1. Introduction
In 4C-HSDPA, one of the open issues to be resolved is the design of the HS-DPCCH codebook for the case when 3 carriers are configured without MIMO on any of the carriers. This configuration was considered by some to be a more common scenario and one which warranted further optimization.
In this contribution, we show link level simulation results for the HS-DPCCH design options proposed for the case when the UE is configured with 3 carriers without MIMO on any of them.
2. HS-DPCCH Design Schemes
When the UE is configured with 3 carriers without MIMO, ACK/NACK and CQI feedback is required for carriers C1, C2 and C3. There are two design approaches proposed for this case. They are detailed below.
It should be noted that both the design approaches are essentially independent of whether the spreading factor SF256 or SF128 is chosen for this configuration. However, it is desired that a consistent approach to HS-DPCCH design is followed -- i.e., the same HS-DPCCH design is used when the three carriers are configured or are active based on an HS-SCCH order based activation/deactivation.
In this scheme, the ACK/NACK information is jointly coded into a single codebook. The codebook used in this simulation corresponds to Option 2 in [1]. The codebook contains 24 code words and is backward compatible with the Rel-8 DC-HSDPA ACK/NACK codebook which assures a measure of robustness upon deactivation. This scheme was simulated using a single code with SF256.
The CQI information is encoded using legacy Rel-8 DC-HSDPA CQI and Rel-5 SC-HSDPA CQI codebooks. The CQI information for carriers C1 and C2 are jointly encoded in the Rel-8 DC-HSDPA codebook and is transmitted in a TDM fashion with the CQI information corresponding to carrier C3 which is encoded using the Rel-5 SC-HSDPA codebook.
Figure 1: Block diagram representation of a single codebook using 1xSF256 when the UE is configured with 3 carriers without MIMO
NodeB processing for Scheme 1:
The NodeB is aware of the fact that the HS-DPCCH channel carries feedback information pertaining to three carriers.
DTX detection is performed based on the energy over the entire slot
HS-DPCCH decoding is also based on the energy over the entire slot.
The codebook used contains 24 code words and corresponds to Option 2 in [1]
In this scheme, legacy codebooks are re-used to encode the ACK/NACK information for the 3 carriers. This design is considered by some to be a simple and straightforward solution [3, 4]. This scheme necessitates the use of SF128 in order to transmit all the information in a single slot. The codebook used to encode the carriers is the Rel-8 DC-HSDPA codebook. Carriers C1 and C2 are encoded in the first half slot and carrier C3 is encoded into the second half slot.
The CQI information is encoded using legacy Rel-5 SC-HSDPA CQI codebooks and is transmitted in a TDM fashion. This CQI transmission scheme is based on the proposal in [3]. Figure 2 illustrates the HS-DPCCH design for Scheme 2.
Figure 2: Block diagram representation of the usage of legacy codebooks on 1xSF128 when the UE is configured with 3 carriers without MIMO
NodeB processing for Scheme 2:
The NodeB is aware of the fact that the HS-DPCCH channel carries feedback information pertaining to only three carriers.
DTX detection is performed based on the energy over the entire slot
A POST codeword is used in place of a half-slot DTX when less than three carriers are scheuled
HS-DPCCH decoding is also based on the energy over each half slot.
For carriers C1 and C2: The codebook is the Rel-8 DC-HSDPA codebook which contains 8 code words. Therefore, the NodeB considers a total of 9 code words as possible hypotheses (including the POST codeword).
The code words considered at the NodeB receiver are: A, N, A+D, N+D, A+A, A+N, N+A, N+N and POST
For carrier C3: The codebook is the Rel-8 DC-HSDPA codebook which contains 8 code words. However, since the NodeB is aware that feedback information for a single carrier C3 is to be received, the codebook can be restricted to only three code words including POST.
The code words considered at the NodeB receiver are: A, N and POST
3. HS-DPCCH ACK/NACK Simulation Assumptions
3.1 Simulation Metrics
False Alarm
This event occurs when the NodeB falsely detects data when the UE transmits only DTX. For all the results shown in this document, the target false alarm probability is set to be 10%
Misdetection
This event occurs when the NodeB does not detect data and the UE transmits data. The probability of misdetection is plotted in for all the schemes and scenarios simulated in the next section.
Misdetection or Decoding error
This event occurs when one of the following events occur
The NodeB does not detect data when the UE transmits data, OR
The NodeB correctly detects data but decodes it incorrectly.
The probability of this event is shown for the schemes and scenarios simulated in the next section.
4. HS-DPCCH ACK/NACK Simulation Results
In this section, simulation results are shown for Schemes 1 and 2 described in Section 2. The simulation assumptions and the metrics used are described in Section 3.
Figure 3 shows the probability of misdetection and the probability of misdetection or decoding error for the two schemes outlined in Section 2.. The results are shown for the AWGN channel for a target False Alarm Probability = 0.1. Results for the PA3, PB3, VA30 and VA120 channels can be found in Figures 5-8 in the Annex.
Figure 3: Simulation results for Schemes 1 and 2; 3 Carriers without MIMO; AWGN; FAR = 10%
Observation
The following observations can be made from Figures 3, 5 -- 8:
For all the channels simulated, Scheme 1: 1xSF256 with a single codebook performs better than Scheme 2: 1xSF128: Legacy Codebooks when both misdetection and decoding error is considered.
