IEEE 802.16m-09/0226r3
Project
IEEE 802.16 Broadband Wireless Access Working Group <http://ieee802.org/16>
Title
Proposed Text of Channel Coding and HARQ for the IEEE 802.16m Amendment
Date
Submitted
2009-01-15
Source(s)
Yu-Chuan Fang, 1Zheng Yan-Xiu,
Ren-Jr Chen, Chang-Lan Tsai,
Chung-Lien Ho, Richard Li, Chien-Yu Kao,
Cheng-Ming Chen, 3Yu-Tao Hsieh,
Jen-Yuan Hsu, Pang-An Ting,
ITRI
Wern-Ho sheen, NCTU/ITRI
Voice:
1
+886-3-5914715
2
+886-3-5917596
3
+886-3-5914854
E-mail:
1
zhengyanxiu@itri.org.tw
2
richard929@itri.org.tw
3
ythsieh@itri.org.tw
Re
IEEE 802.16m-08/053r1, “Call for Comments and Contributions on Project 802.16m
Amendment Working Document”.
Target topic: “11.13 Channel Coding and HARQ”.
Abstract
The contribution proposes the text for HARQ bit-rearrangement
Purpose
To be discussed and adopted by TGm for the 802.16m amendment.
Notice
Release
Patent
Policy
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.
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.
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>.
Proposed Text of Channel Coding and HARQ
for the IEEE 802.16m Amendment
Yu-Chuan Fang, Zheng Yan-Xiu, Ren-Jr Chen, Chang-Lan Tsai, Chung-Lien Ho, Richard Li,
Chien-Yu Kao, Cheng-Ming Chen Yu-Tao Hsieh, Jen-Yuan Hsu, Pang-An Ting,
ITRI
Wern-Ho sheen, NCTU/ITRI
ITRI
1
IEEE 802.16m-09/0226r3
1. Introduction
The contribution proposes the text of bit re-arrangement to be included in the 802.16m amendment. The
proposed text is developed so that it can be readily combined with IEEE P802.16 Rev2/D7 [1], it is compliant to
the 802.16m SRD [2] and it follows the style and format guidelines in [3]. In section 11.13.2.2 of [4], bit
re-arrangement includes a bit-level interleaver and an inverter. Referring to [5,6,7,8], a scheme of a simple bit
circular shifter with bit-priority mapping and a bit inverter is proposed for the bit re-arrangement. For chase
combining (CC) mode, the bit circular shifter and inverter change the bit positions and bit reliabilities of a
subpacket transmitted previously for retransmissions. Besides, the bit-priority mapping can improve the error
rate of initial transmission and overall throughput. For incremental redundancy (IR) mode, the bit-priority
mapping can protect relative significant bits in different redundancy versions more robustly and bit circular
shifter can interleave the selected coded bits to achieve more diversity gain.
2. Motivation
In the legacy transmitter the CTC encoder encodes NEP bits to generate a codeword of 3×NEP coded bits. After
the interleaver and the bit selection including puncturing or repetition, a subpacket is formed and then fed into
the modulator. The bit selection may product several subpackets as redundancy versions for retransmission.
While receiving a NACK, the transmitter retransmits the subpacket or another subpacket of the codeword to
enhance the decoding ability. However, it does not satisfy the requirement of better system performance and
higher throughput. The bit rearrangement can help arrange coded bits by several BitRe versions to obtain
diversity gain. Fig. 1 illustrates the transmitter with the bit rearrangement block. Through the bit selection and
bit rearrangement the transmitter may make several versions of subpackets.
data
CRC NEP bits
adder
CTC
encoder
Interleaver
Bit selection
Bit Rearrangement
Modulation
Tx
ACK/NACK
Fig. 1
An illustration of the transmitter with bit rearrangement
In order to improve the system performance, the diversity gain deserves attention for the bit re-arrangement. We
introduce several familiar diversities.

Constellation diversity:
The bits have different reliabilities on some constellations. Taking 16-QAM for example, four bits, b3b2b1b0 ,
are mapping into a position on the constellation to form a 16-QAM symbol. b3 and b2 modulate the
in-phase carrier and b1 and b0 modulate the quadrature carrier. Bits b3 and b1 respectively modulate the
phases of the in-phase and quadrature carriers. In particular, the phase is 0 when b3 or b1 is 0 and 180
when b3 or b1 is 1. Bits b2 and b0 respectively modulate the amplitudes of the in-phase and quadrature
carriers, where the amplitude is greater when b2 or b0 is 1 than otherwise. Thus, b3 and b1 have higher
reliabilities than b2 and b0 . The diversity occurs due to different reliabilities.
2
IEEE 802.16m-09/0226r3

Frequency diversity:
After the modulation, the modulation symbols may be allocated into the resource units with different
subcarrier indices. Over the fading channel, the bits at different subcarrier indices should suffer from
different fading, as generates diversity.

