Transcript IEEE C802.16m-09/2488r1

Data Transmission Issues of Uplink OLPC Mode 2
IEEE 802.16 Presentation Submission Template (Rev. 9)
Document Number: IEEE S802.16m-09/2488r1
Date Submitted:
2009-11-15
Source:
Rongzhen Yang, Hujun Yin
Yang-seok Choi, Apostolos Papathanassiou |
Intel Corporation
Dongcheol Kim, Wookbong Lee, HanGyu Cho
LG Electronics
Venue:.
RE: Comments on P802.16m/D2
E-mail: [email protected]
E-mail: [email protected],[email protected]
Purpose:
To be discussed and adopted into P802.16m/D2 by WG LB
Notice:
This document does not represent the agreed views of the IEEE 802.16 Working Group or any of its subgroups. It represents only the views of the participants listed in
the “Source(s)” field above. It is offered as a basis for discussion. It is not binding on the contributor(s), who reserve(s) the right to add, amend or withdraw material
contained herein.
Release:
The contributor grants a free, irrevocable license to the IEEE to incorporate material contained in this contribution, and any modifications thereof, in the creation of an
IEEE Standards publication; to copyright in the IEEE’s name any IEEE Standards publication even though it may include portions of this contribution; and at the IEEE’s
sole discretion to permit others to reproduce in whole or in part the resulting IEEE Standards publication. The contributor also acknowledges and accepts that this
contribution may be made public by IEEE 802.16.
Patent Policy:
The contributor is familiar with the IEEE-SA Patent Policy and Procedures:
<http://standards.ieee.org/guides/bylaws/sect6-7.html#6> and <http://standards.ieee.org/guides/opman/sect6.html#6.3>.
Further information is located at <http://standards.ieee.org/board/pat/pat-material.html> and <http://standards.ieee.org/board/pat >.
Background
• In current IEEE P802.16m/D2 version, 2 uplink OLPC modes defined:
P(dBm)  L  SINRTarget  NI  OffsetAMS perAMS  OffsetABS perAMS

SINRMIN ( dB )


 
10 log 10 max 10^ (
),  IoT  SINRDL        10 log 10(TNS ), OLPC Mode 1
SINRT arg et  
10

 


