hash stringlengths 32 32 | doc_id stringlengths 5 12 | section stringlengths 5 1.47k | content stringlengths 0 6.67M |
|---|---|---|---|
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.6.2.2 Scenario 2: 12-reuse without Frequency Hopping | Scenario 2, 12-reuse without frequency hopping, is one of the sparse reuse scenarios investigated and could for example be users on a broadcast channel (see Note) that is sparsely planned to ensure secure signalling operation.
NOTE : three non-hopping frequencies are used per cell in this scenario, thus it is not ident... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.6.2.3 Scenario 3: 12-reuse with Random Frequency Hopping | Scenario 3, 12-reuse with frequency hopping, is the other of the sparse reuse scenarios investigated and could for example be users on a traffic channel that is sparsely planned due to more generous spectrum availability.
Since frequency hopping does not make very large differences in performance for packet data servic... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.6.2.4 Scenario 4 and 5: 1/3-reuse | Figure 135 shows the average session bit rate for different user percentiles (10th, 50th and 90th percentile). Dashed lines are with frequency hopping, solid lines are without frequency hopping. As in [4], the gains for the 90th percentile of users (session bit rates) are small since they are near the peak bit rate alr... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.6.2.5 Scenario 6 and 7: 3/9-reuse | Figure 137 shows the average session bit rate for different user percentiles (10th, 50th, and 90th percentile). Dashed lines are with frequency hopping, solid lines are without frequency hopping. Again, the gains for the 90th percentile of users (session bit rates) are small since the bit rates are near the peak bit ra... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.6.2.6 The Impact of Power Back Off | In the above results, the power back off factors given in table 50 have been used. In a similar manner as with frequency hopping, it has also been investigated how the removal of the power back off factors affects the results. This means that all the three modulations (GMSK, 8PSK, and 16QAM) will be using the same outp... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.7 Increased Peak Throughput with 16QAM and Turbo Codes | In this subclause two new MCSs are defined that increase the peak throughput compared to the conventional MCSs in EDGE today. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.7.1 Modulation, Coding and Interleaving | To increase the peak throughput higher than MCS-9, new modulation and coding schemes need to be defined. Two new MCSs have been considered, MCS-10 and MCS-11, the former one has also been evaluated in combination with turbo codes (MTCS-10). MCS-11 has been chosen to be un-coded and thus the addition of turbo codes will... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.7.1.1 Coding | Each RLC block is encoded separately (joint coding is for further study). The same mother convolutional code as used today in combination with linear puncturing has generated the coded header and data bits. For MCS-10 with turbo coding (MTCS-10) the turbo code defined for UTRAN has been used [15]. The puncturing scheme... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.7.1.2 Interleaving | For the MCS used today, depending on what MCS is used the number of bursts over where the interleaving is performed differs. For MCS-7 the data bits in the RLC block is interleaved over all four bursts while for MCS-8-9 the interleaving is performed over two bursts for each RLC block. The reason for this is the coding ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.7.2 Link Performance | A state-of-the art link level simulator for GSM/EDGE has been used to evaluate the performance of the new modulation and coding schemes. The simulator parameters are shown in table 52.
Table 52: Link simulator settings
Parameter
Value
Channel profile
Typical Urban (TU)
Terminal speed
3 km/h
Frequency band
900 MHz
Frequ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.7.3 System Performance | System simulations in subclause 8.4.3.6 have shown significant gains in spectral efficiency and average session bit rates when introducing HOM and/or TC for robustness (MCS-5-9), see [8], [9] and [29]. It was shown that the spectral efficiency could be increased up to 40 % to 60 % depending on the frequency reuse in th... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.8 16QAM with Alternative Transmit Pulse Shaping | Since decoding of turbo codes is computationally complex, it may not be feasible in all legacy BTS equipment without hardware impacts. Hardware impacts to legacy networks should be avoided according to the objectives of the GERAN Evolution Feasibility Study [2]. It is therefore of interest to look at alternative improv... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.8.1 Link Performance | Link simulations have been run to evaluate the performance of different RRC Tx pulses. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.8.1.1 Simulation Conditions | The simulation assumptions are summarised in table 56.
