3GPP TSG-RAN WG1 #64 R1-111208

Taipei, Taiwan, 21
st
- 25
th
February 2011
Source: NTT DOCOMO
Title: CoMP simulation assumptions
Agenda Item: 6.3.1.3
Document for: Discussion and decision
1 Introduction
The following simulation assumptions are defined for the CoMP evaluations. Rapporteur recommends to treat
contributions on how to address following issue.
- Exact modeling of higher latency and lower capacity
2 Simulation assumptions
The table below captures the simulation assumptions. The red text highlights the changes or additions compared to
Table A.2.3-1 in TR 36.814.
Table 1: System simulation parameters for CoMP Evaluation
Parameter VALSB, ALesValues used for evaluation
Performance metrics
Full buffer traffic: Cell capacity, Cell-edge user throughput
Non full buffer traffic: see Section A.2.1.3.2 in TR36.814
Jain Index may be provided for information.
] var[
2
2
T T
T
J
+

Deployment scenarios
1. Homogeneous network with intra-site CoMP
2. Homogeneous network with high Tx power RRHs
The central entity can coordinate 9 cells as a baseline (Reference
layout is given in Appendix)
Choose between 3, 19, 21 cells as a potential optional value
(Examples are shown in R1-110585)
Method for modelling of the out-of-coordinated area interference is to
be described
3. Heterogeneous network with low power RRHs within the macrocell
coverage
transmission/reception points created by the RRHs have different cell
IDs as the macro cell
Coordination area includes:
- Alt.1: 1 cell with N low-power nodes as starting point
- Alt.2: 3 intra-site cells with 3*N low-power nodes
FFS until RAN1#64
Benchmark is non-CoMP Rel. 10 eICIC framework with the different
cell IDRel-10 eICIC framework (association bias values FFS)
4. Network with low power RRHs within the macrocell coverage where the
transmission/reception points created by the RRHs have the same cell IDs as
the macro cell.
Coordination area includes:
- 1 cell with N low-power nodes as starting point
- 3 intra-site cells with 3*N low-power nodes
Alt.1: 1 cell with N low-power nodes
Alt.2: 3 intra-site cells with 3*N low-power nodes
FFS until RAN1#64
Benchmark is Rel-10 eICIC framework (association bias values FFS)
n on-CoMP Rel. 10 eICIC framework with the different cell ID
Baseline for association bias values,
0 dB only applied for RSRP as baseline
Any other values applied either for RSRP or RSRQ as optional
These association values are applied only for non-CoMP simulation
Simulation case
Deployment scenarios 1, 2:
Need to check the impact of shadowing correlation on the CoMP performance
Baseline:
3GPP-Case1
could be revisited depending on the impact of the shadowing correlation on the
CoMP performance
OptionalRecommended:
ITU UMi channel model (200m ISD)
Proposal change ITU channel model (200m ISD) from optional to
recommended
Deployment scenarios 3, 4:
Alt. 1
Need to check whether fast fading model is consistent with pathloss model 2
as defined in TR36.814
[ITU UMa for Macro, UMi for low power node]
Alt. 2
Need to check whether fast fading model is consistent with pathloss model 1
as defined in TR36.814
[3GPP Case 1 UMa (high spread) for Macro, UMi for low power node]
Alt. 3
[3GPP Case 1]
Baseline:
ITU UMa for Macro, UMi for low power node
UMa
- UE speed : 3km/hr
- No outdoor in-car penetration loss
UMi
- Carrier Frequency : 2GHz
- 100% UE dropped outdoors
- No outdoor to indoor penetration loss
Additional Clarifications
- ITU UMa and UMi penetration, pathloss, and shadowing generation
methodology is used for Macro to UE and Pico/RRH to UE repectively
- Do not use values in TR36.814 for pathloss, penetration and shadowing
- Revisit if significant issues are identified until RAN1 #65
Further Work to converge until RAN1 #65 for optional enhancement
- Extension with Outdoor to Indoor modeling
- Extending ITU UMa with Outdoor to Indoor Model
- Agree on default Outdoor/Indoor dropping probability
- Consider low power node antenna hight
Optional:
3GPP Case 1 Model1 for TR36.814, SCME Urban Macro 15 degrees angle
spread for fast fading (both Macro-to-UE and low power node-to-UE)
Further study is needed with aim to converge on a single model as a baseline
for comparison at RAN1 #64
Note: fast fading model for Heterogeneous based on spatial channel model
Number of low power node per
