3GPP TSG RAN1#56
`
`
`
`
`
`
`
`
`
` R1-090792
`
`Athens, Greece
`
`February 9-13, 2009
`
`Agenda Item:
`Source:
`Title:
`Document for:
`
`12.1
`Motorola
`Control Signalling Design for Supporting Carrier Aggregation
`Discussion
`
`1. Introduction
`With the agreement on MAC-to-PHY mapping reached in the last meeting, i.e., “there is one transport
`block (in absence of spatial multiplexing) and one HARQ entity per scheduled component carrier (from the
`UE perspective)”, the question of how to design the control signalling mechanism to support the downlink
`and uplink transmission (including HARQ retransmission) of such multiple TBs naturally arises. In this
`document, we discuss various DL control signalling design options to support bandwidth extension for LTE
`Advanced.
`
`2. PDCCH
`PDCCH carries the information of DL or UL resource allocation which is now for multiple TBs with each
`being mapped to a component carrier. One question of first importance is the corresponding relationship
`between PDCCH(s) and TB(s), for example, whether a PDCCH message can contain resource assignment
`of multiple component carriers.
`In addition to the issue of PDCCH content, the transmission structure itself (i.e., how resources are
`organized for the transmission of control channels) is also important and its design affects the deployment
`flexibility, system robustness, and UE processing requirements.
`
`2.1. PDCCH Structure
`Given that each TB is confined to a single component carrier, it makes sense to agree on that each PDCCH
`transmission is also confined to a single component carrier due to the following observations:
`- Any benefit of additional frequency diversity gain by transmitting PDCCH across multiple carriers for
`improving the reliability of PDCCH could be minimal.
`- When PDCCH is sent across multiple component carriers, the mechanisms that attempt to reduce UE
`power consumption by allowing the UE to monitor only a single component carrier, especially when
`resources of other carrier do not have to be used, cannot be efficiently supported.
`- Any new PDCCH structure that enables CCE allocation/aggregation over multiple component carriers
`can be quite restricted considering that the new allocation rules for that structure must not affect the
`existing Rel8 CCE allocation mechanisms for Rel8 UEs, especially if a component carrier needs to
`support Rel8 UEs.
`
`Samsung Ex. 1007
`Page 1 of 8
`
`

`

`While confining a PDCCH to a single component carrier is more consistent with the design of confining
`one TB to one component carrier, we believe that, allowing one PDCCH transmitted from one DL
`component carrier to schedule resources on other DL component carriers (or other UL component carriers
`that are not the default pairing with that DL component carrier) can have the following desirable benefits.
`In a system where both Rel8 compatible and Rel8 non-compatible carriers are configured, the
`1.
`PDCCH in Rel8 carriers can be utilised to grant resources in non Rel8 compatible carriers
`(especially in a design/configuration where the non Rel8 compatible carrier carries no PDCCH or
`even no control signalling at all)
`2. The scheduler has the flexibility of assigning resources for multiple component carriers from the
`best or designated component carrier (e.g., the component carrier with strongest CQI or the
`“anchor” carrier). This flexibility helps to improve control signalling efficiency and robustness of
`system operations.
`In the case of asymmetric aggregation such as two UL component carriers and one DL carriers,
`PDCCH will need to grant resources for 2 UL TBs.
`We note that all these benefits may not be realisable in some particular deployment scenarios such as
`scenarios where schedulers for different component carriers cannot operate in a co-ordinated manner.
`Considering this, our view on resource assignment methodology is
`In addition to having an option that uses one PDCCH to schedule resources in one component
`carrier, which is the simplest extension of Rel8 operation, the option of allowing a PDCCH
`transmitted from one DL component carrier to schedule resources in other DL component carriers
`should also be considered.
`
`3.
`
`2.2. PDCCH Content
`Many contributions (e.g.,[1][2][3][4]) have given some overhead analysis for the two options:
`1. Separate PDCCH for each component carrier: Each PDCCH assigns only a single TB (in absence
`of spatial multiplexing), as done in Rel-8. Existing DCI formats may be reused. If each separate
`PDCCH is sent on the same component carrier as the corresponding TB, then the PDCCH-TB
`correspondence is implicitly implied. If the DL component carrier used for PDCCH transmission
`is allowed to be different from the DL component carriers occupied by the TB, additional bits to
`convey the PDCCH-TB correspondence will be needed, which may make it impossible to reuse
`the exact Rel-8 DCI formats.
`2. Common PDCCH for multiple component carriers: Resource assignment and MCS information
`for all TBs, corresponding to multiple component carriers, is signalled in a common PDCCH.
`In [2], a slightly modified option (denoted as option “2a” here) where one PDCCH that
`•
`contains only common fields and another PDCCH containing a joint resource assignment
`was also proposed. The overhead of this option is a bit larger than having only one
`PDCCH, but can potentially improve the robustness of PDCCH detection reliability.
`The two options are illustrated in Figure 1.
`
`Samsung Ex. 1007
`Page 2 of 8
`
`