Table 56: Simulation parameters
Parameter
Value
Modulation/coding scheme
MCS8-16QAM
Channel
TU3 iFH (co-channel)
TU50 noFH (sensitivity)
Tx and Rx pulse shaping filter
RRC, rolloff=0.3, rectangular window, length=51 symbols
RRC Tx and RX pulse single sided 3 dB-ban... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.8.1.2 Simulation Results | Figure 145 shows co-channel interference limited link level performance with different RRC Tx pulses and Rx filters. Performance with a linearised GMSK Tx pulse is shown as reference. The interferer is a legacy GMSK modulated signal.
Figure 145: Co-channel performance for MCS8-16QAM with different Tx pulse shaping filt... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.8.2 Spectrum | Figure 147 shows the spectrum of 16QAM with different Tx pulse shaping filters. It can be seen that the RRC pulse with 110 kHz ssb bandwidth fulfils the GSM spectrum mask.
Figure 147: Spectrum of 16QAM with linearised GMSK pulse and different RRC pulses
The 8PSK spectrum mask for MS [8] is also shown |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.8.3 Discussion | Figure 145 shows that a RRC Tx pulse with a SSB bandwidth of 135 kHz gives gains of 1.5 dB in a co-channel interference limited scenario. Figure 146 shows that the gain in a sensitivity limited scenario is 2.5 dB for the same Tx pulse. With a narrower RRC Tx pulse, the gain is reduced. A pulse with a SSB bandwidth of ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.9 Higher order modulation than 16-QAM | Increasing the modulation order will make the symbols more susceptible to interference and thus the higher order the modulation the better radio conditions are needed to gain in performance. Also, increasing the modulation order to 32QAM, or maybe even to 64QAM, will probably result in such an increase of receiver comp... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.9.1 Modulation, coding and interleaving | The new MCSs for 32QAM are used to increase the robustness of the previously defined 16QAM MCSs, MCS-10 and MCS-11, but also to increase the peak throughput even further to 99.2 kbps with MCS-12.
The MCSs are summarized in table 57. For comparison previously defined MCS-7-11 for 16QAM are also included.
Table 57: Defin... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.9.1.1 Coding | Each RLC block is encoded separately (joint coding is for further study). The same mother convolutional code as used today in combination with linear puncturing has generated the coded header and data bits. For the turbo coded MCSs the code defined for UTRAN has been used [15]. The puncturing schemes for the turbo code... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.9.1.2 Interleaving | The interleaving depth used for an MCS will basically depend on the number of RLC blocks in one radio block and the code rate of the data. For example, conventional MCS-9 is uncoded and will therefore not gain from an increased diversity. Thus, each RLC data block is interleaved over 2 bursts: interleaving depth 2. MCS... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.9.2 Interference Rejection Combining, IRC | Interference rejection combining, IRC, is a diversity combining method that can be used in a multiple antenna system for suppressing mainly CO-channel interference. The suppression is possible by utilizing the cross-covariance of the interference received in the different antennas. Figure 149 shows the basic principle ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.9.3 Results | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.9.3.1 Link simulator settings | Simulations have been conducted using a state-of-the-art GSM/EDGE link simulator. The new modulation schemes utilizing 32QAM have been evaluated but also previously defined MCSs in combination with IRC. The simulation parameters are summarized in table 150.
Table 60: Link simulator settings
Parameter
Value
Channel prof... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.9.3.2 Link Simulations | In this subclause the link level results of the 32QAM MCSs are shown. Both results with and without turbo coding are presented.
Figure 150: Link performance (without antenna diversity)
of MCS-10/11/12-32QAM and MTCS-10/11-32QAM
It can be seen that the turbo coding gives an additional gain in performance of approximatel... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.9.3.3 System simulator settings | The same simulator as in subclause 8.4.3.6.1 has been used. The system level scenarios are summarised in table 63a.
Table 63a: Summary of system simulation parameters.