Antenna diversity:
While multiple antennas are used, the coded bits are separated into several streams. Different channels may
cause different effects on the bits.
The bit rearrangement should have two properties to take advantage of diversities, which are introduced as
follows.

Bit-priority mapping:
The coded bits may be given different priority according to their relative significance, such as systematic or
parity bits. More significant bits should be protected more robustly to improve the decoding performance.
Therefore, more significant bits may be mapped into more reliable positions on the constellation, or mapped
into streams passing better channels. The distribution of priority may change for multiple transmissions of a
codeword. Especially, the bit-priority mapping can provide benefit at the initial transmission.

Averaging reliability:
If some specific bits are always given the highest priority at multiple retransmissions, the performance
improvement would degrade. Thus, we let the bits exchange the priorities or reliabilities, i.e. the bit with
higher/lower reliability at previous transmission is given lower/higher reliability at current transmission.
After several transmissions, the bit reliability seems to be averaged.
In the next section we propose a bit re-arrangement method which utilizes the bit-level interleaver and bit
inverter to achieve the properties and advantage mentioned above.
3. Proposed bit re-arrangement method
The proposed bit rearrangement method comprises the bit-level interleaver and bit inverter. For convenience of
introduction the bit-level interleaver is separated into two blocks, bit-priority mapper and circular shifter. Thus,
the block diagram shown in Fig. 2 has three function blocks, bit-priority mapper, circular shifter and bit inverter.
The bit rearrangement for CC-HARQ and IR-HARQ modes are respectively introduced in two sections and its
operations are also described in the sub-clauses.
Bit rearrangement
Bit-level interleaver
Bit selection
ak
Bit-priority b k
Mapper
Fig. 2
Circular
shifter
ck
bit inverter
An illustration of the bit rearrangement method
The parameters are defined as follows.
3
dk
Modulator
IEEE 802.16m-09/0226r3
k : the subpacket index
NTX : the number of transmission
N EP : the number of bits in the encoding packet (before encoding)
Lk : the number of coded bits for the k -th subpacket
mk : the modulation order for the k -th subpacket (2 for QPSK, 4 for 16-QAM and 6 for 64-QAM)
N CMS  Lk mk : the number of coded modulation symbols per subpacket
N str : the number of streams
N DT : the number of data tones per RU (resource units)
N PT : the number of pilot tones per RU
NT : the number of tones per RU (=N DT  N PT )
TS : the symbol time occupied per RU (  6)
Lscd : the fundamental length of subcarriers for data, Lscd
3.1
3.1.1
17, if N str  1

 16, if N str  2
14, if N  4
str

CC-HARQ mode
Bit-priority mapper
After the bit selection, the coded bits are divided into two groups for 16-QAM and three groups for 64-QAM,
and then the more/less significant bits are mapped to more/less reliable positions on the constellation as shown
in Fig. 3. The coded bits of the subpacket at the initial transmission comprises systematic and parity parts,
which are respectively regarded as the more and less significant parts. Assume the input is
a k  (a0 , a1 ,
, aLk 1 ) and the output is b k  (b0 , b1 ,
, bLk 1 ) .
Let the new address is denoted as BPM.
 2i
BPM (i)  floor 
 mk
Then, bi  aBPM (i ) , for i  0, 1,
 2 Lk

 mk
m

  i mod k
2


.

