U.S. patent application number 12/682110 was filed with the patent office on 2010-08-19 for method of transmitting data using constellation rearrangement.
Invention is credited to Han Gyu Cho, Jin Soo Choi, Jae Hoon Chung, Jong Young Han, Jong Min Kim, Suk Woo Lee, Sung Ho Moon, Seung Woo Nam, Hyung Ho Park.
Application Number | 20100211842 12/682110 |
Document ID | / |
Family ID | 40549750 |
Filed Date | 2010-08-19 |
United States Patent
Application |
20100211842 |
Kind Code |
A1 |
Moon; Sung Ho ; et
al. |
August 19, 2010 |
METHOD OF TRANSMITTING DATA USING CONSTELLATION REARRANGEMENT
Abstract
A data retransmission method using hybrid automatic repeat
request (HARQ) includes transmitting a data block, receiving a
retransmission request signal for the data block, generating a
retransmission block by performing swapping or inversion between
bits constituting the data block according to the retransmission
request signal, and transmitting the retransmission block.
Inventors: |
Moon; Sung Ho; (Gyeongki-do,
KR) ; Nam; Seung Woo; (Gyeongki-do, KR) ; Kim;
Jong Min; (Gyeongki-do, KR) ; Park; Hyung Ho;
(Gyeongki-do, KR) ; Cho; Han Gyu; (Gyeongki-do,
KR) ; Choi; Jin Soo; (Gyeongki-do, KR) ;
Chung; Jae Hoon; (Gyeongki-do, KR) ; Han; Jong
Young; (Gyeongki-do, KR) ; Lee; Suk Woo;
(Gyeongki-do, KR) |
Correspondence
Address: |
MCKENNA LONG & ALDRIDGE LLP
1900 K STREET, NW
WASHINGTON
DC
20006
US
|
Family ID: |
40549750 |
Appl. No.: |
12/682110 |
Filed: |
October 9, 2008 |
PCT Filed: |
October 9, 2008 |
PCT NO: |
PCT/KR08/05947 |
371 Date: |
April 8, 2010 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
60981111 |
Oct 19, 2007 |
|
|
|
Current U.S.
Class: |
714/748 ;
714/E11.113 |
Current CPC
Class: |
H04L 1/08 20130101; H04L
1/1819 20130101; H04L 1/1893 20130101 |
Class at
Publication: |
714/748 ;
714/E11.113 |
International
Class: |
H04L 1/18 20060101
H04L001/18; G06F 11/14 20060101 G06F011/14 |
Foreign Application Data
Date |
Code |
Application Number |
Oct 9, 2007 |
KR |
10-2007-0101690 |
Claims
1. A data retransmission method using hybrid automatic repeat
request (HARQ), comprising: transmitting a data block; receiving a
retransmission request signal for the data block; generating a
retransmission block by performing swapping or inversion between
bits constituting the data block according to the retransmission
request signal; and transmitting the retransmission block.
2. The method of claim 1, wherein the swapping between bits is
performed by swapping a bit having a high bit reliability and a bit
having a low bit reliability in previous transmission.
3. The method of claim 1, wherein the data block is a transmission
block having a matrix format and consisting of rows whose number is
equal to the number of transmit antennas, each row of the
transmission block is transmitted through one transmit antenna,
bits constituting one row represent bits of one data symbol, and
the data symbol is modulated using at least two modulation
schemes.
4. The method of claim 3, wherein the swapping between bits is
performed by cyclically shifting bits constituting each column of
the transmission block, and each column has a different cyclic
shift amount.
5. The method of claim 1, wherein the retransmission block is
mapped according to the number of retransmissions when using a
16-ary quadrature amplitude modulation (QAM) or 64-QAM scheme as
expressed in the following table: TABLE-US-00011 # of trans.
Mapping set # of trans. Mapping set 0 i.sub.1q.sub.1i.sub.2q.sub.2
0 i.sub.1q.sub.1i.sub.2q.sub.2i.sub.3q.sub.3 1
i.sub.2q.sub.2i.sub.1q.sub.1 1 i.sub.3q.sub.3 .sub.2
q.sub.2i.sub.1q.sub.1 2 i.sub.1q.sub.1 .sub.2 q.sub.2 2
i.sub.2q.sub.2 .sub.3q.sub.3i.sub.1q.sub.1 3 i.sub.2q.sub.2 .sub.1
q.sub.1 3 i.sub.1q.sub.1 .sub.2q.sub.2i.sub.3q.sub.3 (a) 16 QAM (b)
64 QAM
where, in i.sub.a and q.sub.a of the table, `a` denotes an index
indicating coordinates on a constellation.
6. The method of claim 1, wherein the retransmission block is
mapped according to the number of retransmissions when using a
16-QAM or 64-QAM scheme as expressed in the following table:
TABLE-US-00012 # of re- # of re- transmission Optimal mapping
transmission Optimal mapping 0 i.sub.1,1q.sub.1,1i.sub.1,2q.sub.1,2
0 i.sub.1,1q.sub.1,1i.sub.1,2q.sub.1,2i.sub.1,3q.sub.1,3
i.sub.2,1q.sub.2,1i.sub.2,2q.sub.2,2
i.sub.2,1q.sub.2,1i.sub.2,2q.sub.2,2i.sub.2,3q.sub.2,3 1
i.sub.1,2q.sub.1,2i.sub.2,1q.sub.2,1 1 i.sub.2,3q.sub.1,3 .sub.2,2
q.sub.1,2i.sub.2,1q.sub.1,1 i.sub.2,2q.sub.2,2i.sub.1,1q.sub.1,1
i.sub.1,3q.sub.2,3 .sub.1,2 q.sub.2,2i.sub.1,1q.sub.2,1 2
i.sub.1,1q.sub.1,1 .sub.2,2 q.sub.2,2 2 i.sub.2,2q.sub.2,2 .sub.2,3
q.sub.2,3i.sub.2,1q.sub.1,1 i.sub.2,1q.sub.2,1 .sub.1,2 q.sub.1,2
i.sub.1,2q.sub.1,2 .sub.1,3 q.sub.1,3i.sub.1,1q.sub.2,1 3
i.sub.1,2q.sub.1,2 .sub.2,1 q.sub.2,1 3 i.sub.2,1q.sub.1,1 .sub.2,2
q.sub.1,2i.sub.2,3q.sub.1,3 i.sub.2,2q.sub.2,2 .sub.1,1 q.sub.1,1
i.sub.1,1q.sub.2,1 .sub.1,2 q.sub.2,2i.sub.1,3q.sub.2,3 (a) 16-QAM
(b) 64-QAM
where, in i.sub.a,b and q.sub.a,b of the table, `a` denotes an
antenna index and `b` denotes an index indicating coordinates on a
constellation.
7. A data retransmission method in a multiple-antenna system,
comprising: transmitting a first data symbol modulated with a first
modulation scheme through a first transmit antenna and transmitting
a second data symbol modulated with a second modulation scheme
through a second transmit antenna; generating a new first data
symbol and a new second data symbol by performing swapping or
inversion on bits constituting the first data symbol and bits
constituting the second data symbol; and transmitting the new first
data symbol through the first transmit antenna and transmitting the
new second data symbol through the second transmit antenna.
8. The method of claim 7, further comprising receiving a
retransmission request signal for the first data symbol and the
second data symbol.
9. An apparatus for wireless communication, comprising: a radio
frequency (RF) unit for transmitting and receiving a radio signal;
and a processor coupled with the RF unit, for performing
HARQ-increment redundancy (IR), and configured to: sequentially
configure data blocks in a cyclic buffer of the HARQ-IR according
to a retransmission request, and transmit the data blocks, wherein
if all or some parts of the data block are wrapped around with a
previously transmitted data block, constellation rearrangement is
performed on the wraparound parts.
10. The apparatus of claim 9, wherein the processor is configured
to use 16-QAM as a modulation scheme, and perform constellation
rearrangement according to the number of wraparounds as described
in the following table: TABLE-US-00013 # of wraparound Mapping rule
0 i.sub.1 q.sub.1 i.sub.2 q.sub.2 1 i.sub.2 q.sub.2 i.sub.1 q.sub.1
2 i.sub.1 q.sub.1 .sub.2 q.sub.2 3 i.sub.2 q.sub.2 .sub.1
q.sub.1
where, in i.sub.a and q.sub.a of the table, `a` denotes an index
indicating coordinates on a constellation.
11. The apparatus of claim 9, wherein the processor is configured
to use 16-QAM as a modulation scheme, and perform constellation
rearrangement according to the number of wraparounds as described
in the following table: TABLE-US-00014 # of wraparound Mapping rule
0 i.sub.1,1q.sub.1,1i.sub.1,2q.sub.1,2
i.sub.2,1q.sub.2,1i.sub.2,2q.sub.2,2 1
i.sub.1,2q.sub.1,2i.sub.2,1q.sub.2,1
i.sub.2,2q.sub.2,2i.sub.1,1q.sub.1,1 2 i.sub.1,1q.sub.1,1 .sub.2,2
q.sub.2,2 i.sub.2,1q.sub.2,1 .sub.1,2 q.sub.1,2 3
i.sub.1,2q.sub.1,2 .sub.2,1 q.sub.2,1 i.sub.2,2q.sub.2,2 .sub.1,1
q.sub.1,1
where, in i.sub.a,b and q.sub.a,b of the table, `a` denotes an
antenna index and `b` denotes an index indicating coordinates on a
constellation.
12. The apparatus of claim 9, wherein the processor is configured
to use 64-QAM as a modulation scheme, and perform constellation
rearrangement according to the number of wraparounds as described
in the following table: TABLE-US-00015 # of wraparound Mapping rule
0 i.sub.1q.sub.1i.sub.2q.sub.2i.sub.3q.sub.3 1 i.sub.3q.sub.3
.sub.2 q.sub.2i.sub.1q.sub.1 2 i.sub.2q.sub.2 .sub.3
q.sub.3i.sub.1q.sub.1 3 i.sub.1q.sub.1 .sub.2
q.sub.2i.sub.3q.sub.3
where, in ia and qa of the table, `a` denotes an index indicating
coordinates on a constellation.
13. The apparatus of claim 9, wherein the processor is configured
to use 64-QAM as a modulation scheme, and perform constellation
rearrangement according to the number of wraparounds as described
in the following table: TABLE-US-00016 # of wraparound Mapping rule
0 i.sub.1,1q.sub.1,1i.sub.1,2q.sub.1,2i.sub.1,3q.sub.1,3
i.sub.2,1q.sub.2,1i.sub.2,2q.sub.2,2i.sub.2,3q.sub.2,3 1
i.sub.2,3q.sub.1,3 .sub.2,2 q.sub.1,2i.sub.2,1q.sub.1,1
i.sub.1,3q.sub.2,3 .sub.1,2 q.sub.2,2i.sub.1,1q.sub.2,1 2
i.sub.2,2q.sub.2,2 .sub.2,3 q.sub.2,3i.sub.2,1q.sub.1,1
i.sub.1,2q.sub.1,2 .sub.1,3 q.sub.1,3i.sub.1,1q.sub.2,1 3
i.sub.2,1q.sub.1,1 .sub.2,2 q.sub.1,2i.sub.2,3q.sub.1,3
i.sub.1,1q.sub.2,1 .sub.1,2 q.sub.2,2i.sub.1,3q.sub.2,3
where, in i.sub.a,b and q.sub.a,b of the table, `a` denotes an
antenna index and `b` denotes an index indicating coordinates on a
constellation.