macro-cell
From TR36.814:
Configuration #4b with N low power nodes per macro cell
Configuration #1 with N low power nodes per macro cell
Baseline:
N = 4
Optional:
N = 1, 2, 10
High power RRH Tx power (Ptotal) 46/49dBm in a 10, 20MHz carrier
Low power node TX power (Ptotal)
30 dBm and 37 dBm for both FDD and TDD in 10MHz carrier, with higher
priority for 30 dBm
Number of UEs per cell
Full buffer traffic model: 10 for Homogeneous networks; dependent on the
targeted resource utilization for non-full-buffer traffic model.
Same as TR 36.814 for Heterogeneous networks
System bandwidth 10 MHz, 20MHz
Possible transmission schemes in
DL
SU-MIMO
MU-MIMO
SU-MIMO with intra-eNB CS/CB
MU-MIMO with intra-eNB CS/CB
SU-MIMO with intra-eNB JP-CoMP
MU-MIMO with intra-eNB JP-CoMP
Legacy UE impact
How to assess the legacy impact is FFS
Companies are encouraged to bring contributions how to address this issue
Impairments modelling
The following impairments are modelled. The modelling needs to be described.
- impairments of JP-CoMP
- Collision between CRS and PDSCH
- Different control regions
[- time/frequency synchronization impairments - FFS]
- Modeling of actual propagation delay differences depending on UE location
and timing error [FFS] would need to be included as a multipath effect
Baseline timing error is 0us; recommended to provide results for additional
case with timing error modelled as uniform distribution between +/- x us [revisit
to decided exact value of x on Friday].
Methods that offset the propagation delay are not precluded.
X =
Alt.1: 0.5
Ericsson, ST-Ericsson, Samsung, LG, Intel, Qualcomm
Alt. 2: 1.5
ALU, ASB
- Frequency offset sensitivity analysys is recommended
- [- PDCCH overhead/capacity should be taken into account in the comparison
- FFS]Analysys of PDCCH and SRS overhead/capacity is recommended
Companies are encouraged to bring contributioins how to address these
issues
Network synchronization Synchronized
Number of antennas at transmission
point
Macro and high Tx power RRH: 1, 2, 4, 8 (2 and 4 antennas are baseline for
FDD, 2 and 8 antennas are baseline for TDD)
Low power node: 1, 2, 4 (2 and 4 antennas are baseline).
Inclusion of 2 or 4 antennas as baseline may be revisited
Baseline vValues for combinations (number of antennas at macro node,
number of antennas at low-power node) are FFS until RAN1#64. Candidate
values (to be further down-selected) are:
(2, 2), (4, 4), (4, 2), (2, 4), (8, 2)
(2, 2), (4, 4) for FDD, (2, 2), (8, 2) for TDD as baseline, (2, 4) for FDD, (4, 2) for
TDD as optional
Number of antennas at UE 2, 4, with higher priority for 2 antennas.
Antenna configuration For macro eNB and high power RRH, In priority order for each number of
antennas:
2 Tx antennas
1. 1 column, cross-polarized: X
2. 2 columns, closely-spaced vertically-polarized: | |
4 Tx antennas
1. 2 columns, cross-polarized on each column, closely-spaced: X X
2. 2 columns, cross-polarized on each column, widely-spaced: X X
3. 4 columns, vertically-polarized, closely-spaced: | | | |
8 Tx antennas
1. 4 columns, cross-polarized on each column, closely-spaced: X X X X
2. 4 columns, cross-polarized on each column, 2 widely-spaced sets of
closely-spaced columns: X X X X
3. 8 columns, vertically-polarized, closely-spaced: | | | | | | | |
For low power node
1 Tx antenna: vertically-polarized
2 Tx antennas:
cross-polarized: X
4 Tx antennas:
1. 0.5 -spaced cross-polarized: X X
2. 0.5 -spaced vertically-polarized: | | | |
Array orientation needs to be defined (e.g., random for 4 antennasTx)
When cross-polarized antenna configuration is applied to transmission point, it
is also applied to the receiverUE. When co-polarized antenna configuration is
applied to transmission point, it is also applied to the receiverUE.
For scenarios 3 and 4 and more that 1 antenna at the low power node, when
cross-polarized antenna configuration is applied at the macro, it is also applied
at the low power node; when co-polarized antenna configuration is applied at
the macro, it is also applied at the low power node
Antenna pattern
For macro eNB and high-power RRH:
3D as baseline
2D as additional
Follow Annex A 2.1.1.1 Table A.2.1.1-2 in TR36.814
For low-power node:
2D as baseline
3D as optional
Horizontal plane: omnidirectional
Vertical plane:
( )
1
1
]
1

