement
The number of columns, , of the frequency interleaver is , where denotes the number of CBs in the TB (including scaling, if any), denotes the number of OFDM symbols of the PMCH,
Working assumption:
m = , and denotes the operation to compute the greatest common divisor.
Conclusion
In this contribution we presented our views on Time-Frequency Interleaving for 5G Broadcast. Our proposals and observations are summarized below.
Proposal 1: Regarding configuration for time interleaving transmission, the following Option is agreed:
Option 1: Per PMCH configuration, via pmch-Config
Observation 1: Per PMCH configuration of time-interleaving parameters simplifies the design of new MSI periodicities, which are required to minimize resource wastage on account of mismatch between -length units and the legacy MSI periodicities
This is because all MBMS sessions within a MCH scheduling period would have the same value of
Observation 2: For the supported values of when time interleaving is enabled, the set of unique value of the basic time-interleaving unit for scheduling, , are
With a per PMCH configuration of , each MCH scheduling period will comprise a single value of
Proposal 2: To reduce resource wastage, specify the following set of new MSI periodicities: {rf7, rf14, rf27, rf53, rf106, rf211}.
Proposal 3: For time-interleaved PMCH, specify the starting pointer for each CB of the TB as , where and is the redundancy version of the n-th subframe.
and are defined as follows:
For ,
For
For
Observation 3: To maintain consistency between the broadcast transmission and the (multiple, potentially different category) UEs, the LBRM parameter(s) must be configured to the UE via control signaling, e.g., via PMCH-Config for each PMCH.
Observation 4: To determine the soft buffer size for a time-interleaved PMCH transport block in each subframe (i.e., the value , the following differences with respect to unicast need to be noted
The term is replaced by , where denotes the number of interleaved TBs
Observation 5: Signaling the values of and from an exhaustive set of candidate values supported in the current specifications for DL-SCH, provides the broadcast network maximum flexibility in configuring the LBRM parameters.
Proposal 4: For PMCH with time-interleaving, a UE shall determine the value of for LBRM according to the equation below:
is configured in PMCH-Config from:
{1, 3/2, 6/5, 8/5, 2, 12/5, 8/3, 3, 5, 32}
is configured in PMCH-Config from
{1827072, 3654144, 5481216, 7308288, 9744384, 12789504, 14616576, 17052672, 19488768, 24360960, 29233152, 34105344, 36541440, 38977536, 42631680, 47431680, 303562752}
M is the number of TBs for time-interleaving
Observation 6: For PMCH with time-frequency interleaving, the mapping of the coded bitstream of a TB to physical resources is identical
Observation 7: In [1], the mapping of the coded bitstream to physical resources employed a successive cyclic shift of one codeblock worth of bits, when mapping to the successive subframes of the time-interleaved TB.
Observation 8: For general values of and , for mapping of bits to physical resources, successive cyclic shifts of codeblocks’ worth of bits should be employed, when mapping to the successive subframes of the time-interleaved TB
This maximizes the frequency-domain span of each codeblock, across the subframes.
Observation 9: Cyclically shifting the coded bit sequence for a TB at the subframe , by CBs’ worth of bits provides up to a gain, versus the current time-frequency interleaver design
Using a shift of CBs’ worth of bits for the subframe (as in [1]) incurs up to a penalty relative to the above design
Proposal 5: Cyclically shift the bit sequence in Section 6.3.1 of TS 36.211 for the subframe of the time-interleaved TB by bits, where
,
denotes the number of subframes to which the time-interleaved TB is mapped
denotes the number of CBs in the time-interleaved (scaled) TB
Proposal 6: Confirm the working assumption on the value of m, in the following agreement from RAN1#120bis
Agreement
The number of columns, , of the frequency interleaver is , where denotes the number of CBs in the TB (including scaling, if any), denotes the number of OFDM symbols of the PMCH,
Working assumption:
m = , and denotes the operation to compute the greatest common divisor.