Description
TECHNICAL FIELD
[0001] The present invention relates to wireless communications,
and more particularly, to a method and apparatus using
constellation rearrangement in a wireless communication system.
BACKGROUND ART
[0002] Current development in advanced wireless communication has
led to the requirement of high spectral efficiency and reliable
communication. Unfortunately, packet errors by fading channel
environment and interferences originated from various sources make
the capacity of overall system to be limited.
[0003] Hybrid Automatic Repeat Request (HARQ) which is ARQ protocol
combined with Forward Error Correction (FEC) is strongly considered
as one of cutting edge technologies for future reliable
communication. The HARQ scheme can largely be classified into the
type of two. One is HARQ-Chase Combining (CC) which is disclosed in
D. Chase, Code Combining: A maximum-likelihood decoding approach
for combining an arbitrary number of noisy packets, IEEE Trans. on
Commun., Vol. 33, pp. 593-607, May 1985. The other scheme is
HARQ-Increment Redundancy (IR). In the HARQ-CC, when a receiver
detects an error through cyclic redundancy checking (CRC) while
decoding the transmitted packet, the same packet with the same
modulation and coding is sent to the receiver, repeatedly.
Meanwhile, HARQ-IR retransmits different packets in order to
achieve the coding gain, in which parity bits can be manipulated
through puncturing and repetition.
[0004] Multiple Input Multiple Output (MIMO) systems are regarded
as one of the most promising research areas of wireless
communication. Spatial diversity provided by multiple antenna
configurations for both transmitter and receiver is known to
tremendously increase system capacity without additional bandwidth.
As a result, various approaches have been studied to use the
benefit of transmit diversity and received diversity.
[0005] Constellation Rearrangement disclosed by the PCT
international application No. PCT/KR2007/003625 filed by this
applicant provides additional gain through averaging the difference
of inherent reliability between component bits.
[0006] There is a need to improve performance of system by applying
constellation rearrangement to various schemes such as MIMO system,
HARQ, multi-level modulation, etc.
DISCLOSURE OF INVENTION
Technical Problem
[0007] The present invention provides a method and apparatus for
performing constellation rearrangement in a wireless communication
system.
[0008] The present invention also provides a method of performing
constellation rearrangement in a multiple-antenna system using a
multi-modulation scheme.
Technical Solution
[0009] In an aspect, a data retransmission method using hybrid
automatic repeat request (HARQ) includes transmitting a data block,
receiving a retransmission request signal for the data block,
generating a retransmission block by performing swapping or
inversion between bits constituting the data block according to the
retransmission request signal, and transmitting the retransmission
block.
[0010] The swapping between bits may be performed by swapping a bit
having a high bit reliability and a bit having a low bit
reliability in previous transmission.
[0011] In addition, the data block may be a transmission block
having a matrix format and consisting of rows whose number is equal
to the number of transmit antennas, each row of the transmission
block may be transmitted through one transmit antenna, bits
constituting one row may represent bits of one data symbol, and the
data symbol may be modulated using at least two modulation
schemes.
[0012] In another aspect, a data retransmission method in a
multiple-antenna system includes transmitting a first data symbol
modulated with a first modulation scheme through a first transmit
antenna and transmitting a second data symbol modulated with a
second modulation scheme through a second transmit antenna,
generating a new first data symbol and a new second data symbol by
performing swapping or inversion on bits constituting the first
data symbol and bits constituting the second data symbol, and
transmitting the new first data symbol through the first transmit
antenna and transmitting the new second data symbol through the
second transmit antenna.
[0013] In still another aspect, an apparatus for wireless
communication includes an radio frequency (RF) unit for
transmitting and receiving a radio signal, and a processor coupled
with the RF unit, for performing HARQ-increment redundancy (IR),
and configured to sequentially configure data blocks in a cyclic
buffer of the HARQ-IR according to a retransmission request, and
transmit the data blocks, wherein if all or some parts of the data
block are wrapped around with a previously transmitted data block,
constellation rearrangement is performed on the wraparound
parts.
ADVANTAGEOUS EFFECTS
[0014] An additional diversity gain can be obtained in a multiple
input multiple output (MIMO) system by using constellation
rearrangement. Therefore, performance of a wireless communication
system can be improved.
BRIEF DESCRIPTION OF THE DRAWINGS
[0015] FIG. 1 is a block diagram showing a transmitter and a
receiver according to an embodiment of the present invention.
[0016] FIG. 2 shows an example of an operation of an adaptive
mapper for multi-modulation transmission in case of using two
transmit antennas.
[0017] FIG. 3 is a flow diagram showing a data transmission method
using a wireless communication system of FIG. 1.
[0018] FIG. 4 shows a transmission block in case of using multiple
antennas.
[0019] FIG. 5 shows a signal constellation of an M-ary quadrature
amplitude modulation (M-QAM) scheme.
[0020] FIG. 6 is a graph showing changes in a minimum squared error
(MSE) of average bit reliability with respect to the number of
retransmissions according to a proposed bit swapping scheme.
[0021] FIG. 7 shows a bit swapping and inversion (BSI) scheme in a
2.times.2 multiple input multiple output (MIMO) system using 16-QAM
and 64-QAM according to an embodiment of the present invention.
[0022] FIG. 8 shows an example of a bit shuffling between antennas
(BSA) scheme in single modulation transmission.
[0023] FIG. 9 shows a BSA scheme in a 5.times.5 MIMO system using
16-QAM and 64-QAM according to an embodiment of the present
invention.
[0024] FIG. 10 shows an example of applying a BSI scheme and a BSA
scheme to a 2.times.2 MIMO system using 16-QAM and 64-QAM.
[0025] FIG. 11 is a graph showing a result obtained by performing
simulations in a 2.times.2 MIMO system using 16-QAM and 64-QAM.
[0026] FIG. 12 shows gray mapping for 16-QAM.
[0027] FIG. 13 shows decision boundary to calculate a bit error
probability.
[0028] FIG. 14 shows a swapping operation.
[0029] FIG. 15 shows an inversion operation.
[0030] FIG. 16 shows decision boundary of 8-pulse amplitude
modulation (PAM).
[0031] FIG. 17 shows frame error rate (FER) performances according
to the number of bits shuffled between transmit antennas.
[0032] FIG. 18 shows a comparison result of FER performances of
optimal bitwise mapping and conventional mapping.
[0033] FIG. 19 shows hybrid automatic repeat request
(HARQ)-incremental redundancy (IR) using a swapping operation.
[0034] FIG. 20 shows HARQ-IR using a swapping operation and an
inversion operation.
[0035] FIG. 21 is a block diagram showing an apparatus for wireless
communication according to an embodiment of the present
invention.
MODE FOR THE INVENTION
[0036] The technology described below can be used in various
wireless communication systems such as code division multiple
access (CDMA), frequency division multiple access (FDMA), time
division multiple access (TDMA), orthogonal frequency division
multiple access (OFDMA), single carrier frequency division multiple
access (SC-FDMA), etc. The CDMA can be implemented with a radio
technology such as universal terrestrial radio access (UTRA) or
CDMA-2000. The TDMA can be implemented with a radio technology such
as global system for mobile communications (GSM)/general packet
ratio service (GPRS)/enhanced data rate for GSM evolution (EDGE).
The OFDMA can be implemented with a radio technology such as
institute of electrical and electronics engineers (IEEE) 802.11
(Wi-Fi), IEEE 802.16 (WiMAX), IEEE 802-20, evolved UTRA (E-UTRA),
etc. The UTRA is a part of a universal mobile telecommunication
system (UMTS). 3rd generation partnership project (3GPP) long term
evolution (LTE) is a part of an evolved UMTS (E-UMTS) using the
E-UTRA. The 3GPP LTE uses the OFDMA in downlink and uses the
SC-FDMA in uplink. LTE-advance (LTE-A) is an evolution of the 3GPP
LTE.
[0037] This technology can be used in downlink or uplink. In
general, a downlink denotes a communication link from a base
station (BS) to a user equipment (UE), and an uplink denotes a
communication link from the UE to the BS. The BS is generally a
fixed station that communicates with the UE and may be referred to
as another terminology, such as a node-B, a base transceiver system
(BTS), an access point, etc. The UE may be fixed or mobile, and may
be referred to as another terminology, such as a mobile station
(MS), a user terminal (UT), a subscriber station (SS), a wireless
device, etc.
[0038] The communication system may be a multiple-antenna system
having a plurality of transmit antennas. Hereinafter, a
multiple-input multiple-output (MIMO) system denotes a system using
a plurality of transmit antennas and/or a plurality of receive
antennas.
[0039] FIG. 1 is a block diagram showing a transmitter and a
receiver according to an embodiment of the present invention. A
transmitter 100 and a receiver 200 implement hybrid automatic
repeat request (HARQ). The transmitter 100 and the receiver 200 can
be regarded as a transceiver that performs both a transmission
function and a reception function. For clear explanation of data
retransmission, one side that transmits and retransmits data is
referred to as the transmitter 100, and the other side that
receives data and requests data retransmission is referred to as
the receiver 200. In downlink, the transmitter 100 may be a part of
a BS, and the receiver 200 may be a part of a UE. In uplink, the
transmitter 100 may be a part of the UE, and the receiver 200 may
be a part of the BS. The BS may include a plurality of receivers
and a plurality of transmitters. The UE may include a plurality of
receivers and a plurality of transmitters.
[0040] Referring to FIG. 1, a transmitter 100 includes a channel
encoder 110, an adaptive mapper 120, a spatial encoder 130, a
controller 150, and a receive circuitory 180. Further, the
transmitter 100 includes Nt modulators 140-1, . . . , 140-Nt and Nt
transmit antennas 190-1, . . . , 190-Nt, where Nt is greater than
one (i.e., Nt>1).
[0041] The channel encoder 110 receives a stream of information
bits and encodes the received stream of information bits according
to a predetermined coding scheme. As a result, coded data is
generated. The adaptive mapper 120 modulates the coded data
according to a predetermined modulation scheme and thus provides a
data symbol. The adaptive mapper 120 can use at least two
modulation schemes. The adaptive mapper 120 maps the coded data to
the data symbol representing a position on a signal constellation.
Further, the adaptive mapper 120 adaptively remaps the coded data
in response to a retransmission request message of the controller
150. There is no limit in the modulation scheme used by the
adaptive mapper 120. The modulation scheme may be an M-ary
quadrature amplitude modulation (M-QAM). Examples of the M-QAM
include 16-QAM, 64-QAM, and 256-QAM. Detailed operations of the
adaptive mapper 120 will be described below.
[0042] The spatial encoder 130 processes data symbols output
through the adaptive mapper 120 according to a MIMO pre-processing
scheme. The modulators 140-1, . . . , 140-Nt modulate symbols
output from the spatial encoder 130 and transmit the modulated
symbols through the respective transmit antennas 190-1, . . . ,
190-Nt. When the modulators 140-1, . . . , 140-Nt perform an
inverse fast Fourier transform (IFFT), orthogonal frequency
division multiplexing (OFDM) symbols are output. The receive
circuitory 180 receives signals transmitted from the receiver 200
through the transmit antennas 190-1, . . . , 190-Nt. The receive
circuitory 180 digitizes the received signals and then transmits
the digitized signals to the controller 150.
[0043] The controller 150 controls overall operations of the
transmitter 100. The controller 150 extracts information from
signals received from the receive circuitory 180. An operation of
extracting the information includes general modulation and
decoding. The extracted information may include a retransmission
request signal. The controller 150 prepares a retransmission symbol
by controlling the adaptive mapper 120 in response to the
retransmission request signal.
