See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/326886938

Preamble Design with Interference Cancellation for Channel Estimation in

MIMO-FBMC/OQAM Systems

Article  in  IEEE Access · December 2018

DOI: 10.1109/ACCESS.2018.2864221

CITATIONS READS

10 167

4 authors:

Han Wang Lingwei Xu

Yichun University Qingdao University of Science and Technology
28 PUBLICATIONS   150 CITATIONS    79 PUBLICATIONS   404 CITATIONS   

SEE PROFILE SEE PROFILE

Xianpeng Wang Sohail Taheri

Hainan University Viavi Solutions
83 PUBLICATIONS   695 CITATIONS    11 PUBLICATIONS   76 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

MIMO wireless systems View project

Cooperative communication networks View project

All content following this page was uploaded by Sohail Taheri on 17 September 2018.

The user has requested enhancement of the downloaded file.

 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2864221, IEEE Access

Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2018.Doi Number

Preamble Design with Interference Cancellation

for Channel Estimation in MIMO-FBMC/OQAM
Systems
Han Wang1, (Member, IEEE), Lingwei Xu2, Xianpeng Wang3, and Sohail Taheri4
1
College of Physical Science and Engineering, Yichun University, Yichun 336000, China
2
Department of Information Science and Technology, Qingdao University of Science and Technology, Qingdao 266061, China
3
State Key Laboratory of Marine Resource Utilization in South China Sea, Hainan University, Haikou 570228, China
4
Gobham Wireless, Stevenage SG1 1AN, United Kingdom
Corresponding author: Han Wang, Lingwei Xu (e-mail: hanwang1214@126.com;gaomilaojia2009@163.com).
This work of H. Wang was supported in part by the Jiangxi Provincial Education Office Science and Technology Project under Grant GJJ170915. This
work of L. W. Xu is supported in part by Shandong Province Natural Science Foundation and Shandong Province Postdoctoral Innovation Project under
Grant ZR2017BF023,201703032. This work of X. P. Wang is supported in part by the National Natural Science Foundation of China under Grant 61701144.

ABSTRACT Multiple-input multiple-output filter bank multicarrier with offset quadrature amplitude
modulation (MIMO-FBMC/OQAM) system utilizes the multiple antennas technology, and it can provide
more advantages than conventional FBMC/OQAM system. However, due to the inherent imaginary
interference, the convention preamble based channel estimation (CE) methods in MIMO-FBMC/OQAM
systems cannot achieve high CE performance and mostly suffer from high peak to average power ratio
(PAPR). Motivated by these problems, in this paper, we focus on the efficient preamble design for CE in
MIMO-FBMC/OQAM systems. We propose an extended preamble structure which exploits the symmetry
pattern to cancel interference and the interference weights in the symmetric structure is taken into account.
Since the preamble length is extended, the additional interference effect from the middle of the preamble
structure is also considered. The simulation results demonstrate that the proposed preamble method has low
PAPR, better bit error ratio (BER) and better mean square error (MSE) performance compared to the
conventional preamble methods.

INDEX TERMS Channel estimation, FBMC/OQAM, multiple-input multiple-output (MIMO), preamble.

I. INTRODUCTION mitigate the drawbacks of CP-OFDM, and has drawn

Over the past decades, multicarrier communication
technique has been widely used in many communication
systems. Cyclic prefix-orthogonal frequency division
multiplexing (CP-OFDM) technique is certainly one of the
most adopted multicarrier techniques in many wireless
communication systems [1], [2]. This is because of its
immunity to multipath fading, simplicity of channel
estimation and suitability for multiple input multiple output
(MIMO) systems. However, it suffers from some inherent
shortcomings, such as high out of band radiation, sensitivity
to carrier frequency offset (CFO), and cyclic prefix (CP)
overhead, and the insertion of CP will cause a loss in spectral
efficiency.
As an alternative modulation scheme to CP-OFDM, filter
bank multicarrier employing offset quadrature amplitude
modulation without inserting CP, called FBMC/OQAM, can
increasing attentions in recent years [3]–[6]. FBMC/OQAM
utilizes time frequency localization (TFL) property pulse
shaping [7], [8] via an IFFT/FFT-based filter bank, and
staggered OQAM symbols, i.e.. real symbols at twice the
symbol rate of FBMC/QAM, are loaded on the subcarriers[9].
Despite the higher complexity compared to CP-OFDM,
FBMC/OQAM can provide increased robustness against
CFO, remarkably reduced out of band emissions, and better
spectral containment in bandwidth sensitive applications [10].
However, unlike OFDM signals that transmit complex-
valued symbols at a given symbol rate, FBMC/OQAM
signals transmit real-valued symbols at twice this symbol rate.
As FBMC/OQAM only holds the orthogonality in the real
field, the received symbols are contaminated with pure
imaginary interference resulting from neighboring
subcarriers and symbols.