References
[1] R1-2502582, “Design of Time-frequency Interleaver for 5G Broadcast”, by Shanghai Jiao Tong University, ABS, NERC-DTV, PCL, CUC, in RAN1#120bis
[2] R1-2500872, “Time-Frequency interleaver design for LTE-based 5G broadcast”, by Samsung, in RAN1#120 |
3GPP TSG RAN WG1 #121 R1-2504654
St Julian’s, Malta, May 19th – 23th, 2025
Agenda item: 9.13.1
Source: Qualcomm Incorporated
Title: Time and Frequency Interleaving for 5G Broadcast
Document for: Discussion and Decision
Time Interleaver
Per PMCH configuration of time-interleaving parameters
In the last meeting, the following agreement was made, with down-selection to be performed, with additional necessary conditions to support Option 2, marked in red:
Agreement
The agreement in the previous meeting is updated as follows:
Regarding configuration for time interleaving transmission, the following options can be considered:
- Option 1: Per PMCH configuration, e.g., via pmch-Config
- Option 2: Per MBMS session, e.g., via MBMS-SessionInfo or via MAC CE
- Alt1: If per MBMS session configuration is done, a T_proc of 3 ms between the end of the MSI and the first MTCH in the MCH scheduling period is needed for UE de-rate matching.
- Alt2: If per MBMS session configuration is done, other options (than Alt1 above) are not precluded, e.g. that allow the UE to know the M, N values of the first MTCH scheduled after the MSI
For the sake of simplicity (e.g., to avoid requiring a T_proc), and keeping in mind that all other physical layer parameters are specified “per PMCH”, we propose to configure time-interleaving parameters “per PMCH”. This way, all the MBMS sessions within the PMCH will have the same values of time interleaving parameters .
This will also simplify the design of the “new periodicities” for the MSI, to reduce resource wastage.
Proposal 1: Regarding configuration for time interleaving transmission, the following Option is agreed:
Option 1: Per PMCH configuration, via pmch-Config
Observation 1: Per PMCH configuration of time-interleaving parameters simplifies the design of new MSI periodicities, which are required to minimize resource wastage on account of mismatch between -length units and the legacy MSI periodicities
This is because all MBMS sessions within a MCH scheduling period would have the same value of
New MSI periodicities to minimize resource wastage
In RAN1#120bis, the following agreement was made, regarding introducing new MSI periodicities to minimize resource wastage:
Agreement
In case of time interleaving, in order to reduce the resource wastage due to the potential mismatch between MSI periodicity and the (MxN) pattern:
Adopt an extended set of configurable periodicities for MCH scheduling period for R19 PMCHs that includes the intermediate values among the periodicities defined for the legacy configurations.
May include the case of scheduling different M/N within the same MSI periodicity
FFS the values of the new periodicities.
RAN1 to specify UE behavior, if needed, in case of residual mismatch even after new periodicities are adopted, taking into account the amount of subframe available at the end of the periodicity
Option 1: Drop the time-interleaved TBs which do not have all the N RVs transmitted within the MSI periodicity
Option 2: Drop the time-interleaved TBs which could not be successfully decoded.
It is up to network to decide whether to schedule or not schedule MTCH transmission in the residual mismatch part of the MSI periodicity. UE knows this from MSI (legacy behaviour)
Other Options not precluded
We note that the supported values for are {4,8,16,32} and those for are {1,2,4,8,16}. Further, would imply that a corresponding value of doesn’t exist—i.e., there is no time interleaving. Consequently, the set of (unique) values of are , which will serve as a guide to determine the new MSI periodicities that best accommodate (multiples of) these -length units within MSI periods, while minimizing resource wastage. As stated in the previous subsection, assuming a “per PMCH” configuration, there will only be a single value of within a MCH scheduling period.
Observation 2: For the supported values of when time interleaving is enabled, the set of unique value of the basic time-interleaving unit for scheduling, , are
With a per PMCH configuration of , each MCH scheduling period will comprise a single value of
In Table 1 below, we determine—for each value of —the resource wastage that can occur with different legacy MSI periodicities (considering the overhead from the CAS, the MSI subframe itself), and wherever necessary, propose new periodicities that would reduce this resource wastage.
Table 1: Determination of new MSI periodicities
Proposal 2: To reduce resource wastage, specify the following set of new MSI periodicities: {rf7, rf14, rf27, rf53, rf106, rf211}.
and reading bits from the circular buffer
With regards to the value of , the following agreement was made in RAN1#120.
Agreement
The equation of k0 for time interleaving is modified to avoid puncturing of systematic bits.
Further study the exact equation for k0 for the n-th subframe of a transmission belonging to a same TB , ), including
Option 1: for each CB of the TB, , where (as per TS 36.212), and is the redundancy version of the n-th subframe.