[0044] A channel quality indicator (CQI) may be included in the
information extracted from the signal received from the receive
circuitory 180. The CQI may be information on a channel condition
from the receiver 200 to the transmitter 100 or index information
on a modulation and coding scheme. The CQI can be used by the
controller 150 to control the channel encoder 110 or the adaptive
mapper 120. Thus, a coding scheme of the channel encoder 110 or a
mapping scheme of the adaptive mapper 120 can be adaptively
changed.
[0045] Meanwhile, a receiver 200 includes a spatial decoder 220, a
demapper 230, a channel decoder 250, an error detector 260, a
controller 270, and a transmit circuitory 280. Further, the
receiver 200 includes Nr receive antennas 290-1, . . . , 290-Nr,
where Nr is greater than one (i.e., Nr>1).
[0046] Signals received from the receive antennas 290-1, . . . ,
290-Nr are demodulated by demodulators 210-1, . . . , 210-Nr and
then are input to the spatial decoder 220. The spatial decoder 220
processes the received signals according to a MIMO post-processing
scheme in response to a MIMO control signal. The MIMO control
signal controls decoding according to a space time coding (STC)
scheme of the receiver 200. The MIMO control signal may be
pre-defined by a memory (not shown) of the controller 270.
Alternatively, the MIMO control signal may be received from the
transmitter 100.
[0047] The demapper 230 demaps data symbols from coded data
according to a demapping control signal provided from the
controller 270. The demapping control signal controls the demapper
230 according to a mapping scheme used in the adaptive mapper 120
of the transmitter 100. The demapping control signal may be
pre-defined by the memory of the controller 270. Alternatively, the
demapping control signal may be received from the transmitter
100.
[0048] The receiver 200 may include a combination unit 240 that
combines a retransmitted symbol and a previous symbol. That is, in
case of using an HARQ scheme such as HARQ-chase combining (CC) or
HARQ-incremental redundancy (IR), the combination unit 240 combines
retransmitted symbols and previous symbols. A combining scheme in
use may be an equal-gain combining scheme in which combination is
performed using an average value by assigning the same weight
factor to both previous data and retransmitted data. The combining
scheme may be a maximal ratio combining (MRC) scheme in which
weight factors are assigned to respective pieces of data. There is
no limit in the combining scheme, and thus other various schemes
may also be used.
[0049] The technical features of the present invention are not
limited to the HARQ-CC or HARQ-UR scheme, and can also apply to an
HARQ scheme in which channel decoding is performed using only
retransmitted symbols without being combined with previous symbols.
In this case, the receiver 200 may not include the combination unit
240 as indicated by a dotted line in the figure.
[0050] The channel decoder 250 decodes coded data according to a
predetermined decoding scheme. The error detector 260 detects an
error from a decoded data bit by using cyclic redundancy check
(CRC).
[0051] The controller 270 controls overall operations of the
receiver 200 and provides a retransmission request signal or the
like to the transmit circuitory 280. For this, the controller 270
can perform general channel encoding, modulation, etc. The
controller 270 receives a result of error detection from the
channel decoder 250 and determines whether to request
retransmission. The controller 270 may generate a positive
acknowledgement (ACK) signal if no error is detected, and may
generate a negative acknowledgement (NACK) signal if an error is
detected. The NACK signal may be the retransmission request
signal.
[0052] Further, the controller 270 can provide a CQI signal by
measuring channel quality from received signals. The CQI signal is
a feedback signal to be fed back to the transmitter 100. The
feedback signal indicates channel quality such as a signal-to-noise
ratio (SNR) or an error rate. The transmit circuitory 280 receives
the retransmission request signal or the like from the controller
270, and transmits the received signal through the receive antennas
290-1, . . . , 290-Nr.
[0053] I. Constellation Rearrangement for Multi-Modulation
Transmission.
[0054] Different modulation schemes (hereinafter, multi-modulation
transmission) can be used for respective transmit antennas in a
MIMO system. For example, a 1st transmit antenna may use 16-QAM and
a 2nd transmit antenna may use 64-QAM. A method of performing
constellation rearrangement in a multiple-antenna system in which
various modulation schemes co-exist for respective transmit
antennas will be described.
[0055] FIG. 2 shows an example of an operation of an adaptive
mapper for multi-modulation transmission in case of using two
transmit antennas.
[0056] Referring to FIG. 2, the 1st transmit antenna 190-1 uses
16-QAM and the 2nd transmit antenna 190-2 uses 64-QAM. In initial
transmission, four bits constitute one data symbol in the 1st
transmit antenna 190-1, and six bits constitute one data symbol in
the 2nd transmit antenna 190-2. Data symbols for initial
transmission in the 1st transmit antenna 190-1 are expressed by
{b.sub.1, b.sub.2, b.sub.3, b.sub.4}. Data symbols for initial
transmission in the 2nd transmit antenna 190-2 are expressed by
{b.sub.5,b.sub.6,b.sub.7,b.sub.8,b.sub.9,b.sub.10}. When an error
occurs according to a channel condition, data symbols transmitted
from the respective transmit antennas are retransmitted. The
adaptive mapper 120 configures bits constituting a retransmission
symbol for each antenna by performing swapping or inversion of each
bit on a signal constellation of the symbol in comparison with
initial transmission. Unlike the initial transmission, in
retransmission, the 1st transmit antenna 190-1 uses 64-QAM and the
2nd transmit antenna 190-2 uses 16-QAM.
[0057] According to this example, in retransmission, data symbols
of the 1st transmit antenna 190-1 are remapped to {b.sub.3,
b.sub.5, b.sub.1, b.sub.7}, and data symbols of the 2nd transmit
inversion operation.
[0058] FIG. 3 is a flow diagram showing a data transmission method
using the wireless communication system of FIG. 1.
[0059] Referring to FIG. 3, the transmitter 100 transmits a
transmission (Tx) block S.sup.(O) (step S110). The Tx block is
formed in an Nt.times.2B.sub.max matrix format, and is a data block
max mapped onto a signal constellation by the adaptive mapper 120.
Nt denotes the number of transmit antennas. 2B.sub.max denotes the
number of index bits for a modulation scheme having a highest order
among modulation schemes used in multi-modulation transmission. If
a maximum modulation order is M-QAM, 2B.sub.max is log.sub.2M. For
example, 2B.sub.max is 4 in 16-QAM. The Tx block expressed in a
mathematical form consists of a predetermined number of rows,
wherein the predetermined number is equal to the number of transmit
antennas. Each row is transmitted through one transmit antenna.
Bits constituting one row represent bits of one data symbol. In one
Tx block, the data symbol is modulated using at least two
modulation schemes.
[0060] The superscript of the Tx block S denotes the number of
retransmissions. For example, S.sup.(0) denotes a Tx block for
initial transmission, and S.sup.(1) denotes a Tx block for 1st
retransmission.
[0061] The receiver 200 detects an error from the received Tx block
S.sup.(0) (step S120). If no error is detected, the receiver 200
transmits an ACK signal to the transmitter 100, and waits for
transmission of a next Tx block. However, it will be assumed herein
that the receiver 200 detects the error and thus transmits a NACK
signal as a retransmission request signal (step S130).
[0062] Upon receiving the NACK signal, the transmitter 100
transmits a retransmitted Tx (ReTx) block S.sup.(1) (step S140).
Upon receiving the NACK signal, the controller 150 controls the
adaptive mapper 120 to remap the Tx block S.sub.(0) in a bitwise
manner and/or in a spatial manner, and thus configures the ReTx
block S.sup.(1). Various schemes can be used as a remapping scheme
used in retransmission, which will be described below.
[0063] The receiver 200 detects an error from the received ReTx
block S.sup.(1) (step S150). In this case, the combination unit 240
can combine the previous Tx block S.sup.(0) and the ReTx block
S.sup.(1).
[0064] If no error is detected, the receiver 200 transmits an ACK
signal to the transmitter 100, and waits for transmission of a next
symbol. However, it will be assumed herein that the receiver 200
detects the error and thus transmits a NACK signal as a
retransmission request signal (step S160).
[0065] Upon receiving the NACK signal, the transmitter 100
transmits a remapped ReTx block S.sup.(2) (step S170). The adaptive
mapper 120 remaps the Tx block S.sup.(0) in a bitwise manner and/or
in a spatial manner, and thus configures the ReTx block
S.sup.(2).
[0066] The receiver 200 detects an error from the received ReTx
block S.sup.(2) (step S180). The receiver 200 transmits an ACK
signal or a NACK signal to the transmitter 100 according to an
error detection result (step S190). When the ACK signal is
transmitted, retransmission for a corresponding Tx block is
finished. A retransmission request in response to the NACK signal
can be repeated up to a predetermined number M of times, where M is
an iteration number greater than 0 (i.e., M.gtoreq.1). If errors
are continuously detected even after M-th retransmission is
performed, a retransmission process can be reset and then a next Tx
block can start to be transmitted. Alternatively, transmission of a
current Tx block may be resumed.
[0067] FIG. 4 shows a Tx block in case of using multiple
antennas.
[0068] Referring to FIG. 4, T.sub.0 denotes initial transmission,
T.sub.1 denotes 2nd transmission, that is, 1st retransmission, and
T.sub.m denotes (m+1)th transmission, that is, m-th retransmission.
S.sup.(0) denotes a Tx block for initial transmission. S.sup.(1)
denotes a Tx block for 1st retransmission. S.sup.(m) denotes a Tx
block for m-th retransmission.
[0069] If different modulation schemes are used by the respective
transmit antennas, and if 2B.sub.n denotes the number of index bits
for a modulation scheme of an n-th transmit antenna, then an n-th
row s.sup.(0) constituting the Tx block S.sup.(0) can be expressed
with bits indicating an in-phase (I)-axis and a quadrature (Q)-axis
as shown:
MathFigure 1
s.sub.n.sup.(0)=[i.sub.n,1 . . . i.sub.n,B.sub.n q.sub.n,1 . . .
q.sub.n,B.sub.n] [Math.1]
[0070] where i and q respectively denote bits indicating the I-axis
and the Q-axis on the signal constellation. The positions of the
I-axis and the Q-axis are not absolute positions. That is, if one
axis on the signal constellation is referred to as the I-axis, the
other axis is referred to as the Q-axis.
[0071] Thus, the Tx block S.sup.(0) can be expressed as
follows:
MathFigure 2 S ( 0 ) = [ I ( 0 ) Q ( 0 ) ] = [ i 1 , 1 i 1 , B 1 x
x q 1 , 1 q 1 , B 1 x x i 2 , 1 i 2 , 2 i 1 , B 2 x q 2 , 1 q 2 , 2
q 1 , B 2 x x x i Nt , 1 i Nt , 2 i Nt , 3 i Nt , B max q Nt , 1 q
Nt , 2 q Nt , 3 q q Nt , B max ] [ Math . 2 ] ##EQU00001##
[0072] where Nt denotes the number of transmit antennas, and x
denotes an empty element which indicates that no value exists in a
corresponding location. Likewise, retransmission blocks S.sup.(1),
S.sup.(m) can be expressed with an Nt.times.2B.sub.max matrix.
[0073] In a case where the respective transmit antennas use
different modulation schemes or the same modulation scheme, the
retransmission blocks S.sup.(1), S.sup.(m) are searched for,
wherein bit swapping and inversion are performed on these blocks to
reduce a bit error probability in each retransmission.
[0074] A link performance gain can be obtained in a retransmission
method based on the adaptive mapper for the following two reasons.