C / N,
OLPC Mode 2
• OLPC Mode 1 link adaption model:
– AMC (Adaptive Modulation and Coding) support the fast channel link
adaption;
– Mode 1 Power Control support the slow channel adaption to
• control the interference level
• keep the necessary transmission power in pretty stable manner;
• But in OLPC Mode 2, the key difference:
– C/N value used in Mode 2 equation is directly related to assigned MCS level
from ABS;
– Mixed Power Control and AMC in Mode 2 will bring some design issues need
to be discussed
Data Transmission Issues List of Mode 2
OLPC Mode 2 Equation and its model:
P( dBm )  L  C / N  NI  Offset _ AMS perAMS  Offset _ ABS perAMS
C/N value for each MCS level obtained from predefined table
L
P(dBm )
MS
Uplink Channel
C/N
For OLPC Mode 2, all key parameters have issues:
• C/N table:
–
–
•
Post SINR
Data block of
assigned MCS level
BS
MS has to use C/N table to determine Tx power offset. C/N table depends on both MS and BS. It is
impossible to define one C/N table for all MSs and BSs, per MS-BS table takes too much overhead;
Power adjustment tightly coupled with MCS selection, makes MCS adaption very difficult – may
have to disable fast MCS adaptation for stable power control
L: Pathloss
–
•
BS Receiver
Power is only adjusted based on whole band pathloss (L) and C/N table, it does not accurately reflect
the operating SINR of localized mode. Does not work well with frequency selective scheduling
NI: Broadcast Issue
–
16e supports NI broadcast every frame to allow stable mode 2 operation, same way in 16m will cost
much higher signaling overhead for MAC broadcast message.
The detail of issues will be discussed one by one
C/N Table Issue Summary
• AMS Tx power changes according to MCS level assigned by BS:
– Power adjustment tightly coupled with MCS selection, makes MCS
adaption very difficult – may have to disable fast MCS adaption for
stable power control (detail in next page)
– 16e Solution: “Active UL open-loop power control” to replace the role
of AMC, by the price of high PER (detail in page 6&7)
• C/N table depends on both MS and BS. It is impossible to
define one C/N table for all MSs and BSs:
– It highly depends on the base station receiver implementation and
antenna status (antenna number and correlation);
– It highly related to uplink stream number and related channel
correlation;
– 16e solution: the table broadcast mechanism in UCD (9 bytes) to
override the predefined table, but for 16m:
• It is problematic for multi-stream transmission due to channel correlation
between/among MSs;
• Signaling overhead issue: Multi stream C/N table is much bigger than
single stream in 16e: 4x16 = 64 table items for data transmission;
Issue Detail Discussion (C/N Table)
(1) AMC (Adaptive Modulation and Coding) + Mode 2
Power adjustment tightly coupled with MCS selection, makes MCS
adaption very difficult – or potentially unstable – may have to disable fast
MCS adaption for stable power control, examples:
Case 1: if bad signal quality -> AMC select the lower MCS level -> OLPC mode 2
will set lower Tx power based on C/N table -> Worse received signal quality,
then lower MCS level be selected until lowest MCS level
Case 2: if good signal quality -> AMC select the higher MCS level -> OLPC
mode 2 will set higher Tx power based on C/N table -> better received signal
quality, then higher MCS level be selected until maximum MCS level
Until now, all known simulations (details in C80216m-09_0845.ppt)
for Mode 2, AMC is disabled and 16e “Active UL open-loop power control” is enabled
OffsetMS is dependent on the HARQ feedback :
+1 dB for NAK
- “PER Target” dB for ACK
Issue Detail Discussion (C/N Table)
(2) 16e “Active UL open-loop power control” Issue
•
Because accurate C/N table is impossible to be obtained, without 16e “Active UL
open-loop power control”, the simulation of OLPC Mode 2 will get very poor
performance:
User throughput distribution for different values of IoTTarget
1
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.6
CDF
CDF
User throughput distribution for different IoTTarget
1
0.5
0.4
0.5
IoTTarget=4dB
IoTTarget=6dB
0.4
IoTTarget=8dB
IoTTarget=4dB
0.3
0.3
IoTTarget=6dB
IoTTarget=8dB
0.2
IoTTarget=12dB
•
500
User throughput (in kbps)
IoTTarget=14dB
IoTTarget=16dB
0.1
0
0
IoTTarget=12dB
0.2
IoTTarget=10dB
0.1
IoTTarget=10dB
1000
0
0
500
1000
Throughput(in kbps)
1500
Comparison: Samsung “Load Control” method for OLPC Mode 2, without “Active
UL open-loop power control” vs. with “Active UL open-loop power control”
Note: Samsung “Load Control” is the method disable AMC in its design.
Issue Detail Discussion (C/N Table)
(2) “Active UL open-loop power control” Issue – Cont.
•
Because AMC disabled for link adaptation, “Active UL open-loop power control” play the
role for link adaptation by using ACK/NCK feedback, with the price of PER increasing:
FER Distribution for different values of 
FER Distribution for different values of IoTTarget
0.35
0.5
=0
IoTTarget=4dB
0.45
IoTTarget=6dB
=0.4
IoTTarget=8dB
0.4
=0.6
IoTTarget=10dB
0.35
=0.8
0.25
IoTTarget=12dB
0.3
=1.0
IoTTarget=14dB
0.25
=1.2
0.2
IoTTarget=16dB
=1.4
FER
FER
=0.2
0.3
0.15
0.2
0.15
0.1
0.1
0.05
0.05
0
1
2
3
4
5
6
7
MCS index
8
9
10
11
Samsung “Load Control” method
With “Active UL open-loop power control”
•
•
0
1
2
3
4
5
6
7
8
9
10
11
MCS index
OLPC Mode 1 Simulation
As example to show the normal PER distribution
Simulation Setting: IEEE PCLA DG Setting (detail in backup), PER threshold for MCS
selection: 0.2
With “Active UL open-loop power control”, much higher PER than normal level: ~4 times
L: Pathloss (Channel Loss) in Mode 2
• Power is only adjusted based on whole band pathloss (L) and
C/N table, it does not accurately reflect the operating SINR of
localized mode. Does not work well with frequency selective
scheduling
Case 1: When a better channel is assigned to MS, the related higher MCS
level cannot be assigned to MS to get frequency selective gain; the MS
waste the power for low MCS level;
Case 2: When a worse channel is assigned to MS, the relative lower MCS
cannot be assigned to MS to perform link adaptation, the transmission
will fail.
NI Broadcast Issue for OLPC Mode 2
• In OLPC Mode 2, NI value play an important role for received C/N value:
– Example: When the broadcast period is long, the NI value kept in MS will be
out dated, the target C/N value will mismatch the real received C/N value;
– so for Mode 2, the NI broadcast need to be very frequent to update MS, which
requires the high signaling overhead.
• OLPC Mode 1 is insensitive to NI mismatch:
– BS will select MCS level based on received signal quality
– The BS decision of MCS is decoupled from MS Tx power
– The evaluation of Mode 1 for NI long broadcast period is shown at next page
NI Broadcast Period Evaluation for OLPC Mode 1
eITU PedB 3km/h
Performance comparison
0.07
•
0.065
– 1 frame vs. 50 frames for extreme
comparison
– eITU 3 km/h, 30 km/h and 120 km/h
Cell edge SE
0.06
0.055
•
0.05
0.045
Period=1frame
Period=50frames
0.04
0.7
NI Broadcast Period Evaluation:
0.8
0.9
1
1.1
Sector SE
1.2
1.3
The evaluation results show:
NI period 1 frame vs. 50 frames, the
performance difference for different speed
is slight
1.4
eITU VehA 30km/h
eITU VehA 120km/h
Performance comparison
Performance comparison
0.06
0.055
Period=1frame
Period=50frames
Period=1frame
Period=50frames
0.055
0.05
Cell edge SE
Cell edge SE
0.05
0.045
0.045
0.04
0.04
0.035
0.03
0.5
0.035
0.03
0.6
0.7
0.8
0.9
Sector SE
1
1.1
1.2
0.65
0.7
0.75
0.8
0.85
0.9
Sector SE
0.95
1
1.05
1.1
Recommendation
• The OLPC text in C80216m-09/2391 (its latest version) and C80216m09/2659 (its latest version) are recommended as updated uplink power
control text of IEEE 802.16m
Notes:
For OLPC text in 2 contributions are petty aligned, there are minor
difference in CLPC part at current time, the harmonization is ongoing at the
time of this supporting document uploaded.
Uplink SLS Simulation Key Parameters
(decided by PCLA DG as PC EMD)
Parameter
Value
Parameter
Value
Carrier frequency (GHz)
2.5 GHz
Site to site distance (m)
500m
System bandwidth (MHz)
10 MHz
Channel
eITU-Ped B, 3km/h
Reuse factor
1
Max power in MS (dBm)
23dBm
Frame duration
(Preamble+DL+UL)
5ms
Antenna Config
1x2 SIMO
Number of OFDM
symbols in UL Frame
18
HARQ
On (Max retrans: 4/Sync)
FFT size (tone)
1024
Target PER
0.2
Useful tone
864
Link to system mapping
RBIR
Number of LRU
48
Scheduler type
PF
LRU type
DRU
Resource Assignment
Block
8 LRU
Number of users
per sector
10
Penetration loss (dB)
20dB
CMIMO support
no
Control Overhead
0 for SE calculation (not
defined yet)
Samsung: Power Load Algorithm
to support 16e OLPC formula without changes
Load Value Threshold for IoT control:
 Κ  Κ  [dB] if IoTavg  IoTTarget