Option 2: For each CB of the TB, follow the principle of NR TBoMS to determine
Option 1 and Option 2 assumes all CBs of the TB have the same RV index for all CBs of the TB, indexed by
FFS whether different CBs of the TB mapped to each subframe may have different RV indices (e.g., indexed by different per CB)
We note that Option 2 in the agreement above would require maintaining multiple values of (corresponding to the CBs of the TB) for circular buffer management. This is difficult in terms of implementation and maintaining a single pointer across the entire TB is strongly preferred. There is expected to be negligible performance difference, it at all, between the two approaches. To this end, we make the following proposal.
Proposal 3: For time-interleaved PMCH, specify the starting pointer for each CB of the TB as , where and is the redundancy version of the n-th subframe.
and are defined as follows:
For ,
For
For
LBRM for time-interleaved PMCH
In RAN1#120bis, the following agreement was made, regarding LBRM for time-interleaved PMCH.
Agreement
LBRM parameters (e.g. ) are consistent between the broadcast network and the (multiple) UEs that are receiving the broadcast transmission.
These multiple UEs may belong to different UE categories, thus having different values of , .
FFS: how the consistent NIR and NCB is calculated
The agreement lends itself to the following primary observation.
Observation 3: To maintain consistency between the broadcast transmission and the (multiple, potentially different category) UEs, the LBRM parameter(s) must be configured to the UE via control signaling, e.g., via PMCH-Config for each PMCH.
On top of the fact that the LBRM parameters for time-interleaved PMCH have to be signaled (as opposed to implicitly determined based on UE category), the following additional observation needs to be kept in mind, regarding differences from the legacy (unicast) equation for determining the value of , which essentially defines when and how LBRM kicks in.
Observation 4: To determine the soft buffer size for a time-interleaved PMCH transport block in each subframe (i.e., the value , the following differences with respect to unicast need to be noted
The term is replaced by , where denotes the number of interleaved TBs
The above observation leaves us with two degrees of freedom, with respect to the network configuration of the LBRM parameters, to determine —the terms (which denotes how many component carriers (CCs) are used for time-interleaved PMCH) and (which denotes the total number of soft channel bits corresponding to different UE categories). To provide the broadcast network maximum flexibility in selecting the values of and , we propose to signal these fields in control signaling, from among all (practical) candidate values for these parameters, that are supported in the specifications.
Observation 5: Signaling the values of and from an exhaustive set of candidate values supported in the current specifications for DL-SCH, provides the broadcast network maximum flexibility in configuring the LBRM parameters.
Considering the above observations, we make the following proposal.
Proposal 4: For PMCH with time-interleaving, a UE shall determine the value of for LBRM according to the equation below:
is configured in PMCH-Config from:
{1, 3/2, 6/5, 8/5, 2, 12/5, 8/3, 3, 5, 32}
is configured in PMCH-Config from
{1827072, 3654144, 5481216, 7308288, 9744384, 12789504, 14616576, 17052672, 19488768, 24360960, 29233152, 34105344, 36541440, 38977536, 42631680, 47431680, 303562752}
M is the number of TBs for time-interleaving
Codeblock cycling across subframes of a TB
As pointed out in [1, 2], the current design of the time and frequency interleavers result in a localization in frequency (across the subframes of a time-interleaved TB) for each codeblock of the TB. This is because the eventual mapping of coded bits (from the output of codeblock concatenation (Section 5.1.5 of TS 36.212), followed by bit-level scrambling (Section 6.3.1 of TS 36.211)) to resource elements, are identical across the subframes of the TB.
Observation 6: For PMCH with time-frequency interleaving, the mapping of the coded bitstream of a TB to physical resources is identical across the subframes to which a time-interleaved TB is mapped
This leads to localization in frequency for each constituent codeblock (within the coded bitstream), across the subframes, potentially affecting BLER-vs-SNR performance adversely.
To mitigate the potential performance loss, codeblock cycling—i.e., using a different ordering (by means of cyclic shifts) of codeblocks, while mapping the bits to each of the subframes—was proposed and evaluated in [1], demonstrating meaningful performance gains over even a TDL-A channel. The approach in [1] employed cyclically shifting the codeblocks by a unit of a single codeblock across successive subframes—i.e., subframe (RV) 0 of the TB would have a cyclic shift (in units of bits equal to the size of a codeblock) of 0 units; subframe (RV) 1 of the TB would have a cyclic shift of 1 unit; subframe (RV) 2 of the TB would have a cyclic shift of 2 units, and so on.