First, a position of each bit has an unequal bit importance due to
a QAM characteristic. When diversity can be obtained by varying
mapping of the signal constellation, the diversity is called
mapping diversity. The mapping diversity is obtained by performing
swapping of bits constituting a data symbol or by performing bit
inversion when retransmission is made. Second, spatial diversity
can be obtained by shuffling transmit antennas when retransmission
is made using multiple antennas.
[0075] Bit swapping and inversion (BSI) denotes horizontal
rearrangement for obtaining the mapping diversity. Bit shuffling
between antennas (BSA) denotes vertical rearrangement for obtaining
the spatial diversity. If all transmit antennas use the same
modulation scheme, bit mappings obtained using the BSI and the BSA
are independent from each other. On the other hand, if each
transmit antenna uses a different modulation scheme, the obtained
bit mappings are not completely independent since swapping between
bits located in different rows has to be considered using the
BSI.
[0076] There is a need for a mapping scheme capable of optimizing
the bit error probability and the spatial diversity.
[0077] FIG. 5 shows a signal constellation of an M-QAM scheme. It
is assumed that the location of signal constellation conforms to
general gray mapping.
[0078] Referring to FIG. 5, if M.sub.n-QAM denotes a modulation
scheme used in an n-th transmit antenna, a total of M.sub.n signal
constellations exist, and the number of index bits for each
modulation scheme is 2B.sub.n=log.sub.2M.sub.n.
[0079] From the perspective of the I-axis, if D.sub.n denotes a
minimum distance between positions on the signal constellation, a
position I(c) of an arbitrary Tx signal c in one axis can be
expressed by Equation 3 below.
MathFigure 3 I ( c ) = - ( M n + 1 - 2 i ) D n 2 , i = 1 , 2 , , M
n [ Math . 3 ] ##EQU00002##
[0080] From the perspective of the I-axis, A(c) denotes a priori
probability in an arbitrary Tx signal c. The Tx signal is
c=(i.sub.n,1, i.sub.n,2, i.sub.n,3, . . . , i.sub.n,Bn). In
addition, i.sub.n,1 denotes a most significant bit (MSB), and
i.sub.n,Bn denotes a least significant bit (LSB). If y.sub.n
denotes a reception (Rx) signal, a log-likelihood ratio (LLR) of a
bit i.sub.n,b can be expressed by Equation 4 below.
MathFigure 4 LLR ( i n , b y n ) = log { c .di-elect cons. { i n ,
b = 1 } A ( c ) P ( y n I ( c ) ) c .di-elect cons. { i n , b = 0 }
A ( c ) P ( y n I ( c ) ) } .apprxeq. ( 2 y n ( I min , 1 ( c ) - I
min , 0 ( c ) ) + ( I min , 0 ( c ) 2 - I min , 1 ( c ) 2 - I min ,
1 ( c ) 2 ) ) ( N 0 ) , where I min , x ( c ) = arg min I ( c ) ( y
n - I ( c .di-elect cons. { i n , b = x } ) ) [ Math . 4 ]
##EQU00003##
[0081] An Rx signal z.sub.n denotes a signal received from an n-th
antenna and completely restored using a MIMO equalizer. If it is
assumed that the Rx signal z creates an Rx signal y.sub.n by
removing channel information h.sub.n through a zero-forcing (ZF)
equalizer, the Rx signal y.sub.n can be expressed as shown:
MathFigure 5
y.sub.n=x.sub.n+n' [Math.5]
[0082] where x.sub.n denotes a Tx signal. n' denotes an increased
noise generated when passing the ZF equalizer, and has a normal
distribution of N(0, N.sub.0/(2|h.sub.0|.sup.2).
[0083] According to Equations 4 and 5, an average and dispersion of
a received LLR value can be calculated as follows.
MathFigure 6 LLR ( i n , b y n ) .apprxeq. 2 E [ x n ] ( I min , 1
( c ) - I min , 0 ( c ) ) + ( I min , 0 ( c ) 2 - I min , 1 ( c ) 2
) N 0 / h n 2 V [ LLR ( i n , b y n ) ] .apprxeq. ( 2 ( I min , 1 (
c ) - I min , 0 ( c ) ) N 0 / h n 2 ) 2 N 0 2 h n 2 [ Math . 6 ]
##EQU00004##
[0084] If it is assumed that LLR values are calculated by
performing MIMO combining with a bit unit in every retransmission,
the LLR values are summed until rth retransmission is performed.
The obtained sum of LLR values can be expressed by the following
equation.
MathFigure 7 LLR ( comb ) ( i n , b y n ) = i = 0 r LLR ( l ) ( i n
, b y n ) , b = 1 , 2 , , B n [ Math . 7 ] ##EQU00005##
[0085] By using a Q-function and a Chernoff bound, a bit error rate
(BER) value of individual transmission bits constituting each data
symbol can be expressed by the following equation.
MathFigure 8 P n , b c ( r ) = Q ( E [ LLR ( comb ) ( i n , b y n )
] V [ LLR ( comb ) ( i n , b y n ) ] ) = Q ( l = 0 r E [ LLR ( l )
( i n , b y n ) ] l = 0 r V [ LLR ( l ) ( i n , b y n ) ] )
.ltoreq. 1 2 exp { - 1 2 ( l = 0 r E [ LLR ( l ) ( i n , b y n ) ]
) 2 l = 0 r V [ LLR ( l ) ( i n , b y n ) ] } = 1 2 exp { - 1 2 R n
, b c ( r ) } [ Math . 8 ] ##EQU00006##
[0086] Herein, R.sup.(r).sub.n,b|c denotes a ratio of an average
and a dispersion of the summed LLR values expressed in a last part
of Equation 8, and is defined as a conditional bit reliability of a
bit i.sub.n,b. The term "conditional" means that the bit
reliability is at a time when a Tx signal (i.e., data symbol) c is
transmitted. By averaging conditional bit reliabilities for all
possible data symbols, an average bit reliability can be obtained
by the following equation.
MathFigure 9 R n , b ( r ) = 1 M n c R n , b c ( r ) [ Math . 9 ]
##EQU00007##
[0087] According to Equation 9, the conditional bit reliability and
the average bit reliability depending on a transmission bit of the
M-QAM can be obtained.
[0088] Table 1 shows the average bit reliability in the I-axis of
16-QAM when initial transmission is performed. Herein,
D.sup.2.sub.n=E.sup.2.sub.s/10.
TABLE-US-00001 TABLE 1 c I(c) R n , 1 c ( r ) ##EQU00008## R n , 2
c ( r ) ##EQU00009## 00 -3D.sub.n/2 2 D n 2 h n 2 N 0 ##EQU00010##
1 2 D n 2 h n 2 N 0 ##EQU00011## 01 -D.sub.n/2 1 2 D n 2 h n 2 N 0
##EQU00012## 1 2 D n 2 h n 2 N 0 ##EQU00013## 10 3D.sub.n/2 2 D n 2
h n 2 N 0 ##EQU00014## 1 2 D n 2 h n 2 N 0 ##EQU00015## 11
D.sub.n/2 1 2 D n 2 h n 2 N 0 ##EQU00016## 1 2 D n 2 h n 2 N 0
##EQU00017## R n , b ( r ) ##EQU00018## N/A 1.25 D n 2 h n 2 N 0
##EQU00019## 1 2 D n 2 h n 2 N 0 ##EQU00020##
[0089] Table 2 shows the average bit reliability in the I-axis of
64-QAM when initial transmission is performed. Herein,
D.sup.2.sub.n=E.sup.2.sub.s/84.
TABLE-US-00002 TABLE 2 c I(c) R n , 1 c ( r ) ##EQU00021## R n , 2
c ( r ) ##EQU00022## R n , 3 c ( r ) ##EQU00023## 000 -7D.sub.n/2 8
D n 2 h n 2 N 0 ##EQU00024## 2 D n 2 h n 2 N 0 ##EQU00025## 1 2 D n
2 h n 2 N 0 ##EQU00026## 001 -5D.sub.n/2 4.5 D n 2 h n 2 N 0
##EQU00027## 1 2 D n 2 h n 2 N 0 ##EQU00028## 1 2 D n 2 h n 2 N 0
##EQU00029## 010 -D.sub.n/2 1 2 D n 2 h n 2 N 0 ##EQU00030## 2 D n
2 h n 2 N 0 ##EQU00031## 1 2 D n 2 h n 2 N 0 ##EQU00032## 011
-3D.sub.n/2 2 D n 2 h n 2 N 0 ##EQU00033## 1 2 D n 2 h n 2 N 0
##EQU00034## 1 2 D n 2 h n 2 N 0 ##EQU00035## 100 7D.sub.n/2 8 D n
2 h n 2 N 0 ##EQU00036## 2 D n 2 h n 2 N 0 ##EQU00037## 1 2 D n 2 h
n 2 N 0 ##EQU00038## 101 5D.sub.n/2 4.5 D n 2 h n 2 N 0
##EQU00039## 1 2 D n 2 h n 2 N 0 ##EQU00040## 1 2 D n 2 h n 2 N 0
##EQU00041## 110 D.sub.n/2 1 2 D n 2 h n 2 N 0 ##EQU00042## 2 D n 2
h n 2 N 0 ##EQU00043## 1 2 D n 2 h n 2 N 0 ##EQU00044## 111
3D.sub.n/2 2 D n 2 h n 2 N 0 ##EQU00045## 1 2 D n 2 h n 2 N 0
##EQU00046## 1 2 D n 2 h n 2 N 0 ##EQU00047## R n , b ( r )
##EQU00048## N/A 3.75 D n 2 h n 2 N 0 ##EQU00049## 1.25 D n 2 h n 2
N 0 ##EQU00050## 1 2 D n 2 h n 2 N 0 ##EQU00051##
[0090] As shown in Tables 1 and 2, the average bit reliability is
higher in the MSB (i.e., i.sub.n,1) than in the LSB (i.e.,
i.sub.n,2). This implies that more bit errors can be resulted when
the same information is transmitted through an LSB position than
when the same information is transmitted through an MSB position.
Therefore, if a bit transmitted through the LSB in initial
transmission can be transmitted through the MSB in retransmission,
reliability of an overall link performance can be increased.
[0091] Comparing Tables 1 and 2, the average bit reliability has a
significant difference according to a modulation scheme. Therefore,
the link performance can be more improved when bit swapping is
performed between data symbols using different modulation schemes
than when bit swapping is performed between data symbols using one
modulation scheme in every retransmission. Swapping is defined as
positional changes between bits.
[0092] Since all antennas use the same modulation scheme in
single-modulation transmission, data symbols for all transmit
antennas have the same type of average bit reliability. Therefore,
theoretically, the average bit reliability of all data symbols can
be determined to be almost constant by performing several
retransmissions. However, since data symbols for all transmit
antennas have different types of average bit reliability in
multi-modulation transmission, swapping has to be performed so that
a difference of average bit reliability can be reduced as much as
possible.
[0093] To quantize the difference of average bit reliability by
performing bit swapping, a minimum squared error (MSE) is defined
by the following equation.
MathFigure 10 ( r ) = n = 1 N t b = 1 B n ( R n , b ( r ) - E [ R n
, b ( r ) ] ) 2 [ Math . 10 ] ##EQU00052##
[0094] By minimizing the MSE of bit reliability defined in Equation
10, the difference of average reliability can be reduced in every
retransmission. Each bit has the difference of average reliability
in multi-modulation transmission.
[0095] In R.sup.(r-1).sub.n,b obtained by performing (r-1)
retransmissions, a best swapping matrix I(r) is a matrix that
satisfies the following equation.