2169-3536 © 2017 IEEE. Translations and content mining are permitted for academic research only.
VOLUME XX, 2017 Personal use is also permitted, but republication/redistribution requires IEEE permission. 1
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2864221, IEEE Access

The inherent interference complicates signal processing
tasks, such as channel estimation, and conventional channel
estimation methods for CP-OFDM systems cannot be
directly applied to FBMC/OQAM systems. The inherent
interference becomes a source of problem in channel
estimation and equalization processes, especially in MIMO-
FBMC/OQAM systems.
In recent years, a number of preamble based channel
estimation methods for single-input single-output (SISO)
FBMC have been studied in literature [11]-[14]. Interference
approximation method (IAM) has been first introduced in
[11]. The method is capable of computing an approximation
of the interference from neighboring symbols. Then, several
variants of IAM have also been proposed, including IAM-R,
IAM-I, IAM-C and E-IAM-C [12]-[14]. In the variants of
IAM, R denotes real-valued pilots, I and C stand for
imaginary and complex pilots, and E represents the extended
IAM-C. All of these methods can be characterized as aiming
at constructively exploiting interference to improve
estimation performance. In contrast to the IAM approach,
another approach, called interference cancellation method
(ICM), was presented in [15]-[17] by designing the preamble
structure to cancel or avoid the interference. By observing the
interference mainly from the nearest subcarriers, a more
effective preamble structure based ICM approach has been
suggested in [16]. Inspired by IAM and ICM, the authors [18]
proposed an efficient preamble structure for channel
estimation in FBMC/OQAM systems. The simulations
demonstrate that the proposed preamble structure can obtain
better performance than the two convention preamble
approaches.
In [19], the authors provided an overview of MIMO signal
processing in FBMC/OQAM systems. The conclusion is that
FBMC/OQAM can benefit from the advantages and gains of
incorporating MIMO processing, and the application of
FBMC/OQAM to a MIMO context is becoming mature. It is
noted that significant research attention on MIMO-
FBMC/OQAM systems has been focused on how to repress
the influence of multi-antenna interference for effective
channel estimations in the presence of inherent imaginary
interference [20]-[24]. In [20], the IAM variants were
considered and their MIMO extensions with their relative
advantages and practical issues were investigated. The
extended MIMO preamble is designed by simply selecting a
method can outperform than IAM methods. However, the
channel estimation performance is still inferior to the ICM
method. Recently, an efficient novel training sequences
channel estimation approach has been proposed in [24]. The
authors proposed “complex training sequence
decomposition” to facilitate the reconstruction of the
complex-field orthogonality of MIMO-FBMC/OQAM
signals. The channel estimation performance of the proposed
method is comparable to that of the IAM-C method.
In this paper, we investigate the preamble based channel
estimation for MIMO-FBMC/OQAM systems by exploiting
the preamble symmetry pattern to cancel interference and
taking into account the interference weights in the symmetric
structure. The main contribution of this paper is that a novel
efficient preamble structure for channel estimation in MIMO-
FBMC/OQAM systems has been proposed. The proposed
preamble structure exploits the symmetry pattern to obtain
interference cancellation gain and jointly considers the
interference weights to obtain a lower interference power.
Furthermore, additional interference affecting the middle of
preamble structure is also considered. Therefore, the channel
estimation performance of the proposed preamble is less
affected by interference. Compared to the conventional
extended preamble based channel estimation methods, the
proposed preamble can obtain better BER and MSE
performance, as well as low PAPR.
The remainder of this paper is organized as follows.
Section II presents the MIMO-FBMC system model and the
conventional preamble based channel estimation methods. In
Section III, an efficient preamble structure is proposed.
Numerical simulations are presented in Section IV. Finally,
Section V gives the concluding remarks.
II. MIMO-FBMC/OQAM SYSTEM
A. SYSTEM MODEL
FBMC/OQAM systems are implemented by synthesis and
analysis filter banks in transmitter and receiver side,
respectively. The channel is assumed to be invariant in the
duration of the preamble. At first, we give the SISO-
FBMC/OQAM system model. The transmitted signal in
SISO-FBMC/OQAM systems can be written in the
following form
N −1