`

`PDSCH 1
`
`PDSCH 1
`
`DL Carrier 1
`
`DL Carrier 1
`
`PDSCH 2
`
`PDCCH
`
`PDSCH 2
`
`DL Carrier 2
`
`DL Carrier 2
`
`PDCCH 1
`
`PDCCH 2
`
`PDCCH 3
`
`PDSCH 3
`DL Carrier 3
`
`(i)
`
`PDSCH 3
`DL Carrier 3
`
`(ii)
`
`
`
`Figure 1 – Separate and Common PDCCH.
`
`The pros and cons of each option may be seen from the following perspectives:
`• Overhead: The biggest overhead reduction for option-2 comes from the fact that a single CRC (or
`two CRCs for option-2a) is needed as opposed to “M” 16-bit CRC with “M” being the number of
`component carriers.
`• Blind decoding: Option-1 requires “M” times of the number of Rel-8 blind decoding. Option-2
`will result in a DCI format size that depends on the number of component carriers used, i.e.,
`number of TBs. If a UE does not know the number of TBs in the common PDCCH and hence
`must blindly detect all possible DCI formats, there will be an increase of blind decoding too, in
`addition to that required to decode Rel-8 DCI formats (note decoding of Rel-8 DCI may not be
`necessary if LTE-A UEs are required to decode only LTE-A DCI formats).
`Impact to existing PDCCH structure: Given that LTE-A PDCCH may well multiplexed in a
`subframe with Rel-8 PDCCH messages intended for Rel-8 UEs, the PDCCH structure in terms of
`aggregation rule is better kept unchanged. A common PDCCH could sometimes result in a new
`DCI format of large size so that the coding rate under the existing aggregation rule becomes too
`high and thus making PDCCH reception unreliable, especially for cell-edge users that require
`carrier aggregation to satisfy their high data rate needs. Further investigation is needed in this
`topic.
`• PDCCH error event handling: In case of any PDCCH error, option-1 has a localized impact. The
`probability of any one of “M” PDCCHs in error for option-1 can be higher than the error rate of
`option-2 if we can assume the code rate of separate PDCCH and common PDCCH is the same.
`Considering the above three options, our view on PDCCH contents is
`
`•
`
`Samsung Ex. 1007
`Page 3 of 8
`
`

`

`In scenarios where TBs for multiple component carriers are scheduled from a single component
`carrier, resource assignment mechanisms (and PDCCH structures) that exploit commonality
`between individual component carrier assignments are beneficial as they reduce signalling
`overhead.
`
`2.3. PDCCH Beam-forming
`Beam-forming of the PDCCH based on, for example, PMI feedback or dedicated pilots has been shown to
`significantly improve performance and coverage [5]. When compared to open-loop transmit diversity
`techniques such as cyclic shift transmit diversity, beam-forming is most beneficial when four or more
`transmit antennas are available. Therefore, it is recommended that PDCCH performance enhancement
`techniques such as beam-forming should be investigated for LTE-A.
`
`3. SCH and P-BCH
`Structure
`No additions to existing P/S-SCH and P-BCH structure are envisioned. However, the transmission of SCH
`and P-BCH may not always be needed for non Rel-8 compatible carriers. Flexibility of configuring a
`component carrier non-backward-compatible (e.g., LTE-A only) can be desirable for future deployment. If
`so configured, Rel-8 P/S-SCH and P-BCH may not be transmitted, possibly along with other Rel-8
`messages such as SIBs. Note that the overhead saving by removing only P-BCH is rather small, especially
`for large component carrier bandwidth (0.17% for 20MHz and 2.8% at 1.4MHz). It is still possible to
`convey the system configuration information of non Rel-8 component using SIBs sent on a Rel-8
`compatible component carrier.
`Content
`Modified P-BCH may be used to support additional signalling related to bandwidth extension information
`that can be beneficial to LTE-A UEs if such information is made available. Additional spare bits that are
`already available in MIB may be employed for this purpose.
`
`4. P-CFICH Structure
`No additions to existing P-CFICH structure are envisioned. Some co-ordination on values signalled on P-
`CFICH of different component carriers may be helpful, especially if a common PDCCH within an anchor
`component carrier conveys the RA information for PDSCH in different component carriers. UE
`implementation may be slightly simplified if it can assume that same ‘number of control symbols’ is used
`in all aggregated component carriers. Otherwise, the resource allocation for each component carrier may
`need to include information on the starting symbols of each TB.
`
`5. P-HICH Structure
`If a PDCCH on one DL component carrier is allowed to grant resources for multiple UL component
`carriers (e.g., in asymmetric aggregation with more UL components than DL components), then that DL
`component carrier should also signal ACK/NACK corresponding to multiple UL component carriers. In
`such a scenario, additional PHICH resources may be required to support bandwidth extension.
`
`Samsung Ex. 1007
`Page 4 of 8
`
`