MathFigure 11 I ( r ) = arg min .A-inverted. I ( r ) ( r ) [ Math .
11 ] ##EQU00053##
[0096] If the number of transmit antennas and a modulation scheme
used by each transmit antenna are predetermined, an average bit
reliability to be added to an LLR value for each bit in every
retransmission is predetermined. Therefore, when an average bit
reliability for each bit is determined after performing (r-1)
retransmissions, a bit having a lowest bit reliability is shifted
to a position at which a highest bit reliability can be obtained in
rth retransmission, and a bit having a highest bit reliability is
shifted to a position at which a lowest bit reliability can be
obtained in rth retransmission. Remaining bits are also shifted to
positions having a high reliability in an ascending order of bit
reliability. According to this method, conditions for minimizing
the MSE of Equation 10 is satisfied in a state that an average bit
reliability value summed in every transmission is determined for
each transmit antenna and for each bit.
[0097] A minimum size of a bit swapping set according to the
proposed method is defined as a first .eta. satisfying
.epsilon..sup.(.eta.-1)<.epsilon..sup.(.eta.).
[0098] For example, it is assumed that a 1st data symbol modulated
with 16-QAM is transmitted through a 1st transmit antenna, and a
2nd data symbol modulated with 64-QAM is transmitted through a 2nd
transmit antenna.
[0099] FIG. 6 is a graph showing changes in an MSE of average bit
reliability with respect to the number of retransmissions according
to the proposed bit swapping scheme. The graph shows changes of
.epsilon..sup.(r) that is normalized to .epsilon..sup.(0) of
initial transmission.
[0100] Referring to FIG. 6, according to the proposed bit swapping
scheme, a difference of bit reliability can be reduced by half in
1st retransmission. In this case, a minimum size of the bit
swapping set is 3. Therefore, the bit swapping set proposed in this
example can be expressed by the following equation.
MathFigure 12 I ( r ) .di-elect cons. { [ i 1 , 1 i 1 , 2 i 1 , 3 i
2 , 1 i 2 , 2 x ] , [ i 1 , 1 i 2 , 2 i 2 , 1 i 1 , 3 i i , 2 x ] ,
[ i 1 , 1 i 1 , 3 i 2 , 1 i 1 , 2 i 2 , 2 x ] , } [ Math . 12 ]
##EQU00054##
[0101] In each matrix, a first row represents 64-QAM transmission,
and a second row represents 16-QAM transmission. A first element of
the swapping set represents initial transmission. As shown in
Tables 1 and 2 above, in case of 64-QAM and 16-QAM, an MSB of
16-QAM is positioned where a highest bit reliability can be
obtained, followed by an LSB of 16-QAM, a middle bit of 64-QAM, and
an LSB of 64-QAM, in that order. Therefore, a second matrix is
configured such that an LSB of 64-QAM which has a lowest bit
reliability in a first matrix is positioned to an MSB of 16-QAM,
and a middle bit of 64-QAM is positioned to an LSB of 16-QAM.
Accordingly, an MSE of average bit reliability is reduced. In a
third matrix, bits are mapped by reverse-sorting according to the
summed average bit reliability obtained by performing two
retransmissions. Since the average bit reliability is sufficiently
decreased after performing three retransmissions as shown in FIG.
6, the proposed bit mapping of Equation 12 is repeatedly used.
[0102] A method of reducing a difference of average bit reliability
between bits by performing bit swapping is necessary to find a best
BSI scheme. However, it is difficult to reduce a difference of
conditional bit reliability between data symbols when only bit
swapping is used.
[0103] To reduce the difference of conditional bit reliability, it
is necessary to perform an inversion operation in retransmission.
In a case where (1,0) and (1,1) of Table 1 are transmitted, if an
LSB is inversely mapped in transmission, a receiver receives (1,1)
and (1,0) having an opposite LLR value. When using single
modulation transmission, since an MSB of 16-QAM has a total of two
absolute values for different LLR values, an absolute value
difference of LLR between data symbols can be reduced when
transmission is made by performing the inversion operation one time
for initial transmission. However, when using multi-modulation
transmission, rows of a Tx block are configured to have different
sizes. Bit inversion for an nth antenna whose row is less than
B.sub.max has to be repetitively performed. Bit inversion on an
n-th row of the Tx block can be expressed by the following
equation.
MathFigure 13 I ( r ) ( n , : ) .di-elect cons. { [ i n , 1 i n , 2
_ i n , 3 i n , B n ] [ i n , 1 i n , 2 i n , 3 _ i n , B n ] [ i n
, 1 i n , 2 _ i n , 3 _ i n , B n _ ] [ i n , 1 i n , 2 _ i n , 3 i
n , B n ] [ i n , 1 i n , 2 i n , 3 _ i n , B n ] [ i n , 1 i n , 2
_ i n , 3 _ i n , B n _ ] } [ Math . 13 ] ##EQU00055##
[0104] In the above equation, I.sup.(r)(n,:) denotes an nth row of
I.sup.(r). A size of the proposed inversion matrix is 2.sup.Bmax-1.
The number of times of repeating the inversion operation within
I.sup.(r)(n,:) is 2.sup.Bmax-Bn. According to the principal of the
above inversion operation, the bit reliability exactly the same for
all data symbols in multi-modulation transmission.
[0105] By combining bit swapping and bit inversion, combined bit
reliabilities can be approximated with respect to all data symbols
and all bit positions. Therefore, a BSI set size proposed in the
present invention for multi-modulation transmission is
.eta.2.sup.Bmax-1.
[0106] FIG. 7 shows a BSI scheme in a 2.times.2 MIMO system using
16-QAM and 64-QAM according to an embodiment of the present
invention. Herein, Equations 12 and 13 are used in combination by
the 2.times.2 MIMO system using 16-QAM and 64-QAM.
[0107] Referring to FIG. 7, three Tx blocks are obtained by
performing bit swapping according to Equation 12 above. If the
three Tx blocks are extended by performing an inversion operation,
a total of 12 BSI sets are obtained. A total of four Tx blocks are
configured by performing the inversion operation with respect to a
1st transmit antenna using 64-QAM. Four Tx blocks are configured by
repeatedly performing an inversion and non-inversion operation two
times on a second bit with respect to a 2nd transmit antenna using
16-QAM whose modulation order is lower than that of 64-QAM. A
difference of average bit reliability can be significantly reduced
using the inversion operation.
[0108] Now, optimization of the BSA scheme in multi-modulation
transmission will be described.
[0109] In a diversity transmission scheme used in a conventional
MIMO system, the same data symbol is transmitted to a plurality of
transmit antennas to obtain spatial diversity, and the same data
symbol is repeatedly transmitted at different time to obtain time
diversity. In the HARQ scheme, a time difference between
retransmissions is several-millisecond (ms) unit, which is
significantly greater than a symbol unit. Thus, a channel condition
is significantly different between retransmissions. Accordingly,
when retransmission is considered, a higher diversity gain can be
obtained using the spatial diversity in which a transmit antenna is
changed for each bit.
[0110] FIG. 8 shows an example of a BSA scheme in single modulation
transmission.
[0111] Referring to FIG. 8, transmission is achieved by applying
different shift amounts to respective columns so that bits
constituting one data symbol are transmitted respectively through
different antennas in every retransmission. Applying different
shift amounts to respective columns of a Tx block is called
multi-step cyclic shift (MSCS). A maximum diversity gain can be
expected through the MSCS.
[0112] Since respective bits constituting one data symbol can be
transmitted through different transmit antennas, if the number B of
bits constituting one I-axis or Q-axis symbol is greater than the
number Nt of transmit antennas, up to B additional diversities can
be obtained.
[0113] The MSCS can be expressed with the aforementioned matrix
expression according to the following equation.
MathFigure 14
I.sup.(1)=f.sub.MSCS(I.sup.(0)) [Math.14]
[0114] In case of performing two or more retransmissions, a time
difference between retransmissions is relatively large, and thus it
can be regarded that a channel response is independent between
retransmissions. Therefore, it can be regarded that immediately
previous BSA does not have an effect in determining of BSA in every
retransmission. Accordingly, BSA for 2nd or higher retransmission
as shown Equation 15 can also directly use an optimal BSA scheme of
Equation 14 irrespective of previous BSA.
MathFigure 15
I.sup.(r)=f.sub.MSCS(I.sup.r-1)), r=1, 2, . . . , r.sub.max
[Math.15]
[0115] In multi-modulation transmission, the BSA in the
aforementioned single modulation transmission can be extended in
the similar manner. It is desired to use MSCS-based BSA since the
purpose of using the multi-modulation transmission is to obtain a
high spatial diversity by transmitting bits constituting one data
symbol through different transmit antennas. However, since the
number of bits constituting each row of a Tx block is different
unlike in the single modulation transmission, a group MSCS (GMSCS)
scheme is proposed.
[0116] According to the GMSCS scheme, vertical swapping of bits
through MSCS is performed only between rows having the same number
of bits. If a given matrix is divided into an inter-row permutation
matrix P.sub.row and a depermutation matrix D.sub.row, an operation
of GMSCS can be mathematically expressed as follows.
MathFigure 16 I ( r ) = f GMSCS ( I ( r - 1 ) P row ) D row = f
GMSCS ( [ I 2 ( r - 1 ) I 3 ( r - 1 ) I B max ( r - 1 ) ] ) D row =
[ f MSCS ( I 2 ( r - 1 ) ) f MSCS ( I 3 ( r - 1 ) ) f MSCS ( I B
max ( r - 1 ) ) ] D row , r = 1 , 2 , , r max [ Math . 16 ]
##EQU00056##
[0117] Herein, I.sup.(r-1).sub.m denotes a matrix which is newly
created by collecting rows having m bits. The operation of BSA can
maintain independency from the operation of BSI. That is, an LLR
value equally adjusted by the BSI scheme is not changed.
[0118] FIG. 9 shows a BSA scheme in a 5.times.5 MIMO system using
16-QAM and 64-QAM according to an embodiment of the present
invention.
[0119] Referring to FIGS. 9, 1st, 3rd, and 5th transmit antennas
use 64-QAM, and 2nd and 4th antennas use 16-QAM. Therefore, in
initial transmission, a Tx block is expressed in a format of
I.sup.(0).
[0120] A matrix I.sup.(1) is generated by an algorithm using GMSCS.
When bit swapping is performed between 2nd and 4th transmit
antennas using 16-QAM, a second bit of each row is swapped by
1-step cyclic shift. When bit swapping is performed between 1st,
3rd, and 5th transmit antennas using 64-QAM, a second bit is
swapped by 1-step cyclic shift, and a third bit is swapped by
2-step cyclic shift. Matrixes I.sup.(2), . . . , I.sup.(r) for two
or more retransmissions can also be obtained using the same GMSCS
algorithm.
[0121] FIG. 10 shows an example of applying a BSI scheme and a BSA
scheme to a 2.times.2 MIMO system using 16-QAM and 64-QAM. A
combination of the BSI and BSA schemes is used in multi-modulation
transmission. In this example, the matrix set and the BSI scheme of
FIG. 7 are applied. This is because while the proposed BSA scheme
using GMSCS is applied between two or more transmit antennas using
the same modulation scheme, the proposed BSA scheme is not applied
in this example since there are only two transmit antennas using
different modulation schemes.
[0122] The proposed scheme is applied for the number of arbitrary
retransmissions, a combination of arbitrary m-QAM schemes, and the
number of arbitrary transmit antennas.