corresponding Hadamard transform matrix. In [21], the s (t ) = ∑ ∑d m,n g m , n (t ) (1)

m=0 n
authors provided a closed-form estimate of the MIMO
channels while making no assumption on its frequency where d m , n are real valued OQAM symbols, and
selectivity and using a very short preamble for training. The g m , n (t ) represents the synthesis basis which is obtained by
preamble is optimized with respect to the channel estimation the time-frequency translated version of the prototype
mean squared error. The extended IAM preambles can only function g (t ) in the following way
jφm ,n
achieve low channel estimation performance. Besides, an t ) g (t − nτ 0 )ei 2π mF0t e
g m , n (= (2)
overhead reduced preamble based channel estimation
methods for MIMO-FBMC/OQAM systems has been with N being the even number of subcarriers,
proposed in [22], [23] by employing inherent interference =
F0 1=/ T0 1 / 2τ 0 the subcarrier spacing, and φm , n an
cancellation at the transmitter side. The proposed preamble additional phase term. T0 denotes the OFDM symbol

VOLUME XX, 2017 9

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2864221, IEEE Access

duration, and τ 0 denotes the time offset between the real =

y p,q H p,q d p,q + j ∑ H p,q d p + m,q + n g
p,q
p + m,q + n
and imaginary parts of the OQAM symbols. The double ( m , n )∈Ω1,1
(8)
subscript (.) m , n denotes the (m, n) -th frequency-time (FT) H p + m,q + n
∑
p,q
+j d p + m,q + n g p + m,q + n
point, m is the subcarrier index and n is the OQAM symbol ( m , n )∉Ω1,1 H p,q
time index.
p,q
The pulse g is designed so that the associated subcarrier With the increase of m and n , and g becomes
p + m,q + n
functions g m , n are orthogonal in the real field, that is
very close to zero. Then, with noise taken into
ℜ{∑ g m , n (t ) g *p , q (t )} =
ℜ{ g m , n g p , q } = δ m , pδ n , q (3) consideration, (8) can be rewritten as
t y p , q ≈ H p , q c p , q +η p , q (9)
where δ i , j is the Kronecker delta, δ m , p = 1 if m = p and
c= d p,q + j ∑
p,q
where p,q d p + m,q + n g p + m,q + n is the
δ m , p = 0 if m ≠ p . We can find that even in the distortion- ( m , n )∈Ω1,1