Parameter
Value
Scenario 1
Scenario 8
Reuse
1
1/3
Spectrum allocation
7.2 MHz (excluding BCCH)
7.2 MHz (excluding BCCH)
Frequencies per cell
36
12
Trans... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.9.3.4 System simulation results | Figure 153a shows the average session bit rate for different user percentiles (10th, 50th, and 90th percentile) in scenario 1. The gains on the 10th percentile are 15-60% for HOT level 2 (where the higher value corresponds to higher load). On the 50th percentile, the gains are 35-45% while the gain on the 90th percent... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.10 Comparison between DFSE and RSSE Performance | To evaluate the impact on performance when using the RSSE of implementation set C and a DFSE, MCS-9 performance has been simulated on a TU3 channel with ideal frequency hopping at 900 MHz. For the receiver impairments mainly the contribution from phase noise was included while transmitter impairments were not. These si... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.11 Discussion | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.11.1 Link Level Performance | Turbo codes show gains in the order of 1 dB in all evaluated cases (MTCS-5, MTCS-6, MTCS-7-16QAM, MTCS‑8‑16QAM, MTCS-9-16QAM), compared to convolutional codes. No gains have been found for 8PSK‑modulated turbo-coded equivalents of MCS-7 to MCS-9 (simulation results not shown in this report). It is clear that in order t... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.3.11.2 System Level Performance | Turbo codes together with 16QAM will give significant gains in average session bit rate also in a 1-reuse network. The gains are present (and even seem to increase) with higher loads. At 8 kbps/timeslot offered FTP load, the average session bit rate gain is about 20 % to 25 % (see figure 130). 8 kbps/timeslot offered F... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4 Implementation Set D | Source: References [10], [11], [12], [13], [14], [31], [41] and [47]. Note that there are additional details of simulation results in annex C. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1 Performance Characterisation | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.1 Uncoded BER Performance | Figure 155 shows the uncoded BER performance of the basic receiver for 8-PSK and 16-QAM modulation. This shows good alignment of the basic receiver performance with previously reported results by other contributors [6].
Figure 155: Uncoded BER for 8-PSK under TU3 Channel |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.2 BLER Performance of Turbo Coding with 8PSK | This subclause reports the results of simulations using logical channel configurations defined in table 64. These define combinations of Turbo coding with 8-PSK modulation only.
Table 66 shows the relative gains for TU3iFH Co-Channel scenario. As can be observed, the gains are relatively modest, in the region of 1 dB t... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.3 BLER Performance of Turbo Coding with 16QAM | This subclause reports the results of simulations using logical channel configurations defined in table 65. These define combinations of Turbo coding with 8-PSK or 16-QAM modulation.
Table 65: Modulation and Coding Configurations - with 16QAM
Modulation and Coding Scheme
Data Block Length (bits)
Coding
Data Code Rate
I... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.4 Comparison to Ericsson Results | These results have been compared to those reported by [19] for the co-channel scenario. Although, there is not an exact alignment, the Ericsson results also confirm the assertions based on the results presented here. It should be noted that the improvements (both as absolute C/Ico and relative) are larger than those re... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.5 Graphs for Co-Channel Interferer Case (TU3iFH) | Figure 156: TU3iFH Co-Channel Performance (MCS-5)
Figure 157: TU3iFH Co-Channel Performance (MCS-6)
Figure 158: TU3iFH Co-Channel Performance (MCS-7)
Figure 159: TU3iFH Co-Channel Performance (MCS-8)
Figure 160: TU3iFH Co-Channel Performance (MCS-9)
Figure 161: TU3iFH Co-Channel Performance (MCS-10 16-QAM)
Figure 162: ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.6 Graphs for Sensitivity Limited Case (TU3iFH) | Figure 163: TU3iFH Sensitivity Performance (MCS-5)
Figure 164: TU3iFH Sensitivity Performance (MCS-6)
Figure 165: TU3iFH Sensitivity Performance (MCS-7)
Figure 166: TU3iFH Sensitivity Performance (MCS-8)
Figure 167: TU3iFH Sensitivity Performance (MCS-9)
Figure 168: TU3iFH Sensitivity Performance (MCS-10 16-QAM)
Figure... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.7 Throughput Performance Gain | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.7.1 Approximation for Throughput Gain | This subclause presents the performance gain of a number of different logical channel set combinations under the physical channel scenarios considered. It is assumed that there is ideal Link Adaptation. The throughput of a logical channel combination is approximated as:
Throughput = (1-BLER) * DataPayloadPerBlock * Blo... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.7.2 Full Incremental Redundancy | Incremental redundancy is now included for simulation of throughput for EGPRS and the Turbo coded logical channels. This is now using RLC/MAC to perform Incremental Redundancy rather than throughput approximations previously used. Puncturing (or repetition) is applied as defined in 3GPP RAN [15]. For a target coding ra... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.7.2.1 Impact of Mobile Speed | The variety of mobile speeds in the network is taken into consideration here. Performance results are included for speeds of 3 km/h, 50 km/h and 120 km/h. An example of expected distribution of users as a function of speed is taken from the Nortel OFDM study that pre-dates the current RAN LTE work [23]. Table 70 shows ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.7.2.2 Hilly Terrain Channel | Figure 185 shows the throughput for MCS9 and MCS9-T4-16QAM in the HT 100 km/h channel. The throughput performance gain is shown in figure 186. It can be seen that for SNR above 10 dB that throughput improvements of up to 50 % are achievable.