, Lk  1 . Note that b k is used at the initial transmission.
4
IEEE 802.16m-09/0226r3
Coded bits of the subpacket
Significant bits
s0 s1 s 2
Insignificant bits
…
…
i 0 i1 i 2
…
Bit-priority
mapping
…
…
Bit sequence based on priority
H L H L H L H L H L H L H L H L H L H L H L H L
s 0 i 0 s1 i1 s 2 i 2
reliability order on the 16-QAM constellation
(a)
Coded bits of the subpacket
Significant bits
s0 s1 s 2
Middle significant bits
…
m 0 m1 m 2
…
…
…
s 0 m 0 i 0 s 1 m1 i 1 s 2 m 2 i 2
…
Insignificant bits
i 0 i1 i 2
…
…
Bit-priority
mapping
Bit sequence based on priority
H M L H M L H M L H M L H M L H M L H M L H M L H M L H M L H M L H M L
reliability order on the 64-QAM constellation
(b)
Fig. 3
3.1.2
Bit-priority mapping (a) for 16-QAM (b) for 64-QAM
Bit circular shifter
For retransmission, the bit sequence based on priority is shifted rightward by a shift value N shift to obtain ck ,
and then ck is fed into the next block. Or, we can circularly access the bit sequence from the new initial index
given by the shift value to obtain ck . The circular shift may change the bit reliability on the constellation, the
corresponding subcarrier indices or the stream allocation to achieve diversity gain.
Taking 16-QAM case for example, the shift value N shift can be controlled by two parameters, q1 and q2, and
can be represented as the function of them, such as N shift  f (q1 , q2 ) . q1 determines the subcarrier (or frequency)
index and q2 determines the bit reliability on the constellation and antenna. Fig. 4 illustrates this example on
condition of 16QAM. If a bit is located at different subcarriers at different transmissions, it may suffer from
different fading. That could decrease the probability of invariant bad fading on the bit. In Fig. 4, q1 and q2 are
equal to zero so that no shift occurs at the initial transmission. At the second transmission q1=2 lets the bit
sequence be shift rightward about half length of the occupied subcarrier indices and q2=2 lets the bit sequence
be shifted two units more across the original modulation symbols to obtain antenna diversity while multiple
antennas are used. At the third transmission q1=3 lets the bit sequence be shifted rightward about three-fourths
length of the occupied subcarrier indices and q2=1 lets the bit sequence be shifted one unit more to change the
bit reliability on the constellation. At the fourth transmission q1=1 lets the bit sequence be shifted rightward
5
IEEE 802.16m-09/0226r3
about one-fourth length of the occupied subcarrier indices and q2=1 lets the bit sequence be shifted one unit
more to change the bit reliability. x represents the logically inverting of x. At the second and third
transmissions, some bits are logically inverted to change the positions on the constellation, as will be introduced
in the next section.
reliability order on the 16-QAM constellation
1 TX
H L H L H L H L H L H L H L H L H L H L H L H L
s 0 i 0 s1 i1 s 2 i 2 …
Bit sequence based on priority
2 TX
f(q1=2, q2=2)
3 TX
s 0 i 0 s 1 i1
f(q1=3, q2=1)
4 TX
f(q1=1, q2=1)
…
s 0 i 0 s 1 i1
…
…
s 0 i 0 s1 i1
Note that x represents the bit logically inverting of x.
Fig. 4
An illustration of the circular shift for 16-QAM and CC-HARQ case
The shift value is calculated by the following equations.
if ( ( N CMS N str ) mod N DT )  N DT / 2
s0  floor ( N CMS / N str / N DT )  Lscd ;
if s0  0
s0  Lscd ;
else
s0  ceil ( N CMS / N str / N DT )  Lscd ;
,
s1  floor ( s0  q1 / 4 / Lscd );
s2  floor ( s0  q1 / 4) mod Lscd ;
N shift  ( s1  N DT  mk  s2  mk )  N str  q2 ;
where q1 and q2 can be referred to Table 1.
Table 1
The bit re-arrangement parameters for CC-HARQ
Modulation
NTX mod 4
q1
q2
Invert pattern
QPSK
0
1
2
3
0
2
3
1
0
2
0
2
None
None
None
None
Modulation
NTX mod 4
q1
q2
Invert pattern
16-QAM
0
1
2
3
0
2
3
1
0
2
1
1
None
[0 1]
[0 1]
None
Modulation
NTX mod 6
q1
q2
Invert pattern
6
IEEE 802.16m-09/0226r3
64-QAM
3.1.3
0
1
2
3
4
5
0
2
3
1
0
2
0
3
2
1
2
1
None
[0 1 1]
None
None
[0 1 1]
[0 1 1]
Bit inverter
The bit inverter is used to scramble the bit position with lower reliability on the constellation for some BitRe
versions. It logically inverts bits by an invert patterns. For 16-QAM, the invert pattern is [0 1], when NTX mod
4 = 1 or 2, otherwise the invert pattern is none. For 64-QAM, the invert pattern is [0 1 1], when NTX mod 6 =
1, 4 or 5, otherwise the invert pattern is none. [0 1] means the last one per two bits is needed to be inverted and
[0 1 1] means the last two per three bits are needed to be inverted. In other words, [0 1] represents every two
bits are exclusive-ORed by the vector [0 1] and [0 1 1] represents every three bits are exclusive-ORed with the
vector [0 1 1]. Besides, none means no inverting operation. The invert pattern can be referred to Table 1.
Note that Table 1 shows four BitRe versions for QPSK and 16-QAM, and six BitRe versions for 64-QAM.
However, in the later amendment draft we will suggest only first four BitRe versions for 64-QAM to keep the
same signaling overhead for each modulation type.
3.2
3.2.1
IR-HARQ mode
Bit-priority mapper
Due to the different redundancy versions from bit-selection, the first NEP bits of coded bits of different
redundancy versions are regarded as significant bits and other coded bits are regarded as insignificant bits.
Then, the coded bits are divided into two groups for 16-QAM and three groups for 64-QAM, and the more/less
significant bits are mapped to more/less reliable positions on the constellation as shown in Fig. 5.
Let the new address is denoted as BPM.
 2i
BPM (i)  floor 
 mk
 2 Lk

 mk
m

  i mod k
2

7

 , for i  0, 1, L , Lk  1 .

