Transcript IEEE C802.16m-09/0997

MCS signaling for IEEE802.16m
IEEE 802.16 Presentation Submission Template (Rev. 9)
Document Number:
IEEE C802.16m-09/0997
Date Submitted:
2009-04-27
Source:
Chiwoo Lim , Seunghoon Choi, Sung-Eun Park, Songnam Hong, Jaeweon Cho, Jaehee Cho, Heewon Kang, Hokyu Choi
E-mail: {chiwoo.lim, seunghoon.choi, se.park, sn7955.hong, jaeweon.cho, jaehee1.cho, hkang, choihk} @samsung.com
Samsung Electronics, Co., Ltd.
416 Maetan-3, Suwon, 443-770, Korea
Venue:
Category: AWD comments / Area: Chapter 15.3.12 (Channel coding HARQ-PHY)
“Comments on AWD 15.3.12 Channel coding HARQ-PHY”
Base Contribution:
None
Purpose:
To be discussed and adopted by TGm for 802.16m amendment
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 >.
MCS signaling for IEEE802.16m
Chiwoo Lim, Seunghoon Choi, Sung-Eun Park, Songnam Hong,
Jaeweon Cho, Jaehee Cho, Heewon Kang, Hokyu Choi
Samsung Electronics Co., Ltd.
2
About This Contribution
• Goal and scope of this contribution
– Show the performance comparison with 4-bit and 5-bit MCS.
– Show the weak relationship between padding overhead and MCS
resolution, especially when 4 bits and 5 bits of MCS resolution are
compared.
– Propose schemes for reducing A-MAP overhead for 802.16m
3
Performance comparison with
4-bit and 5-bit MCS
MCS levels
• 4bits (16 levels) vs. 5bits (32 levels)
• Mathematical calculation shows the 5.7% gain in 5bits.
• But, we can’t consider the scheduling effect and CQI mismatch (due to
delay).
• So, we require the SLS.
• Simulation conditions based on the current 16m EMD.
• Some specific conditions are as following :
–
–
–
–
5
Cell user : 10 , Scheduling user : 3
Scheduling : Proportional Fair
Cell radius : 1.5km
Variable factors : MCS levels, CQI report period, velocity (channel)
LLS results
•
•
•
•
6
In AWGN
Target FER : 10e-1
SNR range : -10dB ~ 21dB (Equi-distance SNR)
32 levels case. (16 levels case use a half.)
SLS results
• System Throughput (in Ped_B 3km/h)
CQI report
period (frame)
32 levels
(Mbps)
16 levels
(Mbps)
Gains
(in 32 levels)
1
8.18 (97.8%)*
7.93 (98.4%)
+3.2%
4
6.77 (79.1%)
6.75 (83.5%)
+0.3%
8
6.44 (78.4%)
6.49 (82.6%)
-0.8%
* means the success probability in first transmission
7
SLS results
• System Throughput (CQI report period : 1 frame)
Channel
32 levels
(Mbps)
16 levels
(Mbps)
Gains
(in 32 levels)
Ped_B (3km/h)
8.18 (97.8%)*
7.93 (98.4%)
+3.2%
Ped_B (30km/h)
5.93 (80.1%)
5.92 (84.4%)
+0.2%
Veh_A (120km/h)
5.51 (64.8%)
5.50 (70.1%)
+0.2%
* means the success probability in first transmission
Mixed Channel : Ped_B 3 (60%), Ped_B 30 (30%), Veh_A 120 (10%)
Gains(in 5bits) : 2%
8
SLS results
• System Throughput (CQI report period : 4 frame)
Channel
32 levels
(Mbps)
16 levels
(Mbps)
Gains
(in 32 levels)
Ped_B (3km/h)
6.77 (79.1%)
6.75 (83.5%)
+0.3%
Ped_B (30km/h)
5.45 (71.4%)
5.46 (76.9%)
-0.2%
Veh_A (120km/h)
5.44 (67.8%)
5.41 (72.9%)
+0.6%
* means the success probability in first transmission
Mixed Channel : Ped_B 3 (60%), Ped_B 30 (30%), Veh_A 120 (10%)
Gains(in 5bits) : 0.18%
9
SLS results
• System Throughput (CQI report period : 4 frame)
Channel
Equi dist.