free channel and with perfect time and frequency u p ,q

synchronization, there still will be some purely imaginary virtual transimtted symbol at ( p, q ) ,with u p , q being the
inter-carrier interference at the output of analysis filter bank.
imaginary interference from the neigboring FT point.
Thus, we set interference weights
Assuming that the symbol on the receiver FT point
= − j g m,n g p,q
p,q
g m,n
(4) ( p, q) is a prior known preamble symbol, the preamble
based channel estimation is
with g m,n g p,q a purely imaginary term y p,q η p,q
Hˆ p , q = ≈ H p,q + (10)
for (m, n) ≠ ( p, q ) . c p,q c p,q
After passing through the channel, the received signal
with an additive noise can be written as Note that the larger the power of c p , q , the better the channel
M −1 estimation will be.
=r (t ) ∑ ∑d
m=0 n
m,n g m , n (t ) H m , n (t ) + η (t ) (5) TABLE 1
WEIGHTS OF INTERFERENCE ON THE FIRST ORDER NEIGHBORS
with q −1 q q +1
τ max
p −1 (−1) p δ −β (−1) p δ
H m , n (t ) = ∫ h(t ,τ )e −2 jπ mF0τ dτ (6)
0 p −(−1) p γ 1 (−1) p γ
where h(t ,τ ) is the channel impulse response. H m , n (t ) is the p +1 (−1) p δ β (−1) p δ
complex response of the channel at instant t . We assume
that we have a flat fading channel at each subcarrier, which The above SISO-FBMC/OQAM signal formulation can be
means that the channel is constant during the duration of easily extended to the MIMO case. Consider an
the prototype length, then H m , n (t ) = H m , n . N t × N r MIMO-FBMC/OQAM system, as shown in Fig. 1.
When the output is at the p -th subcarrier and q -th The data is send to the antennas after serials to parallel
OQAM symbol, the output signal, in a noiseless case can be conversion. FBMC/OQAM modulation is achieved based on
expressed as the synthesis filter bank in each antenna, and it demodulation
y p,q = r g p,q is achieved based on analysis filter bank. After passing the
(7) MIMO channels, the data can be recovered by
=
y p,q H p,q d p,q + j ∑
p,q
H p + m,q + n d p + m,q + n g p + m,q + n FBMC/OQAM modulation and MIMO equalization. Then,
( m , n ) ≠ (0,0)
with the same assumptions as previously, one can write an
A common definition is that only the first-order equation analogous to (9) for each receive
neighborhood Ω1,1 of a given FT antenna j = 1, 2, , N r ,
y pj , q ≈ H pj ,,iq (d ip , q + ju ip , q ) +η pj , q (11)
point ( p, q ) =
, Ω1,1 {(m, n), m ≤ 1, n ≤ 1, (m, n) ≠ (0, 0)} ,
where d ip , q are real valued OQAM symbol at the ith transmit
causes the interference. Table 1 shows the number of
p,q
coefficients on the first order neighbors of the point antenna, H pj ,,iq is the channel response from the ith transmit
g m,n
antenna to the jth receive antenna, u ip , q is the imaginary part
( p, q) . The weights of interfernce for the filter
are γ = 0.5004 , β = 0.3183 ,and δ = 0.2501 [18]. We can of the interference, and η pj , q is the corresponding noise
rewrite (7) as component. The MIMO-FBMC/OQAM signal model can be
expressed in the form of the matrix
y p,q ≈ H p,q cp,q +ηp,q (12)

VOLUME XX, 2017 9

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2864221, IEEE Access

1 H m1,1,n 1
d 1
m,n FBMC/OQAM FBMC/OQAM dˆm1 ,n
Modulation Demodulation
H mN,rn,1

. .
. . MIMO
H m1,,Nnt Equalization
. .
Nt Nr
d mN,tn FBMC/OQAM H mN,rn, Nt FBMC/OQAM
dˆmN,tn
Modulation Demodulation
MIMO Channel

FIGURE 1. The MIMO-FBMC/OQAM system.

according to (12), the receiving pilot symbol at q = 2, 4 can
where be written as
T T
y p,q =  y1p , q y 2p , q  y pN,rq  , ηp,q = η 1p , q η p2 , q  η pN,rq  , and c1 c1p ,4 
=  y p, 2 y p, 4  H p, 2  2p ,2 2 
+  ηp, 2 ηp, 4  (13)
 H 1,1
p,q H 1,2
p,q  H 1,p ,Nqt  c p ,2 c p ,4 
 2,1 
H H p2,2, q  H p2,, qNt  For MIMO-FBMC/OQAM systems with good TFL
H p,q =  p,q is the MIMO channel
     