IEEE 802.16m-09/0226r3
Coded bits of the subpacket
Significant bits
s0 s1 s 2
Insignificant bits
…
i 0 i1 i 2
…
s0 s1 s 2
…
…
…
i 0 i1 i 2
Bit-priority
mapping
…
H H H H H H H H H H H H L L L L L L L L L L L L
constellation reliability
(a)
Coded bits of the subpacket
Significant bits
s0 s1 s 2
…
Middle significant bits
m 0 m1 m 2
…
s0 s1 s 2
…
Insignificant bits
i 0 i1 i 2
…
…
m 0 m1 m 2
…
…
…
i 0 i1 i 2
Bit-priority
mapping
…
H H H H H H H H H H H H M M M M M M M M M M M M L L L L L L L L L L L L
constellation reliability
(b)
Fig. 5
3.2.2
Bit-priority mapping (a) for 16-QAM (b) for 64-QAM
Bit circular shifter
For retransmission, the bit sequence is divided into two groups based on priority and shifted rightward by a shift
values N shift of different groups to obtain ck , and then ck is fed into the next block. We can obtain the shift
value in accordance with which group the index BPM (i ) belongs to and access the bit from the position with
the new index ADD (i ) . The circular shift may change the bit reliability on the constellation, the corresponding
subcarrier indices or the stream allocation to achieve diversity gain. The shift value and the new index ADD (i )
are calculated by the following equations.
8
IEEE 802.16m-09/0226r3
ADD(i )  Q  ((( P  N shift )  ( BPM (i)  Q)) mod P)
ADD(i ) : the index of selected bit from transmitted coded packet
where
P : the number of significant or non - significant bits
Q : a pre - defined starting index
i  0,1, 2,...( Lk  1),
if 0  BPM (i )  N EP
 P  N EP , Q  0,

 P  ( Lk - N EP ), Q  N EP , if N EP  BPM (i )  Lk
if [ P  ( Lk /(mk / 2)) ]
N EP : the number of bits in the encoder packet (before encoding )
k : index of the subpacket.
T : additional shifting length
N str : the number of streams for the subpacket
t  1 2,
else if [ P  ( Lk /(mk / 2)) ]
Lk : the number of coded bits for the subpacket
t  0,
mk : the mod ulation order for the kth subpacket
else [ P  ( Lk /(mk / 2)) ]
t  1,
end
and
T  1, if N str  1

