1216 IEEE COMMUNICATIONS LETTERS, VOL. 17, NO. 6, JUNE 2013
16-QAM Golay Complementary Sequence Sets with Arbitrary Lengths
Fanxin Zeng, Xiaoping Zeng, Zhenyu Zhang, and Guixin Xuan
Abstract—In an orthogonal frequency division multiplexing
(OFDM) communication system, the reduction of peak-to-mean
envelope power ratio (PMEPR) is one of crucial issues. It is all
known that in an OFDM system, Golay complementary sequence
(CS) pairs result in lowest upper bound 2 of PMEPR. Hence, this
letter focuses on constructions of 16-QAM Golay CS sets. The
proposed two constructions can produce the required sequences
with arbitrary lengths, including lots of unknown ones in the
existing references, if there exist the quaternary Golay CS sets
of such lengths, and the PMEPRs of the Golay CS pairs from
both are at most 2. As a consequence, the resultant sequences by
this letter can be used as the inputs of using non-radix-2 inverse
fast Fourier transform (IFFT) in the time domain OFDM signals.
Index Terms—Complementary sequences, 16-QAM constella-
tion, PMEPR, sequence length.
I. INTRODUCTION
R
ECENTLY, QAM (quadrature amplitude modulation)
Golay complementary sequence (CS) sets/pairs have
received increased attention due to the fact that such se-
quence sets/pairs f ully suppress multiple access interference
(MAI) in a CDMA communication system [1], and result in
lowest upper bound 2 of peak-to-mean envelope power ratio
(PMEPR) in an orthogonal frequency division multiplexing
(OFDM) communication system [2]. To reduce PMEPR, many
constructions were presented [3]-[14]. More clearly, [3]-[5]
investigated the upper bounds of PMEPR of several construc-
tions of QAM sequences without CS pairs, and showed that
each value of them is g reater than 2, which is the reason why
construction of QAM Golay CS pairs is one of all-important
issues in an OFDM system. Under motivation of this sense,
based on QPSK GDJ sequences [2], [6]-[7] constructed 64-
QAM Golay CS pairs, and [8]-[9] proposed 16-QAM Golay
CS pairs, all of which are included in a general construction
in [10]. Furthermore, analogous to traditional Golay CS pair,
QAM Golay CS pairs are generalized to QAM periodic CS
sets as well [11]-[14]. In general, a CS pair is fit for application
Manuscript recei ved January 18, 2013. The associate editor coordinating
the review of this letter and approving it for publication was L. Dolecek.
F. X. Zeng is with the College of Communication Engineering, Chongqing
Uni versity, Chongqing 400044, China, and also with the Chongqing Ke y Lab-
oratory of Emergency Communication, Chongqing Communication Institute,
Chongqing 400035, China (e-mail: fzengx@yahoo.com.cn).
X. P. Zeng is with the College of Communication Engineering, Chongqing
Uni versity, Chongqing 400044, China (e-mail: zxp@cqu.edu.cn).
Z. Y. Zhang and G. X. Xuan are with the Chongqing Ke y Laboratory of
Emergency Communication, Chongqing Communication Institute, Chongqing
400035, China (e-mail: {cqzhangzy, guixinxuan}@yahoo.com.cn).
This work was supported by the National Natural Science Founda-
tion of China (NSFC) under Grants 60872164, 61002034, 61171089 and
61271003, and the Ministry of Industry and Information Technology of China
(No.Equipment[2010]307).
Digital Object Identifier 10.1109/LCOMM.2013.042313.130148
to an OFDM system, and a CS set (including a pair) is
preferred in a CDMA system.
Up to now, to the best o f the authors’ knowledge, the
existing QAM Golay CS pairs [6]-[10] must be of length 2
m
(the integer m ≥ 2), which is in favor of usage of radix-2
inverse fast Fourier transform (IFFT) algorithms [15] so as
to process the time domain OFDM signals rapidly. However,
there exist non-radix-2 IFFT algorithms, such as prime factor
algorithms, so as to process such signals with other lengths
[15]. Unfortunately, there don’t exist inputs for non-radix-
2 IFFT algorithms, in other words, QAM Golay CS pairs
without power 2 of length are unknown at present. In addition,
the theory of QAM Golay CS pairs is not perfect if not all
lengths are known. This motivates our work to explore such
sequences with arbitrary lengths.
This letter is organized as follows. In Section II, we give
some necessary preliminaries. In Section III, new 16-QAM
Golay CS sets are constructed from known quaternary Golay
CS sets with various lengths, and the upper bound of their
PMEPR is derived. Finally, concluding remark is given in
Section IV.
II. P
RELIMINARIES
A. Golay Complementary Sequences
Let u
r
=(u
r
(0),u
r
(1),u
r
(2), ···,u
r
(N − 1)) be a
complex sequence with length N. For a time shift τ, we refer
to
C
u
r
,u
r
(τ)=
⎧
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎩
N−1−τ
k=0
u
r
(k)u
r
(k + τ)0≤ τ ≤ N − 1
N−1+τ
k=0
u
r
(k − τ)u
r
(k)1− N ≤ τ<0
0 |τ|≥N,
(1)
as an aperiodic autocorrelation function of the sequence u
r
,
where
x denotes the complex-conjugate of x.
Let a complex sequence set u
=(u
0
,u
1
, ···,u
M−1
)
consist of M sub-sequences with each of length N .Ifthe
sum of aperiodic autocorrelations of all sub-sequences in the
sequence set u
satisfies
M−1
r=0
C
u
r
,u
r
(τ)=
⎧
⎨
⎩
M−1
r=0
E
u
r
τ =0
0 other,
(2)
we refer to the sequence set u
as a Golay CS set, where
E
u
r
is the energy of the sub-sequence u
r
, namely, E
u
r
=
N−1
t=0
|u
r
(t)|
2
. In particular, if M =2, we call the sequence
pair u
a Golay CS pair.
1089-7798/13$31.00
c
2013 IEEE