[Note that the scale PathGain represents SNR in dB for GMSK modulation. For 8P... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.7.2.3 Performance of HOMTC in Non Hopping Configuration | A comparison of throughput performance for both hopping and non-hopping layers, and at different mobile speeds, is shown in figure 187 for the MCS9-T4-16QAM logical channel. Also included for reference is the MCS9 throughput for TU 3 km/h non-hopping; remember from figure 180 that MCS9 performance decreases with speed.... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.8 Number of Turbo Decoding Iterations | Figures 188 and 189 (zoomed version of figure 188) show the impact on performance of reduced iterations in the Turbo decoder. As expected there is some degradation of throughput performance with reduced Turbo iterations, but it is by no means catastrophic. This would make use of Turbo coding on the uplink easier to emp... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.9 Improved Cell Edge Performance | This subclause shows initial results of investigation to improve throughput performance at the cell edge. As described earlier, 8PSK modulated Turbo coded schemes have been compared to the current EGPRS GMSK modulated schemes, MCS1 to MCS4. The new MCSs are defined in table 71 such that the new logical channels carry t... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.10 System Performance | A full system analysis has not been performed. However, based on the results it is possible to make some general comments.
In order to understand relevant C/I operating values, the C/I CDFs presented recently in contributions by Ericsson [8] and TeliaSonera [5] have been used. The curves are shown in figure 195.
Turbo ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.11 32QAM Modulation | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.11.1 Logical Channel Configurations Used | The logical channel configurations used for the simulations are shown in tables 73 and 74. The configurations are used to compare performance of 16QAM and 32QAM modulations with that of MCS7, MCS8 and MCS9. A simple rectangular interleaver was used. It is noted that coding schemes with coding rates greater than 0.9 hav... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.1.11.3 Throughput Performance | This subclause presents the throughput performance of the HOMTC logical channel configurations as compared to EGPRS. The configurations used for each set are given in table 77. Type II Incremental Redundancy is included.
Graphs of throughput performance and gain relative to EGPRS are shown in figures 209 and 210 respec... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.2 Comparison of Different Coding Configurations for Higher Order Modulation and Turbo Coding Schemes | For thermal noise limited scenarios, results presented in [7] and [26] indicated poor results of HOMTC as compared with EGPRS, both with and without Rx Diversity. This contradiction to performance results that are reported in subclause 8.4.4.1. This subclause reports comparative performance of configurations used in [1... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.2.1 HOMTC Coding Scheme Configurations | The logical channel configurations used are defined in table 79. Two configurations of HOMTC have been used. MCSx-T4-16QAM is the configuration used in [12], where the payload is Turbo encoded as a single block. MCSx‑T4‑16QAM_2 is the configuration used in [26], where the payload is Turbo encoded as 2 half length block... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.2.2 Performance Characterization | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.2.2.1 Interference Limited Channel | The results from the interference limited simulations are summarized in table 80 which shows the link layer performance in terms of CIR at BLER=10 %.
Table 80: Interference limited results
Modulation and Coding Scheme
EGPRS
T4-16QAM
T4-16QAM_2
C/I@ 10 % BLER
Gain (dB)
C/I@ 10 % BLER
Gain (dB)
MCS7
17.9
14.6
3.3
16.2
1.... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.2.2.2 Sensitivity Limited Channel | The results from the sensitivity limited simulations are summarized in table 216 which shows the link layer performance in terms of SNR at BLER=10 % (after 5.3 dB backoff is taken into account).