T  2, if N str  1
N shift  ( NTX  ( P  t  ( Lk /(mk / 2))  T )) mod P,
Then, ci  a ADD (i ) , for i  0, 1, L , Lk  1 . Note that a k is coded bit sequence selected by the bit selection.
Coded bits of the subpacket
H H H H H H H H H H H H L L L L L L L L L L L L
1 TX
s0 s1 s 2
…
2 TX
s0 s1 s 2
3 TX
4 TX
Fig. 6
i 0 i1 i 2
s0 s1 s 2
s0 s1 s 2
…
…
i 0 i1 i 2
…
i 0 i1 i 2
…
i 0 i1 i 2
…
…
An illustration of the circular shift for 16-QAM and IR-HARQ case
For IR-HARQ mode, the bit inverter is not needed.
4. Simulation Results
Simulation parameters:
Carrier Frequency
2.5GHz
FFT Size (N)
1024
Guard Interval
1/8
MS Velocity
Ped-B 3 km/hr
9
…
IEEE 802.16m-09/0226r3
Permutation Type
RB (18X6), symbol based random permutation
Channel Coding
CTCs with 8 iterations, Max-Log-MAP
HARQ Type
Chase Combining (CC),
Incremental Redundancy (IR)
NEP
Chase case: 16QAM 1/2 : 480 bits * 2 FEC (5 RBs)
64QAM 1/2 : 432 bits * 2 FEC (3 RBs)
IR case: 16QAM 1/2 : 480 bits * 2 FEC (5 RBs)
64QAM 1/2 : 288 bits * 2 FEC (2 RBs)
Transmission Times
For BLER Simulation: 1TX, 2TX, 3TX
For Throughput Simulation: Maximal Trans. = 3 (including initial trans.)
Retransmission Delay
2 frames (10 msec)
MIMO Configuration
SM 2x2
Fig.7
BLER performance for 16QAM 1/2 CC-HARQ under 1TX and 2TX
10
IEEE 802.16m-09/0226r3
Fig.8
BLER performance for 16QAM 1/2 CC-HARQ under 3TX
Fig.9
Throughput performance for 16QAM 1/2 CC-HARQ
11
IEEE 802.16m-09/0226r3
Fig.10
BLER performance for 64QAM 1/2 CC-HARQ under 1TX and 2TX
Fig.11
BLER performance for 64QAM 1/2 CC-HARQ under 3TX
12
IEEE 802.16m-09/0226r3
Fig.12
Fig.13
Throughput performance for 64QAM 1/2 CC-HARQ
BLER performance for 16QAM 1/2 IR-HARQ under 1TX and 2TX
13
IEEE 802.16m-09/0226r3
Fig.14
BLER performance for 16QAM 1/2 IR-HARQ under 3TX
Fig.15
Throughput performance for 16QAM 1/2 IR-HARQ
14
IEEE 802.16m-09/0226r3
Fig.16
BLER performance for 64QAM 1/2 IR-HARQ under 1TX and 2TX
Fig.17
BLER performance for 64QAM 1/2 IR-HARQ under 3TX
15
IEEE 802.16m-09/0226r3
Fig.18
Throughput performance for 64QAM 1/2 IR-HARQ
5. References
[1] IEEE P802.16 Rev2 / D7, “Draft IEEE Standard for Local and Metropolitan Area Networks: Air Interface
for Broadband Wireless Access,” Oct. 2008.
[2] IEEE 802.16m-07/002r6, “802.16m System Requirements”
[3] IEEE C802.16m-08/043, “Style guide for writing the IEEE 802.16m amendment”
[4] IEEE 802.16m-08/003r6, “The Draft IEEE 802.16m System Description Document”
[5] IEEE C802.16m-08/581, “The Hybrid ARQ Physical Layer Architecture.”
[6] IEEE C802.16m-08/1144r1, “New bit-rearrangement method to enhance HARQ performance.”
[7] IEEE C802.16m-08/685r2, “Bit Priority Mapping to enhance CTC IR HARQ performance.”
[8] IEEE C80216m-08_301r2, “Enhancing CTC HARQ performance by bit rearrangement.”
[9] IEEE 802.16m-08/771r1, “Enhanced HARQ scheme with Signal Constellation Rearrangement.”
[10] IEEE C80216m-07_292r1, “Enhanced HARQ technique using Constellation Rearrangement.”
16
IEEE 802.16m-09/0226r3
6. Text proposal for inclusion in the 802.16m amendment
-------------------------------
Text Start
---------------------------------------------------
Insert a new section 15:
15.
Advanced Air Interface
15.3
Physical layer
15.3.A
Channel coding and HARQ
Concatenation rules
BitRe
version
Burst
partition into
FEC blocks
k
Information
Bits
(Burst
including
CRC)
FEC
block CRC
attachment
k'j
Information Bits in
the jth FEC block
FEC
encoder
kj
Information
Bits in the jth
FEC block
(including FEC
block CRC)
Bit selection
&
Repetition
kj/RMC
Mother code
bits for the jth
block where
RMC is
mother code
rate
Collection
kj/R
Coded bits for
the jth block where
R is code rate
Bit rearrangement
k'/R
Coded bits
Modulation
k'/R
Coded bits
(k'/R)/M
QAM symbols
allocated to the burst
where M is the
modulation order
Figure aaa—Block diagram of channel coding supporting HARQ
15.3.A.1
15.3.A.1.1
Channel coding
Double binary turbo code
The CTC in Section 8.4.9.2.3 is adopted.
15.3.A.1.2.
Bit selection
The bit selection is performed to generate the subpacket. The puncturing block is referred as bits selection in the
viewpoint of subpacket generation. Mother code is transmitted with one of the subpackets. The bits of a
subpacket are formed by selecting specific sequences of bits from the interleaved CTC encoder output sequence.
The resulting subpacket sequence is a binary sequence of bits for the bit re-arrangement (if supporting HARQ)
and the modulator. Note that Section 8.4.9.2.3.4.4 shall be not applied.
Let
17
IEEE 802.16m-09/0226r3
k
be the subpacket index when IR HARQ is enabled. k = 0 for the first transmission and increases by
one for the next subpacket. k = 0 when IR-HARQ is not used. When there are more than one FEC
blocks in a burst, the subpacket index for each FEC block shall be the same.
NEP
be the number of bits in the encoder packet (before encoding).
Lk
be the number of bits in the subpacket.
mk
be the modulation order for the k-th subpacket (mk = 2 for QPSK, 4 for 16-QAM, and 6 for
64-QAM).
Also, let the scrambled and selected bits be numbered from zero with the 0-th bit being the first bit in the
sequence. Then, the index of the i-th bit for the k-th subpacket shall be:
F mod(3  N EP ), if F  3  N EP

