3GPP TSG RAN WG1 Meeting #49bis
`Orlando, FL, USA, 25-29 June, 2007
`Agenda item:
`5.13.2
`Source:
`Samsung
`Title:
`Control Signaling Location in Presence of Data in E-UTRA UL
`Document for: Discussion/Decision
`
` R1-073094
`
`Introduction
`1.
`This contribution considers the data-associated PUCCH transmission. In particular, placement aspects
`in the UL sub-frame for the ACK/NAK and CQI signals are considered with respect to performance
`and other general E-UTRA requirements.
`
`The location of the data-associated CQI and ACK/NAK signals in the UL sub-frame has been
`addressed in [1, 2]. In [1], the suggestion is made for the CQI signaling to be placed at the first symbol
`of the sub-frame in order to reduce scheduling latency. However, as the demodulation RS (DM RS) is
`placed in the middle of the slot, this is not applicable. In [2], the ACK/NAK signal is assumed to be
`distributed substantially over the entire sub-frame (in particular, in the first 3 symbols of each slot).
`
`The previous suggestions however have not considered two important aspects of the overall E-UTRA
`UL transmission. The first is that DM RS exists only in the middle symbol of each slot. This results to a
`substantially degraded channel estimate for the symbols at the beginning and end of each slot for the
`higher UE speeds that need to be supported in E-UTRA. This may be acceptable for the data
`transmission which is coded, has a relatively large target BLER of 10% or above, and can benefit from
`HARQ. Conversely, the PUCCH transmission has much stricter performance requirements, particularly
`the ACK/NAK signaling.
`
`The second aspect is the transmission of a sounding RS (SRS) in synchronous systems. Clearly, the
`SRS cannot be transmitted at the middle symbol of each slot where the DM RS is placed. Distributing
`the CQI and ACK/NAK signals substantially over the entire UL sub-frame will either severely restrict
`the SRS placement or introduce additional complexity and performance loss in the reception of
`ACK/NAK and/or CQI signals as puncturing will be dynamically needed in a symbol depending on
`whether or not the SRS is transmitted in that symbol. Having as many as possible locations for SRS
`transmission is desirable because, for proper CQI and power control measurements, the SRS should
`capture interference from data transmission and not from other SRS transmission, that is, SRS
`transmission from neighboring cells and Node Bs should not coincide.
`2. Raw BER Performance per UL Sub-Frame Symbol
`The raw BER per UL sub-frame symbol is now evaluated as a function of the UE speed. Table 1 gives
`the simulation assumptions. Several key ones are chosen so that they provide the most optimistic
`scenario for the raw BER performance loss of symbols further away from the DM RS. They include:
`a)
`Transmission over 1 RB to maximize power per sub-carrier.
`b)
`Large channel frequency selectivity and 2 uncorrelated receiver antennas in order to
`maximize the slope of the raw BER curve.
`BER evaluation per QPSK symbol and no repetition of bits. This leads to higher SINR for a
`target raw BER. For ACK/NAK transmission, where a target raw BER should be achieved
`at SINRs as low as -5 dB or lower, the channel estimation losses will be (much) larger.
`
`c)
`
`1
`
`APPLE 1008
`
`

`

`Parameters
`Numerology
`Modulation
`Data Allocation
`UE Speed
`Transmission Type
`Channel Model
`Number of Receive Antennas
`Number of Transmit Antennas
`
`Table 1: Simulation Assumptions
`Assumptions
`5MHz @ 2.6 GHz
`QPSK
`1 RB
`3, 30, 120 and 350 kmph
`Localized at 3, 30 Kmph
`Frequency Hopping Between Slots at 120 and 350 Kmph
`TU6
`2
`1
`
`
`Figure 1 presents the raw QPSK bit error rate (BER) for the examined cases outlined in Table 1. The
`BER at symbol locations symmetric to the DM RS is typically the same. This is clearly the case at 120
`Kmph and 350 Kmph due to the frequency hopped transmission per slot. At low speeds, such as 3
`Kmph, this is because the channel does not change. However, for some UE speeds such as 30 Kmph,
`the raw BER at equidistant symbols from the RS in a slot, is lower for the ones after the RS of the first
`slot and before the RS of the second slot as the channel estimate, that uses the RS of both slots, is more
`accurate. Nevertheless, for simplicity, the average BER of equidistant symbols from the DM RS is
`shown in Figure 1 at 30 Kmph.
`
`
`
`
`Figure 1: Raw BER as a Function of the Slot Symbol and the UE Speed.
`
`
`Even under the previous, most optimistic, assumptions for the raw BER degradation at symbols further
`away from the DM RS, at 350 Kmph, the BER saturates at the 1st/7th and 2nd/6th symbols. However, the
`impact on the BER of the 3rd/5th symbols is rather contained and saturation is avoided (the difference
`relative to the BER at 3 Kmph is also partly due to the fact that the latter uses both RS in the sub-frame
`for channel estimation, that is, channel estimation is operating with 3 dB more SINR). The BER at 120
`Kmph, relative to the one of the 3rd/5th symbols at about 1% BER, is also degraded by about 3 dB for
`the 1st/7th symbols and by about 1.5 dB for the 2nd/6th symbols. Obviously, due to the flattening of the
`BER curves for the 1st/7th and 2nd/6th symbols, the degradation will be much larger for operating points
`below 1% as needed for the NAK reception.
`
`
`
`2
`
`