Table 81: Sensitivity limited results
Modulation and Coding Scheme
EGPRS
T4-16QAM
T4-16QAM_2
SNR@ 10 % BLER
Gain (dB)
SNR@ 10... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.2.3 Discussion | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.2.3.1 Interference Limited Channel | As can be seen from the results, the BLER performance is affected by partitioning of the block. This is particularly noticeable in the configurations equivalent to MCS7, where coding the payload as two separate blocks cuts the gain roughly in half from 3.3 dB to 1.7 dB. For the MCS8 and MCS9 equivalent cases, the loss ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.2.3.2 Noise Limited Channel | As was seen in the interference limited cases, the noise limited cases are also adversely affected by splitting the payload into 2 blocks. In particular, for the MCS7 equivalent case, the 2 block coding causes a loss in BLER performance as compared to MCS7 - this is in line with the result reported in [26]. However, if... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.3 Impact of Blind Modulation Detection | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.3.1 Blind Modulation Detection | The training sequences for 8PSK modulation in EDGE were selected to be rotated versions of the original GMSK training sequences. For GMSK, the rotation was π/2 between symbols. For 8PSK, it was selected to be 3π/8.
The principle is extended for 16QAM and 32QAM, with different rotations used as shown in Table 81a.
Table... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.3.2 Simulation Configuration | The results shown are for the simulation configuration described in Section 8.3.
Results are presented for the following channel conditions:
1. Sensitivity: TU50iFH, HT100nH
2. Co-Channel: Tu3iFH, TU50iFH |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.3.3 Performance Results | The results given in the Annex B.4 show the performance of the selected logical channels at 10% BLER. The performance is shown for the cases when the modulation is known and when blind modulation detection is used in the presence of 4 possible modulations: GMSK, 8PSK, 16QAM and 32QAM. Also shown on each graph is the mo... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.3.4 Discussion | The loss due to blind modulation is insignificant, being no more than 0.1dB in all cases. As is seen in the performance graphs, the modulation detection is far more robust than the data decoding, even when considering 4 modulations, rather than 2 in EGPRS. There are even some cases that the BLER is improved relative to... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.4.4.3.5 Conclusion | Blind modulation detection does not degrade performance of HOT. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.5 Symbol Mapping of Turbo Coded Bits | Source: Reference [24].
In this subclause, a symbol mapping method of turbo coded bits for 16-QAM modulation is introduced in order to improve the performance of such turbo coded systems. It is noted that this symbol mapping method has already been included as part of the coding chain for HS-DSCH [15]. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.5.1 Symbol mapping for 16-QAM Modulation | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.5.2.1 Concept description | In high order modulation schemes, such as 16-QAM, 32-QAM and 64-QAM, each symbol consists of bits with different reliabilities. Figure 1 shows the signal constellation diagram of 16-QAM used in 3GPP TS 25.213 [25]. A 16‑QAM symbol has a set of four consecutive data bits with (n mod 4)=0 and this set is separated into ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.5.2.2 16-QAM Symbol Mapping of Turbo Coded Bits | An example of transmit architecture for MCS-7, MCS-8 and MCS-9 in downlink is shown in figure 220. Differences from the architecture considered in the feasibility study of GERAN evolution [12, 13, 8] are in fact that a conventional convolutional encoder is replaced by a turbo encoder and an interleaver is replaced by a... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.5.2 Performance Evaluation | Performance of turbo coding with 16-QAM with a symbol mapping is evaluated on link level. Monte Carlo simulations for MCS-7-T4-16QAM, MCS-8-T4-16QAM and MCS-9-T4-16QAM are performed over AWGN and Rayleigh fading channels, respectively.
Table 82 depicts modulation and coding configuration used in simulations. Simulation... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.6 Higher Order Modulation, Turbo Codes Combined with MS Receiver Diversity | Source: Reference [26]. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.6.1 Simulation Model | The simulation model is described in [7]. The following text explains the features that are not treated in that contribution.
The antenna correlation is assumed to be zero and the gain imbalance 0 dB.