, where F  k  Lk  i and i  0, 1,
Sk , i  
( F  N EP ) mod(3  N EP ), if F  3  N EP
, Lk  1 .
The NEP, mk, and Lk are determined by the BS and can be inferred by the SS through the allocation size in the
DL-MAP and UL-MAP. The first transmission includes the systematic part of the mother code. Thus, it can be
used as the codeword for a burst where the HARQ is not applied or when Chase HARQ is applied.
Besides, for different MIMO-HARQ formats, rank should be feed back for adjusting the total number
of transmitted coded bits for retransmissions. Table zzz shows the all MIMO-HARQ formats. The
total transmitted coded bits may be verified according to Ri, Rr and p. The “Ri“ is the rate of the
previous transmission for the subpacket. The “Rr“ is the reduced rate of retransmissions for the same
subpacket. Besides, due to the rate of each retransmission may be reduced, only partial bits may be
retransmitted and equivalent pilot tones may be also reduced in accordance with the reduced rate.
Therefore, additional non-used pilot tones can be used to send more coded bits in current
retransmission. The “p” is defined as a ratio that the number of used tones between initial
transmission and each retransmission in Table zzz. Where,
Table zzz—Parameter Ri ,Rr and p for adaptive bit selection
Ri
Rr
p
1
1
1
2
1
1
2
2
1
3
2
24/23
3
3
1
4
2
24/23
4
3
1
18
IEEE 802.16m-09/0226r3
4
4
1
If the rate of retransmissions is different from the rate of initial transmission, the index of the i-th bit for the k-th
subpacket shall be:
S k ,i  ( Fk  i ) mod(3 N EP )
where
i  0,1, 2,...( Lk  1),
if ( SPIDk =0 )
Fk  0,
Lk  48  N SCHk  mk ,
else
Fk  48  N SCHk  mk ,
j  SPIDk ,
while ( j > 1 )

R  p
Fk  Fk  (48  N SCHk  mk )  r  ,
Ri 

j  j  1,
end

R  p
Lk   48  N SCHk  mk  r 
Ri 

end
Figure ttt-1, Figure ttt-2, and Figure ttt-3 show the different MIMO-HARQ formats with different
rates between initial transmission and retransmissions.
Transmitted
Coded Bits
Allocated
LRU
Transmitted
Coded Bits
Allocated
LRU
Transmitted
Coded Bits
Allocated
LRU
L
2
L
L
P1
P2
Rate = 2,
Spatial Multiplexing
for initial transmission
P1
NULL
Rate = 1,
SFBC
for retransmission
P1
P2
Rate = 2,
Spatial Multiplexing
for retransmission
Figure ttt-1—MIMO-HARQ with rate=2 for initial transmission and rate=1,2 for retransmission
19
IEEE 802.16m-09/0226r3
Transmitted
Coded Bits
Allocated
LRU
Transmitted
Coded Bits
Allocated
LRU
Transmitted
Coded Bits
Allocated
LRU
16  L
23
L
L
P1
P3
P2
NULL
P1
Rate = 3,
Spatial Multiplexing
for initial transmission
P2
P1
P3
P2
NULL
Rate = 3,
Spatial Multiplexing
for retransmission
Rate = 2,
Spatial Multiplexing
for retransmission
Figure ttt-2—MIMO-HARQ with rate=3 for initial transmission and rate=2,3 for retransmission
Transmitted
Coded Bits
Allocated
LRU
Transmitted
Coded Bits
Allocated
LRU
12  L
23
Transmitted
Coded Bits
Allocated
LRU
Transmitted
Coded Bits
Allocated
LRU
3 L
4
L
L
P1
P3
P2
P4
Rate = 4,
Spatial Multiplexing
for initial transmission
P1
P2
Rate = 2,
Spatial Multiplexing
for retransmission
P1
P3
P1
P3
P2
NULL
P2
P4
Rate = 3,
Spatial Multiplexing
for retransmission
Rate = 4,
Spatial Multiplexing
for retransmission
Figure ttt-3—MIMO-HARQ with rate=4 for initial transmission and rate=2,3,4 for retransmission
15.3.A.1.3
Modulation
20
IEEE 802.16m-09/0226r3
Gray-mapped QPSK, 16-QAM and 64-QAM shall be supported for data modulation (referring to Section
8.3.9.4.2) as shown in Fig. bbb. The constellations shall be normalized by multiplying the constellation point
with the indicated factor c to achieve equal average power.
For QPSK, b1 and b0 respectively modulate the in-phase and quadrature carriers. For 16-QAM, four bits,
b3b2b1b0 , are mapping into a position on the constellation to form a 16-QAM symbol. b3 and b2 modulate the
in-phase carrier and b1 and b0 modulate the quadrature carrier. Bits b3 and b1 respectively modulate the
phases of the in-phase and quadrature carriers. In particular, the phase is 0 when b3 or b1 is 0 and 180 when
b3 or b1 is 1. Bits b2 and b0 respectively modulate the amplitudes of the in-phase and quadrature carriers,
where the amplitude is greater when b2 or b0 is 1 than otherwise. Thus, b3 and b1 have higher reliabilities