Observation 7: In [1], the mapping of the coded bitstream to physical resources employed a successive cyclic shift of one codeblock worth of bits, when mapping to the successive subframes of the time-interleaved TB.
While the cyclic shift units applied above depend on the subframe index, for general values of , and (where denotes the number of OFDM symbols in a subframe), a successive cyclic shift of one unit across the subframes is strictly suboptimal. This can be seen from the fact that for a TBS with a large (e.g., much greater than ) , employing successive cyclic shifts of one CB-unit across successive subframes of the TB results in each CB being localized to a fraction of the bandwidth (recall that when , mapping to physical resources happens in a frequency-first time-second manner).
A straightforward modification to the above approach for general values of and may be used to remedy the above issue. Cyclically shifting the codeblocks in successive subframes by a quantity CBs ensures that each CB spans (approximately) the entire bandwidth in frequency.
However, since the frequency interleaver—which will take as input, the modulated symbols from the cyclically shifted codeblock sequence—works on the principle of “codeblock alignment with column boundaries”, “fractional shifts” of the codeblock sequence will adversely affect the performance of the frequency interleaver. As a result, we need to apply the cyclic shift in “integer multiples of codeblocks”, which necessitates a floor ( or ceiling () operation on the quantity described above.
In this case, using the ceiling operation is the right choice, due to the following aspect: a floor operation will, in general, not ensure that each CB cycles through the entire bandwidth, across the subframes. For certain values of , this can result in, e.g., no CB cycling at all, or a limited harnessing of the available cycling diversity in frequency. The ceiling operation mitigates this, by ensuring that each CB cycles through the entire bandwidth (with some minor wraparound, which doesn’t adversely affect performance), all while respecting the principles of the (subsequent) frequency interleaver design, which mandates codeblock alignment with the column boundaries.
With this in mind, we make the following observation.
Observation 8: For general values of and , for mapping of bits to physical resources, successive cyclic shifts of codeblocks’ worth of bits should be employed, when mapping to the successive subframes of the time-interleaved TB
The ceiling (as opposed to floor) operation ensures that each CB cycles through the entire bandwidth, over the subframes
This ensures that “fractional shifts” are not employed, which would adversely affect the performance of the subsequent frequency interleaver, which works on the principle of codeblock alignment with the column boundaries.
The performance benefits that CB cycling across subframes can provide, are demonstrated in Fig. 1, with corresponding simulation parameters listed in Table 2.
Table 2: Simulation parameters to evaluate codeblock cyclic across subframes of time-interleaved TBs
Figure 1: Performance of codeblock cycling across subframes of a TB
Observation 9: Cyclically shifting the coded bit sequence for a TB at the subframe , by CBs’ worth of bits provides up to a gain, versus the current time-frequency interleaver design
Using a shift of CBs’ worth of bits for the subframe (as in [1]) incurs up to a penalty relative to the above design
Based on the above observation, we make the following proposal, regarding codeblock cycling across the subframes of a TB (with an illustrative depiction of cyclically shifting the coded bit sequence in Fig. 2).
Figure 2: Illustration of cyclically shifting a coded bit sequence by bits.
Proposal 5: Cyclically shift the bit sequence in Section 6.3.1 of TS 36.211 for the subframe of the time-interleaved TB by bits, where
,
denotes the number of subframes to which the time-interleaved TB is mapped
denotes the number of OFDM symbols in a subframe
denotes the number of CBs in the time-interleaved (scaled) TB
Frequency Interleaver
Number of columns
In RAN1#120bis, the following agreement was made, regarding the number of columns of the frequency interleaver.
Agreement
The number of columns, , of the frequency interleaver is , where denotes the number of CBs in the TB (including scaling, if any), denotes the number of OFDM symbols of the PMCH,
Working assumption:
m = , and denotes the operation to compute the greatest common divisor.
As described previously, the value of m = in the above equation minimizes the “column span” of the codeblocks (i.e., the number of columns spanned by the codeblocks, during the column by column writing to the frequency interleaving matrix). Any increase beyond this number would lead to a degradation in frequency diversity, when the elements of the matrix are read out row by row. As a result, we propose to confirm the above working assumption on the value of m.
Proposal 6: Confirm the working assumption on the value of m, in the following agreement from RAN1#120bis
Agreement
The number of columns, , of the frequency interleaver is , where denotes the number of CBs in the TB (including scaling, if any), denotes the number of OFDM symbols of the PMCH,
Working assumption:
m = , and denotes the operation to compute the greatest common divisor.