characteristic prototype filters, the inherent interference being
 Nr ,1  mostly came from the first order FT neighbors, we can easily
 H p , q H pN,rq,2  H pN,rq, Nt  find that c1p ,2 = c1p ,4 = c 2p ,2 = −c 2p ,4 ≡ c p and equation (13) can
frequency response matrix. c=
p,q dp,q + jup,q is the be rewritten as
equivalent transmission symbol vector form of the MIMO- c p cp 
 y p, 2 y p, 4  H p, 2 
= + η η 
−c p   p, 2 p, 4 
T
FBMC/OQAM system, with dp,q =  d 1p , q d p2, q  d pN,tq  , c p (14)
=H p, 2 c p A +  ηp, 2 ηp, 4 
T
up,q = u1p , q u 2p , q  u pN,tq  .
where A is the orthogonal matrix
B. PREAMBLE BASED CHANNEL ESTIMATION 1 1 
A=  (15)
For a N r × N t spatial multiplexing MIMO-FBMC/OQAM 1 −1
system, it is clear from (12) that one will need at least
The pilot c p can be pre-calculated as in the SISO case.
N t number of nonzero pilot symbols to estimate the channel
frequency response matrix. Assume the channel is time Hence, the estimate of the channel frequency response matrix
invariant during a period of 2 N t + 1 FBMC/OQAM symbols. at the subcarrier p can be expressed as
For simplicity, let us consider a 2-by-2 antenna scenario.
Hˆ =  y y  1 A −1
0 1 0 -1 0 0 1 0 1 0 p, 2  p, 2 p, 4  c
p
0 -j 0 j 0 0 -j 0 -j 0 (16)
1
0 -1 0 1 0 0 -1 0 -1 0 = H p, 2 +  ηp, 2 ηp, 4  A
0 j 0 -j 0 0 j 0 j 0 2c p
0 1 0 -1 0 0 1 0 1 0 From equation (16), we can see that noise enhancement is
0 -j 0 j 0 0 -j 0 -j 0
again controlled by the magnitude of the pseudo pilot c p . E-
0 -1 0 1 0 0 -1 0 -1 0
0 j 0 -j 0 0 j 0 j 0 IAM-C preamble structure can also be extended into MIMO-
FBMC/OQAM system to improve the channel estimation
antenna 1 antenna 2
performance, and Fig. 3 shows the extension preamble
FIGURE 2. IAM-C preamble structure for a 2x2 MIMO-FBMC/OQAM
system, with N =8 .
structure. However, it should be noted that equation (14)
relies on the assumption that inherent interference only
In SISO IAM-C case, the preamble length is taken equal to
contributed from the first order neighbors. In fact, there is
3τ 0 , d p , q is located in the middle of the preamble. In MIMO nonnegligible interference to the FT point from
case, one antenna can employ the same preamble used in the ( p ± 1, q ± 2) as well. Therefore, the pilot matrix in (14) is
SISO case, but with a repetition, as shown in Fig. 2. The
other antenna employs the same preamble, but with changed c1 c1p 
equal to  2p  , with c p corresponding to the ith transmit
i

signs at the second nonzero FBMC/OQAM symbol. Then, c p c 2p 

VOLUME XX, 2017 9

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2864221, IEEE Access

antenna and c1p ≠ c 2p .This means that the two columns of middle pilot is 1, the preamble interference has
magnitude 1 + j 2( β − γ ) . When the middle pilot is -1, the
2
the channel frequency response matrix are not estimated with
the same accuracy. The pilot matrix is no longer unitary, and preamble interference has magnitude 1 + j 2 β . In antenna 1,
2