,
_

v
dB
etilt
V
SLA A , 12 min
2
3

dB 3
= 40 degrees, SLAv = 20 dB
eNB Antenna tilt
For macro eNB and high-power RRH: Different downtilt values may be
evaluated.
For low-power node: 0 or 10 degrees
Feedback scheme (e.g.
CQI/PMI/RI/SRS)
Overhead is to be reported
The following benchmarks may be used:
- Rel-10 feedback (baseline) (with overhead as close as possible to
overhead of CoMP scheme)
- If CoMP scheme requires more feedback overhead than is possible in
Rel-10, benchmark is a single-transmission/reception-point scheme (to
be fully described) with same feedback overhead as CoMP scheme
Baseline:
Per-transmission-point feedback is implicit
Inter-cell information feedback mechanism to be described
Channel estimation
Non-ideal, based on CSI-RS.
Clarify in detail the following on CoMP evaluation:
- CSI knowledge of eNB
- Feedback scheme and/or UL sounding scheme
[- Feedback error model - FFS]
- Accuracy of CSI
. Quantization error
. Channel estimation error based on CSI-RS
. Channel estimation error based on SRS
Revisit CSI-RS and SRS estimation error modeling before Friday
- Try to capture common mis-calibration modelling at RAN1 #65 for TDD
Until RAN1 #65, no antennas mis-calibration for UL-DL channel
reciprocity as mandatory and antennas mis-calibration for UL-DL channel
reciprocity as recommended for TDD- FFS]
[- Antennas mis-calibration for DL Tx antennas with 0.5 spacing as-
FFS]optional for FDD
- Channel estimation error for demodulation
- Any channel reciprocity modelling to be described.
- Any antenna calibration mechanism to be described.
UE receiver
Channel estimation error based on DM-RS should be modelled
Mandatory: MMSE receiver model option1 in R1-110586
Recommended: Advanced MMSE receiver and/or IRC receiver
Details are described in R1-110586
Companies should specify the modelling of Advanced MMSE/IRC
FFS for phase 2
DL overhead assumption Should be clarified for each transmission scheme, taking into account CSI-RS
and PDSCH muting overhead, as well as PDCCH overhead
corresponding to scheduling
Placing of UEs
Uniform distribution for homogeneous networks
For heterogeneous networks, placement according to the configuration.
Traffic model
Full buffer
Non-full-buffer according to Section A.2.1.3.1 in TR36.814, with the following
modifications:
Model 1 with file size of 2 Mbytes is preferred, however Model 1 with file
size of 0.5 Mbytes and Model 2 with file size of 0.5 Mbytes can be
evaluated instead
Simulations are run for various (for model 1) or K (for model 2) to find
performance metrics covering at least the HM-NCT values (See A.2.1.3.2)
that lead to [10 - 70]% of RU (See A.2.1.3.2) in non-CoMP SU-MIMO.
Note revisit document R1-110895
For full buffer traffic model and non-full buffer traffic model 2
- Fix the total number of users, Nusers, dropped within each macro
geographical area, where Nusers is 30 or 60 in fading scenarios and 60 in
non-fading scenarios.
- Randomly and uniformly drop the configured number of low power
nodes, N, within each macro geographical area (the same number N for
every macro geographical area, where N may take values from {1, 2, 4,
10}).
- Randomly and uniformly drop Nusers_lpn users within a 40 m radius of
each low power node, where ]
/N N P N
users
hotspot
users_lpn

with P
hotspot


defined in Table A.2.1.1.2-5, where P
hotspot
is the fraction of all hotspot
users over the total number of users in the network.
- Randomly and uniformly drop the remaining users, Nusers - Nusers_lpn*N, to
the entire macro geographical area of the given macro cell (including
the low power node user dropping area).
For non-full buffer traffic model 1
- Randomly and uniformly drop the configured number of low power
nodes, N, within each macro geographical area (the same number N for
every macro geographical area, where N may take values from {1, 2, 4,
10}).
- Generate users based on traffic load. Chose the geographical area in
which user will be dropped randomly and with probability of P
hotspot
for the low
power node geographical area, and 1- P
hotspot


for the the entire macro cell
geographical area (including the low power node user dropping area) .
Backhaul assumptions
For deployment scenarios 1, 2 and 3:
Step 1: [point-to-point fiber, zero] latency and infinite capacity
Step 2: higher latency and limited capacity for scenarios 2 and 3
Exact modeling of higher latency and lower capacity is FFSThe latency
values used for CoMP evaluation are {0ms,2ms,10ms}
The latency value here refers to the one-way delay incurred when a
message is conveyed from one node to another
Each company indicate the capacity requirement associated with the proposed
scheme
Link adaptation Non-ideal; details to be provided
Others Note UL assumptions isto be revisited inbased on R1-111176.0803
Appendix
Figure A1- Reference CoMP Coordination Cell Layout for Scenario 2