[0123] FIG. 11 is a graph showing a result obtained by performing
simulations in a 2.times.2 MIMO system using 16-QAM and 64-QAM. The
result of FIG. 11 is obtained by performing one retransmission with
respect to the conventional method, an antenna switching method in
which only BSA is performed in a symbol unit, and the proposed
method. It is shown that a gain of about 3 dB can be obtained in
the proposed method in comparison with the conventional method
irrespective of a speed of a UE.
[0124] II. Constellation Rearrangement for MIMO Systems with
HARQ
[0125] Spatial Multiplexing (SM) system using N transmit antenna
and M receive antennas is considered. Suppose that the information
bits are encoded with a coding rate of R by turbo code and L-coded
bits are mapped into one of the complex constellation points out of
the 2L-ary Quadrature Amplitude Modulation (QAM) symbols. The turbo
code is used as exemplary purpose only, any well-known coding
scheme such as convolution code, etc. may be used. Each symbol is
modulated by Inverse Fast Fourier Transform (IFFT). In the
transmitter, every N OFDM symbols can be transmitted through N
transmit antennas at the same time.
[0126] From now on, we will analyze for 2L-QAM symbol in terms of
unequal error probability over all of component bits. As known
well, in square constellation such as 2.sup.L-QAM symbol, there
exist symmetries between In-phase and Quadrature components.
Therefore, in order to make analysis simpler, we just consider one
dimension 4-PAM constellation for either In-phase or Quadrature
component of 16-QAM.
[0127] FIG. 12 shows Gray mapping for 16-QAM and FIG. 13 shows
decision boundary to calculate the probability of bit error. It is
assumed that 16-QAM is modulated by gray mapping and 4-PAM
constellation for Most Significant Bit (MSB) and Least Significant
Bit (LSB) to calculate the probability of bit error on each
component bit is subject to one dimension, either In-phase or
Quadrature. All possible 4-PAM symbols are s.sub.1, s.sub.2,
s.sub.3 and s.sub.4. Also these symbols are located at -d.sub.1,
-d.sub.2, -d.sub.1, +d.sub.2, d.sub.1, -d.sub.2 and d.sub.1,
+d.sub.2, respectively, in which 2d.sub.1 means the distance
between two decision boundary within 4-PAM constellation for LSB
and 2d.sub.2 represents the distance between two neighboring
symbols within the 4-PAM constellation.
[0128] The probability of bit error for MSB, which is the
probability of bit error when s1 symbol (i.e. 10) is transmitted,
can be expressed as shown:
MathFigure 17 P e = P { n , .PHI. .gtoreq. d 2 = 2 ( d 1 + d 2 ) 2
} = .intg. d 1 + d 2 .infin. 1 2 .pi..sigma. n 2 1 2 .sigma. n 2 u
2 u = .intg. d 1 + d 2 .sigma. n 2 .infin. 1 2 .pi. 1 2 u 2 u = Q (
d 1 + d 2 .sigma. n 2 ) [ Math . 17 ] ##EQU00057##
[0129] where n means white Gaussian noise with variance
.sigma..sup.2.sub.n, and .PSI. is a unit norm vector along the line
between s.sub.1 and s.sub.4. <x, y> represents the inner
product between x and y. The probability of bit error can be
defined as the probability that the component of the noise vector n
along the line connecting between the two associated symbols is
greater than half the distance along this line. The error
probability of s.sub.4 is the same as that of s.sub.1. In the same
way, when s.sub.2 and s.sub.3 is transmitted, the probability of
bit error is given as shown below.
MathFigure 18 Q ( d 1 + d 2 .sigma. n ) [ Math . 18 ]
##EQU00058##
[0130] Thus, the average probability of bit error for MSB can be
obtained as shown:
MathFigure 19 P e = P e s 1 p ( s 1 ) + P e s 2 p ( s 2 ) + P e s 3
p ( s 3 ) + P e s 4 p ( s 4 ) = 1 4 Q ( d 1 + d 2 .sigma. n ) + 1 4
Q ( d 1 - d 2 .sigma. n ) + 1 4 Q ( d 1 - d 2 .sigma. n ) + 1 4 Q (
d 1 + d 2 .sigma. n ) [ Math . 19 ] ##EQU00059##
[0131] where p(s.sub.i) represents the priori probability of
s.sub.i and P.sub.e|si is the probability of bit error when s.sub.i
is transmitted to the receiver. For the second, the probability for
LSB as well as the way to obtain the probability of bit error for
MSB can be calculated. The probability of bit error for LSB is
thereby given as shown below.
MathFigure 20 1 4 ( Q ( d 2 .sigma. n ) - Q ( 2 d 1 + d 2 .sigma. n
) ) + 1 4 ( Q ( d 2 .sigma. n ) + Q ( 2 d 1 + d 2 .sigma. n ) ) + 1
4 ( Q ( d 2 .sigma. n ) + Q ( 2 d 1 + d 2 .sigma. n ) ) + 1 4 ( Q (
d 2 .sigma. n ) - Q ( 2 d 1 + d 2 .sigma. n ) ) [ Math . 20 ]
##EQU00060##
[0132] Now, it can be known from Equations 19 and 20 that the
component bits in either Inphase or Quadrature have the different
probability of bit error. The difference of probability of bit
error between all possible candidate bits to be either MSB or LSB
can be also observed. Furthermore, when the probability of bit
error is calculated, because there are two decision boundaries for
LSB, the decision boundary is easy to be influenced by the wrong
information from the other side as shown in Equation 20.
[0133] Also, let dd.sub.1=2d.sub.2=2d, we can approximate Equation
19 as follows.
MathFigure 21 1 4 Q ( 3 d .sigma. n ) + 1 4 Q ( d .sigma. n ) + 1 4
Q ( d .sigma. n ) + 1 4 Q ( 3 d .sigma. n ) [ Math . 21 ]
##EQU00061##
[0134] Similarly, assuming that the effect from the other decision
boundary can be ignored, the probability of bit error for LSB can
be represented as shown below.
MathFigure 22 1 4 Q ( d .sigma. n ) + 1 4 Q ( d .sigma. n ) + 1 4 Q
( d .sigma. n ) + 1 4 Q ( d .sigma. n ) [ Math . 22 ]
##EQU00062##
[0135] For HARQ-Chase Combing (CC), it is assumed that the
probability of bit error for the case that the current Log
Likelihood Ratio (LLR) values and the previous LLR values are
combined in a demapper. According to the type of channel
environment, the probability of error for HARQ-CC into three sorts
can be divided as follows.
[0136] a) Additive White Gaussian Noise (AWGN) channel:
MathFigure 23 Q ( 1 2 .sigma. n 2 i = 1 M d i 2 ) [ Math . 23 ]
##EQU00063##
[0137] where d denotes a Euclidean distance at i-th retransmission,
and M means the maximum number of retransmissions.
[0138] b) Fading channel with incoherent demodulation:
MathFigure 24 Q ( 1 2 .sigma. n 2 i = 1 M h i 2 d i 2 ) [ Math . 24
] ##EQU00064##
[0139] where h means the channel value at i-th retransmission. For
a receiver with the incoherent demodulation, the probability of
error of HARQ-CC in fading channel environment is given by Equation
24.
[0140] c) Fading channel with coherent demodulation:
MathFigure 25 Q ( 1 2 .sigma. n 2 i = 1 M [ Re { h i } 2 Re { d i }
2 + Im { h i } 2 Im { d i } 2 ] ) [ Math . 25 ] ##EQU00065##
[0141] The probability of error for HARQ-CC in fading channel
environment with coherent demodulation is given by Equation 25.
[0142] Therefore, we consider both the probability of error for
HARQ-CC and the probability of bit error for MSB and LSB in the
course of chase combining as shown in Table 3.
TABLE-US-00003 TABLE 3 component Initial Transmission
Retransmission (CC) bits MSB LSB MSB LSB 10 1 4 Q ( 3 d .sigma. n 2
) ##EQU00066## 1 4 Q ( d .sigma. n 2 ) ##EQU00067## 1 4 Q ( 3 2 d
.sigma. n ) ##EQU00068## 1 4 Q ( 2 d .sigma. n 2 ) ##EQU00069## 11
1 4 Q ( d .sigma. n 2 ) ##EQU00070## 1 4 Q ( d .sigma. n 2 )
##EQU00071## 1 4 Q ( 2 d .sigma. n ) ##EQU00072## 1 4 Q ( 2 d
.sigma. n 2 ) ##EQU00073## 01 1 4 Q ( d .sigma. n 2 ) ##EQU00074##
1 4 Q ( d .sigma. n 2 ) ##EQU00075## 1 4 Q ( 2 d .sigma. n )
##EQU00076## 1 4 Q ( 2 d .sigma. n 2 ) ##EQU00077## 00 1 4 Q ( 3 d
.sigma. n 2 ) ##EQU00078## 1 4 Q ( d .sigma. n 2 ) ##EQU00079## 1 4
Q ( 3 2 d .sigma. n ) ##EQU00080## 1 4 Q ( 2 d .sigma. n 2 )
##EQU00081## Average 1 2 Q ( 3 d .sigma. n 2 ) + 1 2 Q ( d .sigma.
n 2 ) ##EQU00082## Q ( d .sigma. n 2 ) ##EQU00083## 1 2 Q ( 3 2 d
.sigma. n ) + 1 2 Q ( 2 d .sigma. n ) ##EQU00084## Q ( 2 d .sigma.
n 2 ) ##EQU00085##
[0143] Table 3 shows that the difference of bit error probability
still exists during retransmission. Moreover, although a little
improvement of Signal-to-Noise Ratio (SNR) can be obtained, the
difference of bit error probability between MSB and LSB is further
increased comparing to the case of initial transmission. In case of
MSB, the difference of bit error probability between component bits
which are subjected to MSB is also increased.
[0144] Horizontal bitwise optimization makes a role to reduce the
difference of reliability between coded bits. The performance of
decoding depends on how much the distribution of reliability of
bits can be uniform. As the distribution of bits becomes to be
uniform, more coding gain can be obtained. Therefore, horizontal
bitwise operations are used to maximize coding gain in
retransmission. We regard two operations such as swapping and
inversion as horizontal bitwise mapping. The swapping operation
makes the decision boundary corresponding to MSB and LSB to be
exchanged each other during retransmission.
[0145] FIG. 14 shows swapping operation and FIG. 15 shows inversion
operation. In FIG. 13, it is shown that the number of decision
boundary for MSB and LSB is one and two, respectively. When
employing swapping operation during retransmission, the decision
boundary for MSB and LSB is swapped to result in decreasing the
difference of the bit error probability between MSB and LSB as
shown in FIG. 14. Inversion bitwise operation makes the difference
of bit error probability of candidate component bits to be
decreased. In this case, decision boundary is not exchanged between
MSB and LSB. As shown in FIG. 15, inversion for LSB is only valid.
Therefore, when we need to average the probability of bit error
between candidate component bits subjected to LSB, i.e. for the
applying the inversion operation to a component bit which has not
more than two decision boundary, we need swapping operation with
inversion operation.
[0146] Under employing HARQ, the proposed design criterion to
optimize horizontal bitwise mapping for 2.sup.L-QAM is represented
as shown:
MathFigure 26 { .mu. m } = arg Max .mu. i .di-elect cons. .mu. 1 2
L / 2 - 1 L k 2 L / 2 j L / 2 Q ( 1 2 .sigma. n 2 i = 1 m .mu. i (
d j , k ) 2 ) - 1 2 L / 2 - 1 L k 2 L / 2 j L / 2 Q ( 1 2 .sigma. n
2 i = 1 m - 1 .mu. i ( d j , k ) 2 ) [ Math . 26 ] ##EQU00086##
[0147] where d.sub.j,k represents the distance of k-th candidate
component bit subject to j-th component bit (i.e. MSB or LSB etc.).