The interference model is DTS-1 for the single-antenna receiver and DTS-2 for the dual-antenna receiver. The interferer... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.6.2 Simulation Results | The simulations are carried out in interference limited and noise limited environments. In the interference limited scenario, it is assumed that the highest power levels are never reached and no back off is hence needed. In the noise limited scenario, it is assumed that the full transmit power is always used, thus impl... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.6.2.1 Interference Limited Scenario | The results from the interference limited simulations are summarized in table 85, which shows the link layer performance in terms of CIR at BLER=10 %.
Table 85: Interference limited results
MCS
Single antenna MS
Dual antenna MS
EGPRS
16QAM / TC
EGPRS
16QAM / TC
CIR@10%FER
Gain [dB]
CIR@10%FER
Gain [dB]
MCS-5
10.6
10.0
... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.6.2.2 Sensitivity Limited Scenario | The results from the sensitivity limited simulations are summarized in table 86.
Table 86: Sensitivity limited results
MCS
Conventional
MS receive diversity
EGPRS
16QAM / TC
EGPRS
16QAM / TC
SNR@10%FER
Gain [dB]
SNR@10%FER
Gain [dB]
MCS-5
14.1
13.4
0.7
9.5
8.8
0.7
MCS-6
16.1
15.6
0.6
11.3
10.6
0.7
MCS-7
20.9
21.9
-1.0
... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.6.3 Discussion | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.6.3.1 Interference Limited Scenario | As can be seen from the results, the gains from 16QAM/TC and MSRD are rather close to additive, i.e. the total gain is close to the sum of the individual gains in decibel scale. It is also interesting to note that the throughput of the MSRD/16QAM/TC configuration is close to the throughput of the dual-carrier up to a C... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.6.3.2 Noise Limited Scenario | As can be seen from figure 223, the gain from 16QAM/TC is heavily impacted by the inclusion of 5 dB back off for 16‑QAM modulation. For MCS-8 and MCS-9 there is a gain of ~10 %, and for MCS-7 a loss of similar magnitude. This example hence illustrates that the 16QAM/TC does not necessarily bring any improvement, but ca... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.7 Modified 16-ary Constellations for Higher Order Modulation and Turbo Coding Schemes | Source: Reference [32]. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.7.1 Introduction | A number of circular 16APK (Amplitude Phase Keying) constellations are compared to the square 16QAM constellation. The modulations are compared in terms of their PAPR and dynamic range, and their impact on BLER performance. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.7.2 Circular 16APK Constellations | In this subclause two circular constellations are considered. The Circular 16APK(12,4) constellation consists of two concentric circles, the inner one containing four symbols while the outer one contains the remaining 12 symbols. The Circular 16APK(8,8) constellation consists of two concentric circles, with both the in... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.7.2.1 PAPR and Dynamic Range Comparison | Table 87 shows a comparison of PAPR and dynamic range for square 16QAM, 8PSK and circular 16APK in (12,4) and (8,8) constellations with R=1.5 and R=2.0. The values of PAPR (99.99 %) and Dynamic Range (99.99 %) are shown for each modulation.
As can be seen, it is possible to make a substantial reduction in PAPR to aroun... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.7.3 Logical Channel Configurations | The channel configurations used for simulations are shown in table 88. The configurations are used to compare performance of circular 16QAM modulation with that of MCS7, MCS8 and MCS9. For each of the MCS schemes, 3 other options are considered, as taken from table 87.
Table 88: Modulation and Coding Schemes
Modulation... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.7.4 Performance Characterisation | The simulations were carried out for both a noise limited environment, and an interference limited environment. The TU3iFH channel model was used.
It was assumed that for the noise limited case, full transmit power is always used, implying that the power of 8PSK modulated blocks is backed off by 3.2 dB and the power of... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.7.4.1 Uncoded BER Performance | This subclause shows the uncoded BER performance of the 16QAM and 16APK modulations from table 87. The uncoded BER results are shown in figure 226.
It can be seen that for circular 16APK(12,4) with R=2.0, UBER performance is about 0.5 dB better compared to 16QAM - helped by the 1.2 dB advantage in backoff; with R=1.5, ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.7.4.2 BLER Performance | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.7.4.2.1 Sensitivity Limited Channel | Figure 227 shows the BLER performance for MCS7, and the three 16-ary modulated configurations carrying the same payload as MCS7. Figures 228 and 229 show the BLER performance for MCS8 and MCS9 respectively.