16e ext.
Gains
Ped_B (3km/h)
6.75 (83.5%)
6.72(83.0%)
0.4%
Ped_B (30km/h)
5.46 (76.9%)
5.34(76.2%)
2.6%
Veh_A (120km/h)
5.41 (72.9%)
5.32(72.2%)
1.7%
* means the success probability in first transmission
Mixed Channel : Ped_B 3 (60%), Ped_B 30 (30%), Veh_A 120 (10%)
Gains : 1.2%
10
Reasoning for result
•
•
•
•
32 levels cause more HARQ retransmission.
The probability of increasing retransmission is 1/32
So, there are more(3.1%) retransmissions in 32 levels.
In real simulation,
used SNR range : -5dB ~ 21dB, so the value is 1/26 (= 3.8%)
Increasing retransmission region
11
Padding overhead analysis
for MCS resolution
12
Padding Overhead
• Padding Overhead can be determined by the following 3 Factors
– Number of MCS Levels (MCS Resolution)
– Size of Resource Unit (Resource Resolution)
– Number of Nep (Nep Resolution)
• The point of our analysis is only to compare 5bits MCS with 4bits MCS in
terms of padding overhead. The followings are assumed.
– MCS Levels should be determined by following the Equi-Dist SNR
– The Size of Resource Unit is fixed by 96
– Nep has one byte resolution (Let’s ignore the restriction of multiple of 7 for
convenient)
13
Padding Overhead Analysis
• [Example 1] Computing Padding Overhead
– For the given CQI and PDU, MCS Index and RU Size is determined by
considering the following two conditions
[CQI = MCS Index 4, PDU = 15 bytes]
• The MCS Index should be chosen such that MCS Index ≤ 4
• The RU should be chosen such that Nep ≥ PDU
– The MCS Index and RU Size is determined such that padding overhead is
minimized.
• In this example, MCS Index 3 and RU = 2,
• The padding overhead is (19-15)/19 * 100 = 21%
14
RU / MCS
0
1
2
3
4
1
-
-
6
10
13
2
6
9
13
19
26
3
9
13
20
29
38
Padding Overhead Analysis
• [Example 2] Comparison of Padding Overhead, 4bits Vs. 5 bits
(Case 1)
CQI = MCS Index 4 (in 4 bit MCS case), PDU = 20 bytes
• 4Bits MCS : MCS Index 4 and RU = 2,
Padding Overhead = (26-20)/26*100=23.1%
• 5bits MCS : MCS Index 7 and RU = 2,
Padding Overhead = (22-20)/22*100=9.1%
Big difference but special case
(Case 2)
CQI = MCS Index 4 (in 4 bit MCS case), PDU = 80 bytes
• 4Bits MCS : MCS Index 4 and RU = 7,
Padding Overhead = (89-80)/89*100=10.1%
• 5bits MCS : MCS Index 8 and RU = 7,
Padding Overhead = (89-80)/89*100=10.1%
No difference and almost case
RU
4bit
MCS
index
5bit
MCS
index
3
4
6
7
8
1
10
11
13
2
19
22
26
3
29
33
38
4
38
44
51
5
48
54
64
6
57
65
76
7
67
76
89
8
76
87
102
9
86
98
114
10
95
108
127
Padding Overhead Analysis
• Case 1 (blue circle)
– (a) value is larger than (b)
– There is a gain of padding
overhead reduction
– Only a few amount of RU case
• Case 2 (red circle)
– (c) value is smaller than or equal to
(d)
– There is no gain of padding
overhead reduction
– Almost case
16
RU
4bit
MCS
index
5bit
MCS
index
3
4
6
7
8
1
10
11
13
2
19
22
26
3
29
33
38(b)
4
38
44(a)
51
5
48
54
64
6
57
65
76
7
67
76
89(d)
8
76
87(c)
102
9
86
98
114
10
95
108
127
Analysis by Calculation
• For the given MCS Tables in Appendix, we derived the padding
overhead with the same method in Example 1
– 4bits MCS : padding overhead 2.9%
– 5bits MCS : padding overhead 2.7%
* This is because Case 1 is special and few event and Case 2 is almost event.