`

`6. PUCCH under Asymmetric Aggregation
`Spectrum availability could result in asymmetric aggregation in theory. One may think of a case where a
`new acquired TDD spectrum is used either for DL or UL only (i.e., 100% DL or UL) so that an operator
`does not have to deploy a TDD network to make use the spectrum. However, one has to realize that
`regulation may often not allow 100% DL/UL due to coexistence issue with other TDD networks on
`neighboring bands.
`Asymmetric aggregation could also result from spectrum reconfiguration. For example, two adjacent
`20MHz-DL 10-MHz-UL system could be reconfigured as two DL component and one 120MHz-UL
`component for better UL efficiency.
`The above asymmetric aggregation scenarios are system-wide that affects all UEs in the system.
`Asymmetric aggregation can also be UE-specific due to implementation limitation or cost saving. For
`example, a UE can support only one UL component carrier, even though the network has multiple UL
`component carriers. Note that even in this case a UE may still be able to support a different UL component
`carrier at a different time, on a semi-static basis.
`In this section, we will discuss the different configuration options and the associated compatibility
`limitations, as well as the impact on PUCCH design. We will focus on the case of more DL component
`carriers than UL carriers, but the case of more UL carriers may of interest to some applications with high
`UL traffic (enterprise, fixed CPE, etc.).
`
`6.1. Deployment Option 1
`
`
`DL2 (Rel8)
`
`DL1(Rel8)
`
`UL1
`(pairing with Rel8
`DL1, Rel8 DL2, LTE-A
`DL1+DL2)
`
`'
`
`I
`I
`
`Fd1
`(default duplexer gap as
`defined in RAN4)
`Fd2
`(non default gap)
`
`
`
`Figure 2 – Asymmetric aggregation where all DL carriers are Rel8 accessible
`
`As shown in Figure 2, one deployment option is configure all the DL aggregated carriers to be Rel8
`accessible1. Note that since current RAN4 specifications only test a Rel8 UE under a single band specific
`1DF is the band specific default
`Tx-Rx duplexer separation (Section 5.7.4 in [6]). In Figure 2, we show that
`Tx-Rx separation that is tested for Rel8 (a Rel8 UE accessing component carrier DL1 could, by default,
`1DF ). In order to notify a UE that DL
`transmit in an uplink resource cantered with Tx-Rx separation
`component carrier DL2 is also paired with UL1, eNB will have to transmit an “ul-EARFCN” value in SIB2
`2DF to inform the UE to transmit in component carrier UL1. However,
`corresponding to Tx-Rx separation
`
`1 By Rel8 accessible we mean - a Rel8 UE can locate the P-SCH, S-SCH transmitted by that carrier and
`also possibly download the MIB and SIBs
`
`Samsung Ex. 1007
`Page 5 of 8
`
`

`

`Rel8 UE behaviour for non default Tx-Rx separation values is not tested by the RAN4 conformance
`specifications. The issue of whether the DL2/UL1 pairing is called Rel-8 compatible or not is not within the
`scope of RAN1. No change is needed for RAN1 specifications.
`Since in this case two DL component carriers are paired with a single UL component, a number of issues
`needs to be examined as discussed in [7][8], in particular with respects to the splitting of UL resources
`(PRACH, PUCCH, PUSCH, etc.). If resources for a channel are not split in a predetermined manner, such
`as PRACH’s code space, due to concerns of loss of packing/multiplexing efficiency, then issue of whether
`RACH response should be sent from both DL carriers, during initial network entry for example, can arise.
`
`6.2. Deployment Option 2
`
`DL2 (not accessible to Rel8)
`
`DL1(Rel8)
`
`UL1 (Rel8 DL1,
` LTE-A DL1+DL2)
`
`'
`
`I
`I
`
`Fd1
`(default duplexer gap as
`defined in RAN4)
`
`Fd2
`(non default pairing)
`
`
`
`Figure 3 – Asymmetric aggregation where one DL carrier is not Rel8 accessible
`
`In Figure 3, we show an option where one component DL carrier is configured as inaccessible to Rel8 UEs.
`Of course, DL2 will be used for LTE-A UE only. Clearly no Rel8 DL signalling is required for DL2. The
`question of whether DL2 bears its own non Rel-8 signalling or used as a “traffic-only” carrier needs to be
`studied for that case. LTE-A UEs may complete the initial network entry process from DL1 or DL2 (if DL2
`has its own sync channel). Both Rel8 and LTE-A UEs will share UL1 in this case, which should be similar
`to the study needed for option-1.
`
`6.3. PUCCH Design
`The PUCCH transmission scheme should be designed to handle both asymmetric and symmetric bandwidth
`allocation for UL. The logical choice is to utilize the same PUCCH structure as in LTE Release-8. In case
`when the number of carriers to be aggregated in UL is lower than that of DL, the UL PUCCH transmitted
`on one component carrier will carry information for multiple downlink component carriers as shown in
`Figure 4, where only the case of separate PDCCH for DL resource allocation is shown as an example (joint
`PDCCH for DL is also possible).
`
`Samsung Ex. 1007
`Page 6 of 8
`
`