.mu..sub.i denotes a horizontal mapping at i-th retransmission.
.mu. indicates the set of horizontal mapping. The number of
combination that we can compose from bitwise swapping and inversion
is .sub.L/2C.sub.2(2.sup.L/2-1). But, we can reduce the number of
combination through the assumptions that bit inversion operation is
only valid when there are more than two decision boundaries and we
can know the difference of bit error probability between component
bits from the approach-based Q-function. In case of 16-QAM, optimal
horizontal mapping set can be calculated by comparisons of between
4 cases at every retransmission.
[0148] Table 4 shows a comparison of bit error probability of the
swapping operation during first retransmission and that of
conventional chase combining.
TABLE-US-00004 TABLE 4 component Chase combining (none + none)
Chase combining (none + swapping) bits MSB LSB MSB LSB 10 1 4 Q ( 3
2 d .sigma. n ) ##EQU00087## 1 4 Q ( 2 d .sigma. n 2 ) ##EQU00088##
1 4 Q ( 10 d .sigma. n ) ##EQU00089## 1 4 Q ( 10 d .sigma. n )
##EQU00090## 11 1 4 Q ( 2 d .sigma. n ) ##EQU00091## 1 4 Q ( 2 d
.sigma. n 2 ) ##EQU00092## 1 4 Q ( 2 d .sigma. n ) ##EQU00093## 1 4
Q ( 2 d .sigma. n ) ##EQU00094## 01 1 4 Q ( 2 d .sigma. n )
##EQU00095## 1 4 Q ( 2 d .sigma. n 2 ) ##EQU00096## 1 4 Q ( 2 d
.sigma. n ) ##EQU00097## 1 4 Q ( 2 d .sigma. n ) ##EQU00098## 00 1
4 Q ( 3 2 d .sigma. n ) ##EQU00099## 1 4 Q ( 2 d .sigma. n 2 )
##EQU00100## 1 4 Q ( 10 d .sigma. n ) ##EQU00101## 1 4 Q ( 10 d
.sigma. n ) ##EQU00102## Average 1 1 2 Q ( 3 2 d .sigma. n ) + 1 2
Q ( 2 d .sigma. n ) ##EQU00103## Q ( 2 d .sigma. n ) ##EQU00104## 1
2 Q ( 10 d .sigma. n ) + 1 2 Q ( 2 d .sigma. n ) ##EQU00105## 1 2 Q
( 10 d .sigma. n ) + 1 2 Q ( 2 d .sigma. n ) ##EQU00106## Average 2
1 4 Q ( 3 2 d .sigma. n ) + 3 4 Q ( 2 d .sigma. n ) .apprxeq. 3 4 Q
( 2 d .sigma. n ) ##EQU00107## 1 2 Q ( 10 d .sigma. n ) + 1 2 Q ( 2
d .sigma. n ) .apprxeq. 1 2 Q ( 2 d .sigma. n ) ##EQU00108##
[0149] The difference of bit error probability between MSB and LSB
is decreased via swapping operation which is a significant
improvement in terms of the average probability of bit error.
[0150] Through our proposed design criterion for horizontal bitwise
mapping, we can find the optimal mapping sets for both 16-QAM and
64-QAM as shown in Table 5.
TABLE-US-00005 TABLE 5 # of trans. Mapping set # of trans. Mapping
set 0 i.sub.1q.sub.1i.sub.2q.sub.2 0
i.sub.1q.sub.1i.sub.2q.sub.2i.sub.3q.sub.3 1
i.sub.2q.sub.2i.sub.1q.sub.1 1 i.sub.3q.sub.3 .sub.2
q.sub.2i.sub.1q.sub.1 2 i.sub.1q.sub.1 .sub.2 q.sub.2 2
i.sub.2q.sub.2 .sub.3q.sub.3i.sub.1q.sub.1 3 i.sub.2q.sub.2 .sub.1
q.sub.1 3 i.sub.1q.sub.1 .sub.2q.sub.2i.sub.3q.sub.3 (a) 16 QAM (b)
64 QAM
[0151] The types of component bits of 64-QAM are divided into three
sorts which are MSB, SB and LSB, respectively. The corresponding
decision boundary on each component bit is introduced in FIG. 16.
FIG. 16 shows decision boundary of 8-PAM. The difference of bit
error probability between component bits within 64-QAM is larger
than that of 16-QAM. Also, the difference of bit error probability
between candidate component bits is significantly increased. For
the first retransmission, averaging the difference of both the
probability of bit error between MSB and LSB and the probability of
bit error between candidate component bits corresponding to SB
through swapping and inversion, respectively, is the best choice in
terms of average bit error probability. For the second
retransmission, employing second swapping operation between MSB and
LSB after first swapping operation between MSB and SB with
inversion operation of MSB has the best average bit error
probability. Finally, in case of third retransmission, we find the
optimal horizontal mapping that employs the inversion for SB
through Equation 26.
[0152] Hereinafter, vertical bitwise mapping is disclosed.
[0153] HARQ schemes for multiple antenna system can exploit a
spatial diversity. But, HARQ schemes such as Space Time Coding
(STC)-HARQ are hard to obtain enough gain by spatial diversity due
to time latency between retransmissions. Furthermore, because
spatial diversity is only related to the increase of received SNR,
the difference of reliability between component bits cannot be
decreased by spatial selective diversity. It even aggravates the
reliability difference.
[0154] In order to analyze the effect of vertical bitwise mapping,
the channel model for 2 by 2 MIMO systems is defined by h.sub.a,b.
The indexes of transmit antenna and retransmission are denoted by a
and b, respectively. Under an assumption that the selectivity of
time channel is small during retransmission, h.sub.a,b can be
represented by Equation 27.
MathFigure 27
h.sub.a,i.apprxeq.h.sub.a,i+1=h.sub.a [Math.27]
[0155] In addition to the case of time selective channel, low
spatial selective channel can be also represented by Equation
28.
MathFigure 28
h.sub.a,i.apprxeq.h.sub.a+1,i=h.sub.a [Math.28]
[0156] Let us assume the low mobility channel under a rich
scattering environment. In addition, 4-PAM for simpler analysis of
vertical bitwise mapping is considered. The probability of error
for a conventional HARQ-CC can be formulated as shown:
MathFigure 29 1 8 Q ( 2 h 1 , i 2 d .sigma. n ) + 1 4 Q ( 2 h 1 , i
2 d .sigma. n ) + 1 8 Q ( 2 h 2 , i 2 d .sigma. n ) + 1 4 Q ( 2 h 2
, i 2 d .sigma. n ) [ Math . 29 ] ##EQU00109##
[0157] where i indicates the index of retransmission. We know that
there is no additional gain extracted from either time or spatial
channel diversity. Considering MSB shuffling between two transmit
antennas during retransmission, the average probability of bit
error is given as shown below.
MathFigure 30 1 4 Q ( h 1 , i 2 + h 2 , i 2 d .sigma. n ) + 1 2 Q (
2 h 1 , i 2 d .sigma. n ) [ Math . 30 ] ##EQU00110##
[0158] In case of LSB shuffling, the average probability of bit
error is calculated as shown below.
MathFigure 31 1 4 Q ( 2 h 1 , i 2 d .sigma. n ) + 1 2 Q ( h 1 , i 2
+ h 2 , i 2 d .sigma. n ) [ Math . 31 ] ##EQU00111##
[0159] Finally, symbol shuffling is expressed as shown below.
MathFigure 32 1 4 Q ( h 1 , i 2 + h 2 , i 2 d .sigma. n ) + 1 2 Q (
h 1 , i 2 + h 2 , i 2 d .sigma. n ) [ Math . 32 ] ##EQU00112##
[0160] In the above analysis, according to the number of bits that
exchanged between antennas, the corresponding average bit error
probability can be calculated by approach-based Q-function as same
as for horizontal bitwise mapping. Through this analysis, we can
know that MSB shuffling is the lowest average bit error
probability. The following one of low bit error probability is LSB
shuffling. The average bit error probability of symbol level
shuffling is lower than that of conventional HARQ-CC, but is higher
than that of 1-bit shuffling. In the similar way, we can compare
with all cases of bit error probability derived by vertical mapping
under a channel model and find a set of optimal vertical
mapping.
[0161] Considering both horizontal bitwise mapping and vertical
bitwise mapping, the proposed design criterion for the optimal
bitwise mapping is represented as shown:
MathFigure 33 { .gamma. m , .mu. m } = arg Max .mu. i .di-elect
cons. .mu. .gamma. i .di-elect cons. .gamma. 1 2 L / 2 - 1 L k 2 L
/ 2 j L / 2 Q ( 1 2 .sigma. n 2 i = 1 m .gamma. i ( h j , k ) 2
.mu. i ( d j , k ) 2 ) - 1 2 L / 2 - 1 L k 2 L / 2 j L / 2 Q ( 1 2
.sigma. n 2 i = 1 m - 1 .gamma. i ( h j , k ) 2 .mu. i ( d j , k )
2 ) [ Math . 33 ] ##EQU00113##
[0162] where .gamma..sub.i denotes a vertical mapping at i-th
retransmission. .gamma..sub.i subject to a set of vertical mapping
.gamma.. As well as the way to reduce the size of set in Equation
26, we can reduce the size of set of candidate vertical mapping in
course of finding the set of optimal vertical mapping as excluding
the cases with low priority. From our design criterion for a
combination of horizontal bitwise mapping and vertical bitwise
mapping, Table 6 shows the set of optimal mapping.
TABLE-US-00006 TABLE 6 # of re- # of re- transmission Optimal
mapping transmission Optimal mapping 0
i.sub.1,1q.sub.1,1i.sub.1,2q.sub.1,2 0
i.sub.1,1q.sub.1,1i.sub.1,2q.sub.1,2i.sub.1,3q.sub.1,3
i.sub.2,1q.sub.2,1i.sub.2,2q.sub.2,2
i.sub.2,1q.sub.2,1i.sub.2,2q.sub.2,2i.sub.2,3q.sub.2,3 1
i.sub.1,2q.sub.1,2i.sub.2,1q.sub.2,1 1 i.sub.2,3q.sub.1,3 .sub.2,2
q.sub.1,2i.sub.2,1q.sub.1,1 i.sub.2,2q.sub.2,2i.sub.1,1q.sub.1,1
i.sub.1,3q.sub.2,3 .sub.1,2 q.sub.2,2i.sub.1,1q.sub.2,1 2
i.sub.1,1q.sub.1,1 .sub.2,2 q.sub.2,2 2 i.sub.2,2q.sub.2,2 .sub.2,3
q.sub.2,3i.sub.2,1q.sub.1,1 i.sub.2,1q.sub.2,1 .sub.1,2 q.sub.1,2
i.sub.1,2q.sub.1,2 .sub.1,3 q.sub.1,3i.sub.1,1q.sub.2,1 3
i.sub.1,2q.sub.1,2 .sub.2,1 q.sub.2,1 3 i.sub.2,1q.sub.1,1 .sub.2,2
q.sub.1,2i.sub.2,3q.sub.1,3 i.sub.2,2q.sub.2,2 .sub.1,1 q.sub.1,1
i.sub.1,1q.sub.2,1 .sub.1,2 q.sub.2,2i.sub.1,3q.sub.2,3 (a) 16-QAM
(b) 64-QAM
[0163] The proposed bitwise mapping is simulated for 16-QAM with
Convolution Turbo Coding (CTC) of rate 1/2 under a 2 by 2 MIMO
channel. The most parameter for performance evaluation follows the
standard specification of IEEE 802.16e. For channel model, the
typical urban (TU) is considered as frequency selective MIMO
channel. Also, time latency between retransmissions is 5 ms and the
number of retransmissions is considered up to 3. Full usage of the
sub-channels (FUSC) which is one of distributed resource allocation
is considered to obtain the coding gain on frequency domain as
sufficient as possible. The type of receiver is based on linear
Minimum Mean Square Error (MMSE) equalizer. For 10 MHz bandwidth,
the number of subcarriers is 1024.
