4350 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 64, NO. 9, SEPTEMBER 2015
[14] 3rd Generation Partnership Project (3GPP), “Physical channels and
modulation,” Sophia Antipolis, France, 3GPP TS36.211, Dec. 2009.
[15] M. Trivellato, F. Boccardi, and H. Huang, “On transceiver design and
channel quantization for downlink multiuser MIMO systems with limited
feedback,” IEEE J. Sel. Areas Commun., vol. 26, no. 8, pp. 1494–1504,
Oct. 2008.
[16] Z. Shen, R. Chen, J. G. Andrews, R. W. Heath, Jr., and B. L. Evans,
“Low complexity user selection algorithms for multiuser MIMO systems
with block diagonalization,” IEEE Trans. Signal Process., vol. 54, no. 9,
pp. 3658–3663, Sep. 2006.
[17] A. M. Tulino and S. Verdu, “Random matrix theory and wireless
communications,” in Foundations and Trends in Communications and
Information Theory. Delft, The Netherlands: Now, 2004.
A Reduced-Complexity Demapping Algorithm for
BICM-ID Systems
Jiandong Tan, Qi Wang, Chen Qian,
Zhaocheng Wang, Senior Member, IEEE,
Sheng Chen, Fellow, IEEE, and Lajos Hanzo, Fellow, IEEE
Abstract—As a highly efficient decoding and demodulation scheme,
bit-interleaved coded modulation (BICM) is widely adopted in modern
communication systems. In order to enhance the attainable spectral
efficiency, usually, high-order modulation schemes are used for BICM
systems. When BICM is combined with iterative decoding (BICM-ID),
it is capable of further improving the achievable receiver performance.
However, the complexity of the standard max-sum approximation of the
maximum a posteriori probability in log-domain (Max-Log-MAP) invoked
by the iterative demapper is on the order of 2
m
or O
2
m
for a 2
m
-ary
modulation constellation having m bits per symbol, which may become
excessive for high-order BICM-ID systems. The existing simplified algo-
rithms employed for noniterative demappers are based on exploiting the
constellation’s symmetry, which is no longer retained upon the introduc-
tion of the aprioriinformation in BICM-ID systems. Hence, in this paper,
a simplified iterative demapping algorithm is proposed to substantially
reduce the demapping complexity for a binary-reflected Gray-labeled
constellation. Our detailed analysis shows that the simplified demapping
scheme proposed for BICM-ID reduces the computational complexity
to O(m). We demonstrate that this dramatic computational complexity
only imposes modest performance degradation with respect to that of the
optimal high-complexity Max-Log-MAP scheme.
Index Terms—Bit-interleaved coded modulation with iterative decod-
ing (BICM-ID), iterative demapper, maximum a posteriori probability in
log-domain demapping, pulse amplitude modulation (PAM), quadrature
amplitude modulation (QAM).
Manuscript received May 26, 2014; revised September 13, 2014; accepted
October 29, 2014. Date of publication October 31, 2014; date of current
version September 15, 2015. This work was supported in part by the National
Key Basic Research Program of China under Grant 2013CB329203, by the
National Natural Science Foundation of China under Grant 61271266, by the
National High Technology Research and Development Program of China under
Grant 2014AA01A704, and by Beijing Natural Science Foundation under Grant
4142027. The review of this paper was coordinated by Dr. S. K. Jayaweera.
J. Tan, Q. Wang, C. Qian, and Z. Wang are with Tsinghua National
Laboratory for Information Science and Technology (TNList), Department of
Electronic Engineering, Tsinghua University, Beijing 100084, China.
S. Chen is with the School of Electronics and Computer Science, University
of Southampton, Southampton SO17 1BJ, U.K., and also with King Abdulaziz
University, Jeddah 21589, Saudi Arabia.
L. Hanzo is with the School of Electronics and Computer Science, University
of Southampton, Southampton SO17 1BJ, U.K.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TVT.2014.2366195
I. INTRODUCTION
The original concept of bit-interleaved coded modulation (BICM)
was presented by Zehavi [1], who found that the performance of
coded modulation over Rayleigh fading channels can be considerably
improved by adopting bit interleaving. Based on his work, a set of
theoretical modeling and analysis tools has been developed for BICM
for other applications as well [2]. In 1998, Li and Ritcey [3] improved
BICM with the aid of iterative decoding (BICM-ID). In the same
year, ten Brink et al. [4] provided further insights on enhancing the
performance of BICM-ID at reduced complexity. Therefore, state-
of-the-art communication systems often employ BICM to improve
the attainable performance, as exemplified by the second-generation
digital terrestrial television broadcasting standard (DVB-T2) system
[5]. For the sake of achieving high bandwidth efficiency, typically
high-order quadrature amplitude modulation (QAM) [6] is recom-
mended for both noniterative BICM and BICM-ID systems. The Gray-
labeled mapping is also widely applied to high-order modulation
constellation schemes to reduce the bit errors since two adjacent Gray-
labeled constellation points differ in only one bit [7]. The DVB-T2
standard [5], for example, supports binary-reflected Gray-labeled mod-
ulation. However, high-order BICM and BICM-ID systems impose
high-complexity when adopting the optimal log-domain maximum
a posteriori probability (Log-MAP) demapping algorithm [8]. Al-
though the max-sum approximation of the Log-MAP (Max-Log-MAP)
algorithm [9] is capable of substantially reducing the associated com-
putational complexity, it is still on the order of O
2
m
,where2
m
is
the size of the modulation constellation employed that contains m bits
per symbol.
Several simplifications of the Max-Log-MAP algorithm have been
proposed for high-order constellations. Tosato and Bisaglia [10] found
that 2
m
-QAM symbols relying on Gray labeling can be divided
into two independent 2
m/2
-ary pulse amplitude modulation (PAM)
constellations, and with the aid of this partitioning, the complexity of
the noniterative demapper can be reduced to O
2
m/2
without any
performance degradation. In [10], the complexity of the simplified
QAM demapper is further reduced to O
m
by adopting a piece-
wise linear approximation, but this simplification results in modest
performance degradation. Since the aprioriinformation necessary
for iterative demapping is not included in the simplified algorithms in
[10], it cannot be readily applied to BICM-ID systems. The simplified
demapping algorithm in [11] was conceived for universal Gray-labeled
constellations, which reduces the complexity to O
m
by exploiting
the symmetry of Gray-labeled constellations while attaining exactly
the same performance as the Max-Log-MAP algorithm for QAM,
PAM, and phase-shift keying (PSK) constellations. However, this algo-
rithm cannot be applied to the iterative demapper since the symmetry
of Gray-labeled constellations is no longer satisfied after taking into
account the input aprioriinformation. Further efforts invested in
reducing the complexity of BICM-ID systems were reported in [12],
where hard-feedback-based iterative decoding was used to reduce the
complexity of BICM-ID combined with frequency-shift keying modu-
lation, whereas in [13], a mapping scheme combined with low-density
parity-check (LDPC) coding is proposed to merge the demapping
operation into the process of LDPC decoding. However, the methods
in [12] and [13] are only applicable to orthogonal modulation schemes.
By only allowing the extrinsic information for systematic bits to be fed
back to the demodulator, the scheme proposed in [14] was shown to
reduce the complexity of BICM-ID schemes, but the objective of this
scheme is to reduce the decoding complexity of turbo coding rather
than to simplify the demapping algorithm conceived for universal
modulation and coding schemes.
0018-9545 © 2014
IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.