than b2 and b0 . For 64-QAM, each modulation symbol consists of 6 bits, b5b4b3b2b1b0 , where b5 , b4 , and b3
modulate the in-phase carrier, and b2 , b1 , and b0 modulate the quadrature carrier. With three bits modulating
each carrier, the amplitude thereof is defined with a finer resolution. Particularly, b5 and b2 still respectively
modulate the phases of the carriers, but b4 and b3 together modulate the amplitude of the in-phase carrier at
four discrete levels, and b1 and b0 together modulate the amplitude of the quadrature carrier at four discrete
levels. The amplitude of the in-phase carrier is the greatest when b4 b3 =11, is the least when b4 b3 =01, and
has intermediate values when b4 b3 =10 or 00. The modulation of the quadrature carrier is similar and not
explained in detail herein.
Each M interleaved bits (M = 2 for QPSK, 4 for 16-QAM, 6 for 64-QAM) shall be mapped to the constellation
bits b(M – 1) – b0 in MSB-first order (i.e., the first bit shall be mapped to the higher index bit in the
constellation). The interleaved bit sequence denoted as ck  c0c1c2c3c4c5 L is input into the modulator and each
M bits from the first bit c0 are mapped into the constellation point with the corresponding label bM 1 b0 . For
example, the corresponding modulation symbols are [-1-i, -3-i] for 16-QAM, if the input sequence is
ck  10001110 .
21
IEEE 802.16m-09/0226r3
Figure bbb—Block diagram of channel coding supporting HARQ
15.3.A.2
15.3.A.2.1
HARQ
HARQ type
Incremental redundancy Hybrid-ARQ (IR-HARQ) and Chase Combining HARQ (CC-HARQ) are adopted. In
IR-HARQ mode, the k-th subpacket generated by the bit selection in Section 15.3.A.1.2 is transmitted at the
k-th transmission. k=0 is for the initial transmission. CC-HARQ retransmits the subpacket at the initial
transmission.
15.3.A.2.2
Bit selection
The function of bit selection is described as that in Section 15.3.A.1.2. The subpacket is generated according to
the number of transmission, HARQ mode and different MIMO-HARQ formats and then fed into the bit
re-arrangement.
15.3.A.2.3
Bit re-arrangement
For HARQ (re)transmissions, the subpacket after the bit selection may be arranged according to different bit
re-arrangement (BitRe) versions based on the changes of the significance and reliability of each coded bit. The
bit re-arrangement comprises the bit-level interleaver and bit inverter shown as in Fig. ccc. In the bit-level
interleaver the coded bits from bit-selection will be mapped to different bit reliabilities according to their
significance, as is called the bit-priority mapping (BPM) method. In other words, the significant bits will be
mapped to relatively more reliable positions and non-significant bits will be mapped to relatively less reliable
22
IEEE 802.16m-09/0226r3
positions. Then, the bit-level interleaver arranges bits in a simple manner—circular shift. A shift value is
calculated in accordance with the selected BitRe version. Instead of really shifting bits, we utilize a circular
buffer to register the coded bits after BPM and access them from the initial position which is given by the shift
value. Finally, the bit inverter may logically invert some of the coded bits according to the selected BitRe
version.
FEC
Bit rearrangement
ck
Bit-level
ak
Bit inverter
interleaver
dk
Modulator
Figure ccc—bit re-arrangement block diagram
The procedure of the bit-level interleaver can be integrated in the following equations to obtain the new address
ADD(i). The parameters are defined as
ADD(i) : the index of selected bit from transmitted coded subpacket
N shift : the shift value for the k -th transmission
k : the subpacket index
NTX : the number of transmission
N EP : the number of bits in the encoding packet (before encoding)
Lk : the number of coded bits for the k -th subpacket
mk : the modulation order for the k -th subpacket (2 for QPSK, 4 for 16-QAM and 6 for 64-QAM)
N CMS  Lk mk : the number of coded modulation symbols per subpacket
N str : the number of streams
N DT : the number of data tones per RU (resource unit)
TS : the symbol time occupied per RU (  6)
Lscd : the fundamental length of subcarriers for data, Lscd
17, if N str  1