[0164] FIG. 17 shows Frame Error Rate (FER) performances according
to the number of bits shuffled between transmit antennas.
Simulation is performed on all possible combinations which can be
exchanged between transmit antennas by component bits within
64-QAM. Based on the analysis for the vertical bit mapping, we can
know that probability of average bit error for SB shuffling between
transmit antennas has the best performance at 1% FER. The result of
simulation also corresponds to the analysis of probability of bit
error-based Q-function.
[0165] FIG. 18 compares the FER performance of optimal bitwise
mapping and conventional mapping (i.e. HARQ-CC). As opposed to the
typical scheme in which the combining gain is decreased, as the
number of retransmissions is increased, the proposed optimal
bitwise mapping always obtains the combining gain of almost about 2
dB regardless of the number of retransmissions. In case of third
retransmission, the proposed scheme provides about 8 dB gain over
the conventional mapping. This performance is due to the coding
gain which can be obtained by diversity gain extracted from time
selectivity and spatial selectivity through proposed bitwise
mapping.
[0166] III. Constellation Rearrangement for IR Resource
[0167] In case of circular buffer-based HARQ-incremental redundancy
(IR), a redundancy version is transmitted in every retransmission.
Coded bits, which are the same as those previously transmitted, can
be transmitted when the redundancy version is transmitted. In this
case, constellation rearrangement can be performed for these coded
bits. A swapping operation and/or an inversion operation can be
performed for constellation rearrangement.
[0168] FIG. 19 shows HARQ-IR using a swapping operation. A cyclic
buffer 400 stores a code block. An initial data block 410 is a part
of the code block and is initially transmitted. According to a
retransmission request of the initial data block 410, in 1st
retransmission, a 1st retransmission block 420 subsequent to the
initial data block 410 in the cyclic buffer 400 is transmitted. In
2nd retransmission, a 2nd retransmission block 430 consists of a
non-wraparound block 430a and a wraparound block 430b according to
a characteristic of the cyclic buffer 400. The wraparound block
430b denotes a block which has previously been transmitted (or
retransmitted) one time when HARQ is performed. The wraparound
block 430b included in the 2nd retransmission block 430 is wrapped
around with a part of the initial data block 410. By performing the
swapping operation for constellation rearrangement on the
wraparound block 430b, reliability differences for all bits are
averaged. In case of 16-QAM, bit reliabilities of MSB and LSB can
be swapped and then averaged. In case of 64?QAM, bit reliabilities
of MSB, SB, and LSB can be swapped and then averaged. In 3rd
retransmission, since all parts of a 3rd retransmission block 440
belong to a data block which has previously been transmitted one
time, the 3rd retransmission block 440 is a wraparound block and
thus constellation rearrangement is performed thereon. In 4th
retransmission, since all parts of a 4th retransmission block 450
belong to a data block which has previously been transmitted one
time, the 4th retransmission block 450 is a wraparound block and
thus constellation rearrangement is performed thereon.
[0169] The size of cyclic buffer and the size of redundancy version
transmitted according to the maximum number of retransmissions are
for exemplary purposes only, and thus the technical features of the
present invention are not limited thereto. In addition, the
swapping operation for constellation rearrangement is also for
exemplary purposes only, and thus an inversion operation, a spatial
operation, and a combination of the two operations can also be
performed.
[0170] FIG. 20 shows HARQ-IR using a swapping operation and an
inversion operation. A cyclic buffer 500 stores a code block. An
initial data block 510 is a part of the code block and is initially
transmitted. According to a retransmission request of the initial
data block 510, in 1st retransmission, a 1st retransmission block
520 subsequent to the initial data block 510 in the cyclic buffer
500 is transmitted. The 1st retransmission block 520 consists of a
non-wraparound block 520a and a wraparound block 520b according to
a characteristic of the cyclic buffer 500. This is because the
wraparound block 520b included in the 1st retransmission block 520
is wrapped around with a part of the initial data block 510. The
swapping operation for constellation rearrangement is performed on
the wraparound block 520b. In 2nd retransmission, a 2nd
retransmission block 530 consists of one-time wraparound block 530a
and a two-time wraparound block 530b. The one-time wrapped block
530a denotes a block which has previously been transmitted one time
in a wraparound manner. The two-time wrapped block 530b denotes a
block which has previously been transmitted two times in a
wraparound manner. A constellation rearrangement pattern may vary
according to the number of wraparounds. Herein, the one-time
wrapped block 530a uses the swapping operation, and the two-time
wrapped block 530b uses the inversion operation. All parts of a 3rd
retransmission block 540 are 2-time wraparound blocks, and the
inversion operation is performed thereon. The constellation
rearrangement pattern depends on the number of wraparounds and thus
may vary according to the number of wraparounds. However, the
technical features of the present invention are not limited to the
number of wraparounds or the constellation rearrangement
pattern.
[0171] Tables 7 and 8 below show bitwise mapping for each
modulation scheme according to the number of wraparounds in a
process of retransmitting coded bits in a cyclic buffer. 16-QAM is
used in Table 7. 64-QAM is used in Table 8.
TABLE-US-00007 TABLE 7 # of wraparound Mapping rule 0 i.sub.1
q.sub.1 i.sub.2 q.sub.2 1 i.sub.2 q.sub.2 i.sub.1 q.sub.1 2 i.sub.1
q.sub.1 .sub.2 q.sub.2 3 i.sub.2 q.sub.2 .sub.1 q.sub.1
TABLE-US-00008 TABLE 8 # of wraparound Mapping rule 0 i.sub.1
q.sub.1 i.sub.2 q.sub.2 i.sub.3 q.sub.3 1 i.sub.3 q.sub.3 .sub.2
q.sub.2 i.sub.1 q.sub.1 2 i.sub.2 q.sub.2 .sub.3 q.sub.3 i.sub.1
q.sub.1 3 i.sub.1 q.sub.1 .sub.2 q.sub.2 i.sub.3 q.sub.3
[0172] A single antenna system is considered in the HARQ-IR. For
the HARQ-IR of the MIMO system, spatial bitwise rearrangement can
also be performed to obtain spatial diversity. Constellation
rearrangement is performed between transmit antennas whenever coded
bits in a cyclic buffer are transmitted in a wraparound manner in a
retransmission process. For example, when retransmission is first
made in a wraparound manner, a swapping operation is performed on
the signal constellation and also spatial bitwise mapping is
performed between antennas. When retransmission is secondly made in
a wraparound manner, an inversion operation is performed on the
signal constellation and also spatial bitwise mapping is performed
between antennas.
[0173] Tables 9 and 10 below show bitwise mapping for each
modulation scheme according to the number of wraparounds in a
process of retransmitting coded bits in a cyclic buffer in a
multiple antenna system having two transmit antennas. 16-QAM is
used in Table 9. 64-QAM is used in Table 10. In i.sub.a,b and
q.sub.a,b, `a` denotes an antenna index, and `b` denotes an index
indicating coordinates on the constellation.
TABLE-US-00009 TABLE 9 # of wraparound Mapping rule 0
i.sub.1,1q.sub.1,1i.sub.1,2q.sub.1,2
i.sub.2,1q.sub.2,1i.sub.2,2q.sub.2,2 1
i.sub.1,2q.sub.1,2i.sub.2,1q.sub.2,1
i.sub.2,2q.sub.2,2i.sub.1,1q.sub.1,1 2 i.sub.1,1q.sub.1,1 .sub.2,2
q.sub.2,2 i.sub.2,1q.sub.2,1 .sub.1,2 q.sub.1,2 3
i.sub.1,2q.sub.1,2 .sub.2,1 q.sub.2,1 i.sub.2,2q.sub.2,2 .sub.1,1
q.sub.1,1
TABLE-US-00010 TABLE 10 # of wraparound Mapping rule 0
i.sub.1,1q.sub.1,1i.sub.1,2q.sub.1,2i.sub.1,3q.sub.1,3
i.sub.2,1q.sub.2,1i.sub.2,2q.sub.2,2i.sub.2,3q.sub.2,3 1
i.sub.2,3q.sub.1,2 .sub.2,2 q.sub.1,2i.sub.2,1q.sub.1,1
i.sub.1,3q.sub.2,3 .sub.1,2 q.sub.2,2i.sub.1,1q.sub.2,1 2
i.sub.2,2q.sub.2,2 .sub.2,3 q.sub.2,3i.sub.2,1q.sub.1,1
i.sub.1,2q.sub.1,2 .sub.1,3 q.sub.1,3i.sub.1,1q.sub.2,1 3
i.sub.2,1q.sub.1,1 .sub.2,2 q.sub.1,2i.sub.2,3q.sub.1,3
i.sub.1,1q.sub.2,1 .sub.1,2 q.sub.2,2i.sub.1,3q.sub.2,3
[0174] To obtain frequency diversity, when symbols are mapped to
subcarriers in a retransmission process, mapping can be extended
towards noncontiguous subcarriers to perform bitwise rearrangement
on a signal constellation. Frequency diversity can be additionally
obtained by extending mapping towards the noncontiguous
subcarriers. In the mapping from the symbols to the subcarriers, an
additional HARQ gain can be ensured by applying constellation
rearrangement to obtain an average reliability of coded bits.
[0175] FIG. 21 is a block diagram showing an apparatus for wireless
communication according to an embodiment of the present invention.
An apparatus 50 for wireless communication may be a part of a UE.
The apparatus 50 for wireless communication includes a processor
51, a memory 52, a radio frequency (RF) unit 53, a display unit 54,
and a user interface unit 55. The RF unit 53 is coupled to the
processor 51 and transmits and/or receives radio signals. The
memory 52 is coupled to the processor 51 and stores an operating
system, applications, and general files. The display unit 54
displays a variety of information of the UE and may use a
well-known element such as a liquid crystal display (LCD), an
organic light emitting diode (OLED), etc. The user interface unit
55 can be configured with a combination of well-known user
interfaces such as a keypad, a touch screen, etc. With an
implemented physical layer, the processor 51 supports the
aforementioned constellation rearrangement and HARQ. The proposed
method can be implemented by the processor 51.
[0176] All functions described above may be performed by a
processor such as a microprocessor, a controller, a
microcontroller, and an application specific integrated circuit
(ASIC) according to software or program code for performing the
functions. The program code may be designed, developed, and
implemented on the basis of the descriptions of the present
invention, and this is well known to those skilled in the art.
[0177] While the present invention has been particularly shown and
described with reference to exemplary embodiments thereof, it will
be understood by those skilled in the art that various changes in
form and details may be made therein without departing from the
spirit and scope of the invention as defined by the appended
claims. The exemplary embodiments should be considered in
descriptive sense only and not for purposes of limitation.
Therefore, the scope of the invention is defined not by the
detailed description of the invention but by the appended claims,
and all differences within the scope will be construed as being
included in the present invention.
* * * * *