 Κ  Κ  [dB ] if IoTavg  IoTTarget
Load Value Define:
 m  PRxm ,Target[dB]  SINR DL,m [dB]
 NI [dBm ]  C / N mcs [dB]  10 log 10 ( BW )[ dB]
Decision for MCS (BS):
 m   for MCS and BW assigned to m  th MS
Tx Power Calculation (MS):
PTx ,tone[dB]  NI[dBm]  LUL[dB]  C / Nmcs[dB]  OffsetMS
16e Active OLPC:
OffsetMS is dependent on the HARQ feedback :
+1 dB for NAK
- PER target for ACK
* The rule provided by the email from Samsung
Active UL open-loop power control in 16e
Load Control for Mode 2 performance
NRT
(in dB)
Sector
throughput
(in Mbps)
Cell-edge
throughput
(in Kbps)
Sector SE
Cell-Edge
SE
K
4
2.8308
63.024
0.7549
0.0168
-110.6466
6
3.3798
75.0017
0.9013
0.02
-107.6132
8
3.6806
83.4079
0.9815
0.0222
-104.687
10
3.9038
86.5141
1.041
0.0231
-101.9896
12
4.0739
81.5662
1.0864
0.0218
-99.3977
14
4.3318
66.8206
1.1551
0.0178
-96.4913
16
4.2842
56.3248
1.1424
0.015
-93.1296
User Throughput CDF – Load Control for Mode 2
User throughput distribution for different values of IoTTarget
1
0.9
0.8
0.7
CDF
0.6
0.5
IoTTarget=4dB
IoTTarget=6dB
0.4
IoTTarget=8dB
0.3
IoTTarget=10dB
IoTTarget=12dB
0.2
IoTTarget=14dB
IoTTarget=16dB
0.1
0
0
500
1000
Throughput(in kbps)
1500
IoT CDF - Load Control for Mode 2
IoT Distribution for different values of IoTTarget
1
IoT
Target
(in dB)
IoT
mean
(dBm)
IoT Std
(dBm)
0.7
4
3.9988
0.8990
0.6
6
5.9926
1.0860
8
8.0037
1.2059
10
10.0098
1.2694
12
12.0037
1.3353
14
14.0134
1.4078
16
15.9942
1.5207
0.9
CDF
0.8
0.5
IoTTarget=4dB
IoTTarget=6dB
0.4
IoTTarget=8dB
0.3
IoTTarget=10dB
IoTTarget=12dB
0.2
IoTTarget=14dB
IoTTarget=16dB
0.1
0
0
5
10
15
IoT (in dB)
20
25
30
Performance comparison – Samsung/Intel
Performance Comparison
0.07
0.06
Celledge SE
0.05
OLPC Mode 1
OLPC Mode 2 with Load Control Algorithm
0.04
0.03
0.02
0.01
0.7
0.8
0.9
1
1.1
Sector SE
1.2
1.3
1.4
16e OLPC, C/N table and Override in UCD