`

`3. Location of Data-Associated PUCCH
`Based on the results in the previous section, it becomes apparent that the data-associated PUCCH
`should be placed immediately next to the DM RS. Figure 2 shows an example for such placement
`when a UE transmits both ACK/NAK and CQI during a sub-frame by applying TDM with data.
`
`
`Data
`
`RS
`
`ACK/NAK
`
`CQI
`
`Data
`
`Data
`
`Data
`
`1 slot
`RS
`
`Data Data
`
`Data
`Data
`1 sub-frame
`
`Data Data
`
`RS
`
`Data Data
`
`Data
`
`
`
`Figure 2: Placing of Data-Associated ACK/NAK and CQI Transmission
`
`
`To minimize channel estimation losses, ACK/NAK should be placed with priority in the symbol after
`the first RS. Notice that this does not impact demodulation latency as a channel estimate is available
`only after the first RS (clearly, there is no use from earlier transmission with respect to latency).
`
`Subsequently, to address low SINR or coverage issues, the ACK/NAK can be placed in the symbol
`before the second RS. The reason is that for medium UE speeds, this second ACK/NAK placement
`benefits from improved channel estimation and time diversity while for high UE speeds, it benefits
`from frequency and time diversity. The tradeoff is the increased latency which however is not critical
`for the ACK/NAK transmission (data non-associated PUCCH is assumed to be transmitted over the
`entire sub-frame). Even though LTE is optimized at low speeds, the CQI and ACK/NAK performance
`targets are similar at high UE speeds which the most challenging operating conditions.
`
`Based on the results from [2], provisioning for ACK/NAK transmission in 2 symbols is comfortably
`adequate to achieve 1% BER even for the lowest geometries and transmission over 1 RB. Nevertheless,
`since the NAK reception has lower BER requirements, it is appropriate for robustness and to achieve
`time and frequency diversity, to have the ACK/NAK transmission over a number of sub-carriers in 1
`symbol of each slot.
`
`If further ACK/NAK transmissions are needed, because of SINR or coverage issues, or for some
`interference randomization in asynchronous systems, the other symbols next to the RS in the 2 slots
`may also be used. However, the number of symbols where ACK/NAK and CQI transmission may
`occur should be minimized as it directly affects the number of available symbols for SRS transmission
`in synchronous systems (as previously mentioned, for proper SRS operation, SRS from different cells
`or Node Bs should not overlap in time, as much as this is possible).
`
`If latency is a concern for the CQI reception, it may be confined only in the first slot in Figure 2. This
`may be feasible for UEs in good SINR conditions but, in general, it will lead to worse performance due
`to losses from less accurate channel estimation and less diversity. However, as data non-associated CQI
`is already assumed to be transmitted over the entire sub-frame, minimizing latency for the data-
`associated CQI transmission becomes even less important and emphasis should be placed on
`performance.
`
`
`
`3
`
`

`

`4. Conclusions
`This contribution considered the placement of the data-associated PUCCH. Due to the non-uniformity
`of the RS across the sub-frame and the inability, in general, to interpolate across sub-frames, symbols
`closer to the RS are considerably more reliable at medium and high UE speeds, even under optimistic
`evaluation scenarios for the relative BER performance in symbols further away from the DM RS.
`
`Therefore, the placement of ACK/NAK and CQI signaling, having low BER requirements and not
`being able to benefit from HARQ, should be at the symbols immediately next to the RS. To obtain time
`and/or frequency diversity, transmission in both slots should be utilized. CQI reporting latency may be
`minimized, for some performance loss, by confining the corresponding signaling only to the first slot.
`
`The unequal BER at the various symbols of the sub-frame can also be exploited at the receiver by
`appropriately weighting the log-likelihood ratio (LLR) of the data bits provided to the turbo decoder in
`order to improve the BLER at high speeds.
`
`
`References:
`[1] R1-051395, “Mapping Position of Control Channel for Uplink SC-FDMA”, Panasonic
`[2] R1-072313, “ACK/NACK Transmission with UL Data”, Nokia
`
`
`
`
`4
`
`