The results from the sensitivity limited simulations are summarized in table 90. For each configuration the table... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.7.4.2.2 Interference Limited Channel | Figure 230 shows the BLER performance for MCS7, and the three 16-ary modulated configurations carrying the same payload as MCS7. Figures 231 and 232 show the BLER performance for MCS8 and MCS9 respectively.
The results from the interference limited simulations are summarized in table 91. For each configuration the tabl... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.8 Incremental Redundancy for Higher Order Modulation and Turbo Coding Schemes (HOMTC) | Source: Reference [36]. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.8.1 Introduction | The current EGPRS capability includes an ARQ mechanism [22]. This allows for re-transmission of blocks that have failed to be decoded correctly on first or subsequent transmissions, using Link Adaptation and/or Incremental Redundancy. The receiver side signals back to the transmission side using ACK/NACK messages, and ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.8.2 EGPRS ARQ Scheme | The payload structure currently used in EGPRS is currently built around 3 "families" of MCSs as shown in 233. This format was originally proposed during the EDGE Feasibility Study (e.g.[37]). The family structure is constructed such that units of data already segmented from the Logical Link Control (LLC) layer can be t... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.8.3 Concept Proposal for ARQ with HOMTC | As we noted, in EGPRS the payloads for the higher MCSs were split to avoid the poorer performance when using higher coding rate blocks with convolutional coding. In this subclause we first examine the Turbo code structure, and note simulation data showing that for HOMTC, the performance is improved by encoding the payl... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.8.3.1 Turbo Coding Block Structure | The 3GPP RAN Turbo code is a Parallel concatenated code [15]. The structure is shown in figure 234. It is a Rate 1/3 code, with the output bits being systematic bits (the original information bits), plus 2 parity bits. There are 2 separate parity coding structures; the 1st encoder uses the information bits in the origi... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.8.3.2 RLC/MAC Operation for HOMTC | As with EGPRS, the transfer of RLC Data Blocks in the acknowledged RLC/MAC mode can be controlled by a selective type I ARQ mechanism, or by type II hybrid ARQ (incremental redundancy (IR)) mechanism, coupled with the numbering of the RLC Data Blocks within one Temporary Block Flow. The sending side (the MS or the netw... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.8.3.2.1 Type I ARQ for HOMTC with Link Adaptation | The concept for Type I ARQ is almost identical to that for EGPRS. A slight modification is made to take advantage of the improved performance of Turbo coded blocks as the source code block length increases, and also to exploit the potential for higher throughput from the Higher order 16-ary modulation. The new MCSs are... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.8.3.2.2 Type II Hybrid ARQ for HOMTC | The method proposed for Type II HARQ with HOMTC is similar to that for EGPRS, however the limitation regarding how re-transmissions may be done is removed. Currently, on re-segmenting a payload, the new payload parts are separately re-encoded and the re-segmentation is signalled. This occurs for example on re-transmiss... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.8.3.2.2.1 No Link Adaptation | Consider a simplified example of a short block, ignoring tail bits also for simplicity. Suppose we have an information block of 16 information bits, . This is encoded as:
where and are the parity bits output with the kth systematic bit. Suppose we transmit the code word using 16APK with initial code rate 1. Then, unt... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.8.3.2.2.2 With Link Adaptation | This subclause now considers an extension to the case of subclause 8.8.3.2.2.1, where in this case a change of modulation is allowed. As above, the first transmission of the block is as for Transmission 1. Suppose that LA now occurs, and the transmissions will be made using 8PSK instead of 16APK (see note).
NOTE. In th... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.8.3.2.3 Header Format | Some modification of the header format will be required to manage the signalling of a block using different modulation sequences.
Note that it is possible that the header of the first version of the BSN transmission is not decoded correctly, and then the modulation is changed. There are 2 options for how to handle this... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 8.8.3.3 USF Signalling | In most cases that there are EGPRS and HOMTC MSs operating in the same cell, it will be possible to avoid any multiplexing loss. Strategies such as allocation to different hop sequences and timeslots, and intelligent alignments of signalling USFs for EGPRS only mobiles can mitigate most of the impact. However, there wi... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.