`

`A/N
`
`A/N
`
`PUCCH 2
`
`PUCCH 1
`
`PUCCH 2
`PUCCH 1
`UL Carrier 1
`
`
`
`DL Assignment
`
`PDSCH 1
`
`DL Carrier 1
`
`DL Assignment
`
`PDSCH 2
`
`DL Carrier 2
`
`PDCCH 1
`
`PDCCH 2
`
`Figure 4 – Example of UL control with asymmetric UL/DL carriers.
`
`CQI/PMI Transmission Scheme: In LTE Release-8, a UE may be semi-statically configured by the higher
`layers to periodically feedback CQI, PMI and RI on the PUCCH using four different modes. In LTE-A, the
`CQI reports for each TB will be required. One straightforward method is to keep the Rel-8 structure and
`separately code the CQI bits and report CQI/PMI/RI for each component carrier independently.
`A/N Transmission Scheme: The A/N transmission scheme structure should be backward compatible with
`Rel-8 PUCCH structure. The A/N scheme with carrier aggregation may be similar to A/N scheme in TDD
`when the number of DL carriers (sub-frame in TDD) is not the same as UL carriers (sub-frame in TDD).
`On the other hand, multi-code or multi-channel transmission may also be supported if single-carrier
`transmission requirement may be relaxed with increased CM. An example of this multi-channel A/N
`mapping scheme for asymmetrical DL and UL carrier is shown in Figure 4.
`
`7. Conclusion
`Our views on control signalling for carrier aggregation are summarized below
`PDCCH
`-
`o Transmission structure is confined to a single component carrier
`In addition to having an option that uses one PDCCH to schedule TBs in one component
`o
`carrier (which is the simplest extension of Rel8 operation) the option of allowing a
`PDCCH transmitted from one DL component carrier to schedule TBs in other DL
`component carriers should also be considered.
`In scenarios where TBs for multiple component carriers are scheduled from a single
`component carrier, resource assignment mechanisms (and PDCCH structures) that exploit
`commonality between individual component carrier assignments and reduce signalling
`overhead are preferred.
`o PDCCH performance enhancement techniques such as beam-forming should be
`investigated.
`
`o
`
`-
`
`PBCH
`
`Samsung Ex. 1007
`Page 7 of 8
`
`

`

`o
`
`If a PBCH is transmitted in a component carrier, then same structure as Rel8 is
`employed. Otherwise, it may not be transmitted if so configured in non Rel8 component
`carriers.
`o PBCH content may be augmented to support LTE-A specific broadcast signalling
`PCFICH
`o
`
`If a PCFICH is transmitted in a component carrier, then same structure as Rel8 is
`employed.
`
`-
`
`-
`
`PHICH
`
`o Need for additional PHICH resources that corresponding to multi-component carrier
`uplink grants should be further investigated.
`
`-
`
`PUCCH
`o Need to handle both symmetric and asymmetric aggregation
`
`8. References
`[1] R1-090233, “Transport block mapping and DL control signaling in LTE-Advanced”, Nokia, NSN,
`RAN1#55bis, Jan 2009.
`[2] R1-090359, “Multi-carrier control for LTE-A”, Qualcomm, RAN1#55bis, Jan 2009.
`[3] R1-090375, “Control signaling for carrier aggregation”, Ericsson, RAN1#55bis, Jan 2009.
`[4] R1-090337, “Resource Allocation and PDCCH Design Issues in Carrier Aggregation”, CMCC,
`RAN1#55bis, Jan 2009.
`[5] R1-070032 – “Support of Precoding for E-UTRA DL L1/L2 Control Channel”, Motorola,
`RAN1#47bis, Jan 2007.
`[6] 3GPP TS 36.101 v8.4.0, “E-UTRA UE radio transmission and reception”, December 2008.
`[7] R1-082999, “Support of UL/DL asymmetric carrier aggregation”, Panasonic, RAN1 #54, Aug 2008.
`[8] R1-090211, “Considerations on DL/UL Transmission in Asymmetric Carrier Aggregation”, LGE,
`RAN1#55bis, Jan 2009.
`
`
`
`Samsung Ex. 1007
`Page 8 of 8
`
`