• The padding overhead reduction gain in 5 bits MCS is 0.2%.
– The padding overhead reduction gain doesn’t directly have an effect to
throughput but coding gain. (Show the Appendix 3)
• Our analysis shows that 5bits MCS can transmit 0.2% more parity
bits than 4bits MCS, which can be negligible gain.
• We do not have to consider padding overhead when determining the
number of bits for MCS (4bits or 5bits)
Appendix 1
4 bits MCS Table
in IEEE802.16m-09-0510r2
MCS index
Modulation
Code rate
‘0000’
QPSK
31/256
‘0001’
QPSK
48/256
‘0010’
QPSK
71/256
‘0011’
QPSK
101/256
‘0100’
QPSK
135/256
‘0101’
QPSK
171/256
‘0110’
16QAM
102/256
‘0111’
16QAM
128/256
‘1000’
16QAM
155/256
‘1001’
16QAM
184/256
‘1010’
64QAM
135/256
‘1011’
64QAM
157/256
‘1100’
64QAM
181/256
‘1101’
64QAM
205/256
‘1110’
64QAM
225/256
‘1111’
64QAM
237/256
Appendix 2
5 bits MCS Table
19
MCS Index
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Modulation
QPSK
QPSK
QPSK
QPSK
QPSK
QPSK
QPSK
QPSK
QPSK
QPSK
QPSK
16QAM
16QAM
16QAM
16QAM
16QAM
16QAM
16QAM
16QAM
16QAM
64QAM
64QAM
64QAM
64QAM
64QAM
64QAM
64QAM
64QAM
64QAM
64QAM
64QAM
64QAM
Code Rate
31/256
38/256
48/256
58/256
71/256
84/256
101/256
115/256
135/256
150/256
171/256
92/256
102/256
111/256
128/256
138/256
155/256
167/256
184/256
192/256
135/256
148/256
157/256
168/256
181/256
195/256
205/256
217/256
225/256
232/256
237/256
242/256
Appendix 3
The Impact of Padding Overhead reduction (in 5bit MCS)
- The 5bits MCS can transmit the PDU with lower code rate than 4bits MCS
as given in the following figure (Case 1)
4 bits MCS
5 bits MCS
20
A-MAP information related HARQ in 802.16m
• A-MAP information can include the followings:
–
–
–
–
–
–
–
Resource allocation information
MCS index
AI-SN
SPID (SubPacket ID)
CoRe version
ACID (HARQ channel ID)
Etc.
• In conventional MAP transmission, the above information are
transmitted at every transmission.
21
Considerations on A-MAP information
• There are three types of information in A-MAP.
– Information necessary for every transmission
• Resource allocation information (except persistent allocation)
• AI-SN
• ACID
– Information only necessary for initial transmission
• The information for calculation of burst size (Nep)
• For example, modulation and code rate
– Information only necessary for retransmission
• The information can be fixed in initial transmission without any penalty.
• SPID (SPID = 0 in initial transmission)
• CoRe version (CoRe version = 0 in initial transmission)
22
Problem of current MCS table
• Firstly, we should consider supporting adaptive HARQ.
– Assumptions :
• Adaptive HARQ is adapted.
• Same MCS table is adapted in initial and retransmission.
– Problem :
• MCS index in retransmission only give the modulation order information.