When a target of 1% misdetection or decoding error is considered, the difference in performance ranges from over 1dB for VA120 channel to 2dB for the PA3 channel. The data points simulated correspond to the HS-DPCCH C/P values that are permitted in the specification. Therefore, it is clear that even if the quantized C/P values are considered, Scheme 1 has better transmit power savings.
When only misdetection is considered, Scheme 2 performs better than Scheme 1, but the difference in performance is not significant (<0.5dB).
It should be noted that the final design choice of the HS-DPCCH channel should consider the overall transmit power of the UE. Therefore, cubic metric should be considered in this regard. Based on the cubic metric analysis in [5], it is seen that the difference between 1xSF128 and 1xSF256 is small. It should also be noted that 2dB power savings in the HS-DPCCH C/P does not translate directly into gains in the UE headroom. However, it is considered that at the cell edge, the UE is likely to be transmitting small packet sizes where the E-DPDCH T/P would be comparable to the HS-DPCCH C/P. As a result, we consider that Scheme 1 would indeed allow for UE transmit power gains when considered as a whole.
Proposal 1: Adopt a single code book Rel-10 TC-HSDPA solution for the case when the UE is configured with 3 carriers without MIMO.
As mentioned earlier, the codebook simulated in this document corresponds to Option2 in [1]. However, this codebook required that a code word be mapped to three unique A/N signals. This leads to ambiguity at the NodeB when this codeword is decoded. Therefore, it is desired to select a a unique set of 26 codewords with backward compatibility to the Rel-8 DC-HSDPA codebook so that a robust solution can be obtained when one of the carriers is deactivated.
Proposal 2: The code words that are members of the codebook Rel-10 TC-HSDPA are FFS.
Note also that a single code/single codebook with SF256 can be used with a NodeB SF128 de-spreader. For more details, the reader is referred to [6].
5. HS-DPCCH CQI Design
We consider the UE transmit power for the CQI portion of the HS-DPCCH channel under the same scenario -- 3 carriers without MIMO. Revisiting the Schemes that were defined in Section 2, we have
Scheme 1: Since a Rel-8 DC-HSDPA CQI codebook is used, the transmit power is x+2dB where x is the transmit power for the Rel-5 SC-HSDPA.
Scheme 2: In this case, there is a 3dB loss in processing gain due to the reduction in spreading factor. Therefore the required transmit power is x+3dB.
As a result, Scheme 1 is more power efficient when CQI transmissions are considered. Note that further optimizations could be achieved if the CQI transmit power were to be specified on a per-carrier basis. In this case, the CQI information for carrier C3 could potentially be transmitted 2dB lower. However, such a design is not extensible to 4C-HSDPA, where specifying carrier specific CQI power levels would become excessively complicated. Furthermore, the E-TFC selection is based on the worst case power level amongst ACK, NACK and CQI. Since the ACK/NACK portion of the HS-DPCCH channel requires the most amount of power, there would not be any direct gains in UE headroom and consequently throughput as a result of the lower power CQI transmissions. Therefore, it is considered that the same CQI offset (C/P) be used for all the carriers based on the configuration.
Proposal 3: A single CQI value is set for all carriers in a particular configuration. However, the CQI value could be different for different configurations.
In addition to the Schemes considered in Section 2, there have been proposals from companies for alternate CQI designs under the scenario where the UE is configured with 3 carriers without MIMO.
Cross-Combined CQI
It has been proposed in [1, 2], that cross-combined CQI transmissions are used when the UE is configured with 3 carriers without MIMO. It has been shown in [7] that reducing the feedback cycle from 4ms to 2ms does not achieve appreciable gains in the case of MIMO carriers. When carriers are configured without MIMO, the gains are expected to be quite small (<2%). Additionally, cross-combined CQI transmissions result in additional complexities at both the UE and the Node B. It is therefore proposed that the cross-combined CQI transmissions be not considered as a viable option for CQI design.
Minimum CQI Feedback cycle = 2ms by Joint Coding
As an alternative to Scheme 2, it is possible to transmit CQI with 2ms feedback cycle when 3 carriers without MIMO is configured. Figure 4 illustrates this design option.
Figure 4: Block diagram representation of the HS-DPCCH design with CQI feedback cycle = 2ms when 3 carriers without MIMO are configured
It can be seen from Figure 4 that since the Rel-8 DC-HSDPA codebook is used to encode the CQI information for carriers C1 and C2, a power level of x+5dB is required since there is a loss in processing gain in addition to a loss in coding gain. Therefore, it is considered that this design option is also not a serious candidate.
Based on the analysis in Sections 4 and 5, we consider that a single codeword solution with SF256 would offer the ideal trade off between low power, performance and robustness. This design scheme would also allow the NodeB to use a de-spreader with SF128 by the use of a simple symbol-level interleaver as shown in [6].
Proposal 4: The HS-DPCCH ACK/NACK and CQI information is transmitted using a 1xSF256 code when the UE is configured with 3 carriers without MIMO.
6. Conclusions
In this contribution, HS-DPCCH link level simulation results were shown for 2 schemes for the case with the UE is configured with 3 carriers without MIMO. The link results shown are probabilities of misdetection and the probability of misdetection or decoding error for the ACK/NACK feedback. The CQI design for the same scenario was also discussed.
Based on the link simulation results and the discussion presented, the following proposals are made:
Proposal 1: Adopt a single code book Rel-10 TC-HSDPA solution for the case when the UE is configured with 3 carriers without MIMO.