 16, if N str  2
14, if N  4
str

Assume the input sequence is denoted as a k  [a0 , a1 , , aLk 1 ] and the output sequence is denoted as
ck  [c0 , c1 , , cLk 1 ] , where ci  a ADD (i ) , for i  0, 1,
, Lk  1 . ADD(i) can be calculated by the following
equations.
23
IEEE 802.16m-09/0226r3
if [HARQ type = CC]
for i  0, 1,
, Lk  1
if ( ( N CMS N str ) mod N DT )  N DT / 2
s0  floor ( N CMS / N str / N DT )  Lscd ;
if s0  0
s0  Lscd ;
else
s0  ceil ( N CMS / N str / N DT )  Lscd ;
s1  floor ( s0  q1 / 4 / Lscd );
s2  floor ( s0  q1 / 4) mod Lscd ;
N shift  ( s1  N DT  mk  s2  mk )  N str  q2 ;

 2i
ADD(i )   floor 
 mk

 2 Lk

 mk
m

  i mod k
2



  N shift  mod Lk


else if [HARQ type = IR]
for i  0, 1, L , Lk  1
 2i  2 Lk 
m 
j  floor 
  i mod k  ;

2 
 mk  mk 
if 0  j  N EP
P  N EP ; Q  0;
else
P   Lk  N EP  ; Q  N EP ;
if [ P  ( Lk /( mk / 2)) ]
t  1 2;
else if [ P  ( Lk /( mk / 2)) ]
t  0;
else [ P  ( Lk /( mk / 2)) ]
t  1;
if N str  1
T  1;
else
T  2;
N shift  ( NTX  ( P  t  ( Lk /(mk / 2))  T )) mod P;
ADD(i)  Q  ((( P  N shift )  ( j  Q)) mod P);
Herein, q1 and q2 can be referred to Table xxx.
Table xxx
The bit re-arrangement parameters for CC-HARQ
Modulation
NTX mod 4
q1
24
q2
Invert pattern
IEEE 802.16m-09/0226r3
QPSK
16-QAM
64-QAM
0
1
2
3
0
1
2
3
0
1
2
3
0
2
3
1
0
2
3
1
0
2
3
1
0
2
0
2
0
2
1
1
0
3
2
1
None
None
None
None
None
[0 1]
[0 1]
None
None
[0 1 1]
None
None
Finally the bit inverter is used to scramble the bit position with lower reliability on the constellation for some
BitRe versions in the CC-HARQ mode. It logically inverts bits by an invert patterns. For 16-QAM, the invert
pattern is [0 1], when NTX mod 4 = 1 or 2, otherwise the invert pattern is none. For 64-QAM, the invert pattern
is [0 1 1], when NTX mod 4 = 1, otherwise the invert pattern is none. [0 1] means the last one per two bits is
needed to be inverted and [0 1 1] means the last two per three bits are needed to be inverted. In other words, [0
1] represents every two bits are exclusive-ORed by the vector [0 1] and [0 1 1] represents every three bits are
exclusive-ORed with the vector [0 1 1]. Besides, none means no inverting operation. The invert pattern can be
referred to Table xxx. It is noted that the bit inverter is not applied in the IR-HARQ mode.
After all, the rearranged coded bits will be fed into the modulator.
15.3.A.2.4
SU-MIMO Mode Switch
The dynamic or semi-dynamic rank is fed back for enhancing the reliability of retransmission. The retransmitted
MIMO Mode is switched based on the relation of Ri and Rr. Table kkk lists the relation of SU-MIMO Mode
Switching Rule.
Table kkk—MIMO Mode HARQ Switching Table A
Ri
Rr
Rank 1
SISO +
precoding
Rank 1
SISO +
precoding
or
or
Rank 1
SFBC +
precoding
Rank 1
SFBC +
precoding
Rank 2 SM
+precoding
Rank 1
SFBC +
precoding
Rank 2 SM
+
Rank 2 SM
+
25
IEEE 802.16m-09/0226r3
15.3.A.2.5
precoding
precoding
Rank 3 SM
+
precoding
Rank 2 SM
+
precoding
Rank 3 SM
+
precoding
Rank 3 SM
+
precoding
Rank 4 SM
+
precoding
Rank 2 SM
+
precoding
Rank 4 SM
+
precoding
Rank 3 SM
+
precoding
Rank 4 SM
+
precoding
Rank 4 SM
+
precoding
MU-MIMO Mode Switch
In the Multi-User MIMO mode, only the k users who reported a NACK will ask for retransmission. Hence, the
data transmitted to M users in the previous transmission may retransmit to only k users. Each user who
transmits a NACK may retransmit with k user with 1 stream. However, if only one user reported with a NACK
message, the MU-MIMO Mode should be switched to SU-MIMO mode, and the retransmitted stream number
may grow according to SU-MIMO rank. If ABS transmits with Nt=8 and each AMS equips with Nr=4 or Nr=2,
the retransmitted combination is listed in Table uuu.
Table sss—MU-MIMO switch to SU-MIMO HARQ Switching A
Nt
Nr
Rank for NACK user
SU-MIMO Mode
8
4
4
Rank 4 SM + precoding
8
4
3
Rank 3 SM + precoding
8
2/4
2
Rank 2 SM + precoding
8
2/4
1
Rank 1 SFBC +
precoding
If ABS transmits with Nt=4 and each AMS equips with Nr=4 or Nr=2, the retransmitted combination is listed in
Table uuu.
Table uuu—MU-MIMO HARQ Switching B
Nt
Nr
Rank for NACK user
SU-MIMO Mode
4
4
4
Rank 4 SM + precoding
4
4
3
Rank 3 SM + precoding
4
2/4
2
Rank 2 SM + precoding
26
IEEE 802.16m-09/0226r3
4
2/4
1
Rank 1 SFBC +
precoding
If ABS transmits with Nt=2 and each AMS equips with Nr=4 or Nr=2, the retransmitted combination is listed in
Table vvv.
Table vvv—MU-MIMO HARQ Switching C
Nt
Nr
NACK Users k
Each user received rate
2
2/4
2
Rank 2 SM + precoding
2
2/4
1
Rank 1 SFBC +
precoding
-------------------------------
Text End
--------------------------------------------------------------------------
27