• In this case, we use the 4 bits for indicating only modulation order.
23
Problem of current MCS table
• Secondly, we should consider initial A-MAP loss problem.
– If we assumes the adaptive HARQ, we may change the modulation
order and resources in retransmission.
• In one HARQ transmission, there is no change of the Nep.
• So, if AMS don’s receive the initial A-MAP, AMS can’t succeed the decoding
because A-MAP information in retransmission can’t give the Nep information.
• If AMS don’t receive the initial A-MAP, AMS don’t send the ACK/NACK and it
induces the residual error with probability of initial MAP loss.
24
New schemes for 802.16m
• Proposed scheme multiplexes different types of information for
A-MAP overhead reduction.
– Multiplexing information are MCS index and SPID and CoRe version.
– MCS index (Modulation and Code rate)
• The information for calculation of burst size is very important, but burst size should not
be changed during one HARQ transmission. So, this information is only necessary for
initial transmission.
• Modulation can be changed under adaptive HARQ.
• Code rate is used for calculating burst size in initial transmission, but doesn’t have any
information in retransmission.
– SPID and CoRe version
• SPID and CoRe version are the information that can be fixed in initial transmission
without any penalty.
• CoRe may not operate in case that modulation is changed in retransmission.
25
New schemes for 802.16m
• In proposed scheme, MCS index in MCS table have different
information for initial and retransmission.
– Initial transmission : MCS index indicates modulation and code rate.
– Retransmission : MCS index indicates modulation, SPID and CoRe
version.
• Proposed scheme assumes the 4-bit MCS index, 2-bit SPID
and 1 bit CoRe version.
– Additional bits for SPID and CoRe version are possible if the gain is
guaranteed.
– Even if additional bits are used, the bit in current table indicates a part
of whole bits and other bits can be included in other A-MAP field.
26
New schemes for 802.16m
• In proposed scheme, the bits for signaling SPID and CoRe
version can be removed or reduced.
• There is no drawback for conventional operation.
– Only one restriction is that there is no CoRe operation in case that
modulation is changed in retransmission. But, this restriction is
reasonable.
• In MCS table for retransmission
– “+” and “-” are utilized when modulation at retransmission is not same
to that at initial transmission as follows.
• When modulation is QPSK at initial transmission, “-” and “+” represent
64QAM and 16QAM at retransmission, respectively.
• When modulation is 16QAM at initial transmission, “-” and “+” represent
QPSK and 64QAM at retransmission, respectively.
• When modulation is 64QAM at initial transmission, “-” and “+” represent
16QAM and QPSK at retransmission, respectively.
27
New schemes for 802.16m
• For solving the initial A-MAP loss problem, we introduce the
Null detection and the 1-bit MTI (MCS Table Indicator).
• Through the Null detection, ABS can know the initial A-MAP
loss, and ABS can resend the initial A-MAP using MTI.
• MTI can indicate the initial A-MAP to AMS when initial AMAP is lost.
• Null detection can be implemented by applying a few powers
to ACK/NACK channel for initial transmission.
• In our analysis, about 1.8dB power increasing in ACK/NACK
channel for initial transmission can solve the initial A-MAP
loss problem.
28
New schemes for 802.16m
MCS for initial transmission
(MTI = 0)
MCS for retransmission
(MTI =1)
SPID
CoRe
version
0
0
0
1
0
1
1
0
1
1
2
0
2
1
6
3
0
128/256
7
3
1
16QAM
155/256
8
-
0
0
‘1001’
16QAM
184/256
9
-
1
0
‘1010’
64QAM
135/256
10
-
2
0
‘1011’
64QAM
157/256
11
-
3
0
‘1100’
64QAM
181/256
12
+
0
0
‘1101’
64QAM
205/256
13
+
1
0
‘1110’
64QAM
225/256
14
+
2
0
‘1111’
64QAM
237/256
15
+
3
0
MCS
index
Modulation
Code rate
‘0000’
QPSK
31/256
‘0001’
QPSK
48/256
‘0010’
QPSK
71/256
‘0011’
QPSK
101/256
‘0100’
QPSK
135/256
‘0101’
QPSK
171/256
‘0110’
16QAM
102/256
‘0111’
16QAM
‘1000’
MCS
index
2
3
4
5
Modulation
Modulation
is same as
initial
transmission
.