Conclusion
In this contribution we presented our views on Time-Frequency Interleaving for 5G Broadcast. Our proposals and observations are summarized below.
Proposal 1: Regarding configuration for time interleaving transmission, the following Option is agreed:
Option 1: Per PMCH configuration, via pmch-Config
Observation 1: Per PMCH configuration of time-interleaving parameters simplifies the design of new MSI periodicities, which are required to minimize resource wastage on account of mismatch between -length units and the legacy MSI periodicities
This is because all MBMS sessions within a MCH scheduling period would have the same value of
Observation 2: For the supported values of when time interleaving is enabled, the set of unique value of the basic time-interleaving unit for scheduling, , are
With a per PMCH configuration of , each MCH scheduling period will comprise a single value of
Proposal 2: To reduce resource wastage, specify the following set of new MSI periodicities: {rf7, rf14, rf27, rf53, rf106, rf211}.
Proposal 3: For time-interleaved PMCH, specify the starting pointer for each CB of the TB as , where and is the redundancy version of the n-th subframe.
and are defined as follows:
For ,
For
For
Observation 3: To maintain consistency between the broadcast transmission and the (multiple, potentially different category) UEs, the LBRM parameter(s) must be configured to the UE via control signaling, e.g., via PMCH-Config for each PMCH.
Observation 4: To determine the soft buffer size for a time-interleaved PMCH transport block in each subframe (i.e., the value , the following differences with respect to unicast need to be noted
The term is replaced by , where denotes the number of interleaved TBs
Observation 5: Signaling the values of and from an exhaustive set of candidate values supported in the current specifications for DL-SCH, provides the broadcast network maximum flexibility in configuring the LBRM parameters.
Proposal 4: For PMCH with time-interleaving, a UE shall determine the value of for LBRM according to the equation below:
is configured in PMCH-Config from:
{1, 3/2, 6/5, 8/5, 2, 12/5, 8/3, 3, 5, 32}
is configured in PMCH-Config from
{1827072, 3654144, 5481216, 7308288, 9744384, 12789504, 14616576, 17052672, 19488768, 24360960, 29233152, 34105344, 36541440, 38977536, 42631680, 47431680, 303562752}
M is the number of TBs for time-interleaving
Observation 6: For PMCH with time-frequency interleaving, the mapping of the coded bitstream of a TB to physical resources is identical
Observation 7: In [1], the mapping of the coded bitstream to physical resources employed a successive cyclic shift of one codeblock worth of bits, when mapping to the successive subframes of the time-interleaved TB.
Observation 8: For general values of and , for mapping of bits to physical resources, successive cyclic shifts of codeblocks’ worth of bits should be employed, when mapping to the successive subframes of the time-interleaved TB
The ceiling (as opposed to floor) operation ensures that each CB cycles through the entire bandwidth, over the subframes
This ensures that “fractional shifts” are not employed, which would adversely affect the performance of the subsequent frequency interleaver, which works on the principle of codeblock alignment with the column boundaries.
Observation 9: Cyclically shifting the coded bit sequence for a TB at the subframe , by CBs’ worth of bits provides up to a gain, versus the current time-frequency interleaver design
Using a shift of CBs’ worth of bits for the subframe (as in [1]) incurs up to a penalty relative to the above design
Proposal 5: Cyclically shift the bit sequence in Section 6.3.1 of TS 36.211 for the subframe of the time-interleaved TB by bits, where
,
denotes the number of subframes to which the time-interleaved TB is mapped
denotes the number of OFDM symbols in a subframe
denotes the number of CBs in the time-interleaved (scaled) TB
Proposal 6: Confirm the working assumption on the value of m, in the following agreement from RAN1#120bis
Agreement
The number of columns, , of the frequency interleaver is , where denotes the number of CBs in the TB (including scaling, if any), denotes the number of OFDM symbols of the PMCH,
Working assumption:
m = , and denotes the operation to compute the greatest common divisor.
References
[1] R1-2502582, “Design of Time-frequency Interleaver for 5G Broadcast”, by Shanghai Jiao Tong University, ABS, NERC-DTV, PCL, CUC, in RAN1#120bis
[2] R1-2500872, “Time-Frequency interleaver design for LTE-based 5G broadcast”, by Samsung, in RAN1#120 |