Modulation at
initial
transmission.
-
+
QPSK
64QAM
16QAM
16QAM
QPSK
64QAM
64QAM
16QAM
QPSK
• MCS for initial transmission is just an example, so modulation and coderate details are not the
scope of this proposal
Example
• In initial transmission,
– MCS index ‘7’ is indicated : Modulation is 16QAM, SPID is 0 and
CoRe version is 0.
• In retransmission,
– MCS index ‘2’ is indicated : Modulation is same as initial transmission
(16QAM), SPID is 1 and CoRe version is 0.
– MCS index ‘3’ is indicated : Modulation is same as initial transmission
(16QAM), SPID is 1 and CoRe version is 1.
– MCS index ‘9’ is indicated : Modulation is changed by QPSK, SPID is
1 and CoRe version is 0.
– MCS index ’13’ is indicated : Modulation is changed by 64QAM, SPID
is 1 and CoRe version is 0.
Text Proposal to 802.16m amendment
Replace the Table 700 by the following text at the section 15.3.12.1.3 Burst partition
-------------------------------------- Start of Proposed Text --------------------------------------15.3.12.1. Channel Coding
15.3.12.1.3. Burst partition
To determine the modulation and code rate for current transmission, the AMS shall
read the 4-bit ’MCS index’ field in A-MAP.
In downlink, the MCS index represents a different information according to MTI.
For MTI=0 (initial transmission), it denotes the modulation and code rate as shown
in Table xxx, and for MTI=1 (retransmission), it denotes the modulation, CoRe
version, and SPID as shown in Table yyy.
In uplink, the MCS index represents the modulation and code rate as shown in
Table xxx.
In Table yyy, “+” and “-” are utilized when modulation at retransmission is not
same to that at initial transmission as follows.
- When modulation is QPSK at initial transmission, “-” and “+” represent
64QAM and 16QAM at retransmission, respectively.
- When modulation is 16QAM at initial transmission, “-” and “+” represent
QPSK and 64QAM at retransmission, respectively.
- When modulation is 64QAM at initial transmission, “-” and “+” represent
16QAM and QPSK at retransmission, respectively.
31
Text Proposal to 802.16m amendment
Table xxx ― MCS table for downlink (MTI=0) and uplink data channel
32
MCS
index
Modulation
Code rate
‘0000’
QPSK
31/256
‘0001’
QPSK
48/256
‘0010’
QPSK
71/256
‘0011’
QPSK
101/256
‘0100’
QPSK
135/256
‘0101’
QPSK
171/256
‘0110’
16QAM
102/256
‘0111’
16QAM
128/256
‘1000’
16QAM
155/256
‘1001’
16QAM
184/256
‘1010’
64QAM
135/256
‘1011’
64QAM
157/256
‘1100’
64QAM
181/256
‘1101’
64QAM
205/256
‘1110’
64QAM
225/256
‘1111’
64QAM
237/256
Text Proposal to 802.16m amendment
Table yyy ― MCS table for downlink (MTI=1) data channel
MCS
index
SPID
CoRe
version
0
0
0
1
0
1
1
0
1
1
2
0
2
1
6
3
0
7
3
1
2
3
4
5
Modulation
Modulation
is the same
as that of
initial
transmission
.
8
-
0
0
9
-
1
0
10
-
2
0
11
-
3
0
12
+
0
0
13
+
1
0
14
+
2
0
15
+
3
0
-------------------------------------- End of Proposed Text --------------------------------------33