it only has nearly orthogonal rows. The accuracy of channel

estimation is affected. it can be seen that, in the third and the fourth symbols, there
j 1 -j j 1 -j j 1 -j -j -1 j
exists additional interference contributed from its respective
-1 -j 1 -1 -j 1 -1 -j 1 1 j -1
neighborhood. In the third symbol, the preamble interference
-j -1 j -j -1 j -j -1 j j 1 -j at the first subcarrier has magnitude 1-j (2γ +β + 2δ ) ; the
1 j -1 1 j -1 1 j -1 -1 -j 1 interference at the second subcarrier has magnitude 1 + j 2 β ;
j 1 -j j 1 -j j 1 -j -j -1 j the interference at the third subcarrier has
-1 -j 1 -1 -j 1 -1 -j 1 1 j -1 magnitude 1 + j 2(δ − β − γ ) ; and the interference at the
-j -1 j -j -1 j -j -1 j j 1 -j
fourth subcarrier has magnitude 1 + j 2( β + γ ) . In the fourth
1 j -1 1 j -1 1 j -1 -1 -j 1
symbol, the preamble interference at the first subcarrier has
antenna 1 antenna 2
magnitude 1+（ j 2γ +β）; the preamble interference at the
FIGURE 3. E-IAM-C preamble structure for a 2x2 MIMO-FBMC/OQAM
system, with N =8 . second subcarrier has magnitude 1 + j 2(γ − β ) ; the preamble
The preamble can be easily generalized to systems with interference at the third subcarrier has magnitude 1+j 2γ ; and
more than two transmit antennas, with N t being a power of 2. the preamble interference at the fourth subcarrier has
For an N t × N r system, the preamble sequence of each magnitude 1 + j 2( β − γ ) . With the same way, we can also
antenna is composed of 3N t FBMC/OQAM symbols, and calculate the interference magnitude of different subcarriers
the number of symbols per column is the same as that of the at the third and fourth symbols in antenna 2. For the third and
system subcarriers. Three preamble symbols constitute a fourth symbol interference effects, the extended NPS
group of preamble symbols. The first and third preamble structure has partly greater interference weight in antenna 1
sequences of the preamble symbol are all 1, the second compared with the extended ICM method, but interference
column preamble sequence is 1, and the -1 crossover weight in antenna 2 is relatively low. When applying NPS to
sequence. The corresponding preamble structure is selected MIMO-FBMC/OQAM system, channel estimation
according to the number of transmit antennas. Simply select performance will be affected by different interference
the corresponding Hadamard matrix ANt (e.g. for an distributions.
1 -1 1 1 -1 1 1 -1 1 -1 1 -1
1 1 1 1  1 1 -1 1 1 -1 1 1 -1 -1 -1 1
1 −1 1 −1
4 × 4 system, A4 =   ) and design the -1 1 -1 -1 1 -1 -1 1 -1 1 -1 1
1 1 −1 −1 1 -1 1 1 -1 1 1 -1 1 -1 1 -1
  1 -1 1 1 -1 1 1 -1 1 -1 1 -1
1 −1 −1 1  1 1 -1 1 1 -1 1 1 -1 -1 -1 1
preamble accordingly, the channel can also be estimated. -1 1 -1 -1 1 -1 -1 1 -1 1 -1 1
1 -1 1 1 -1 1 1 -1 1 -1 1 -1
III. PROPOSED PREAMBLE STRUCTURE
As we have metioned previously, the preamble for MIMO antenna 1 antenna 2
case can be extended from its SISO case. The extension to FIGURE 4. NPS preamble structure for a 2x2 MIMO-FBMC/OQAM
system, with N =8 .
MIMO system is straightforward. The length of the
preamble in the MIMO case is about two times than it in Based on the idea of maximum interference cancellation
the SISO case. Since the length of the preamble in MIMO gain from all the subcarriers in the transmitter antennas, the
system is extended, there exisits additional interference larger interference power of c p ,2 , c p ,4 and the lower
from the middle of the preamble structure.The conventional interference power of additional interference, we consider
extended preamble structures suffer from complex every four subcarrier to be modulated, and propose a novel
interference effect. preamble structure for MIMO-FBMC/OQAM system. We
At first, we provide the extension version of our previous utilize symmetry pattern to cancel interference and jointly
proposed preamble structure in SISO case from Ref. [18], as consider the additional interference in the middle of the
shown in Fig. 4. In antenna 1 and antenna 2, we can calculate preamble structure. The interference of the four positions
the interference weights for the different symbols in frame. with the greatest interference weights is eliminated, so that
For the second and fifth symbols in antenna 1, when at an the optimal interference result can be obtained. At the same
odd subcarrier and the middle pilot is 1, the preamble time, the maximum interference cancellation is also reflected
interference has magnitude 1-j 2( β +δ ) . When the middle
2
in the middle of the preamble structure, and the influence of
pilot is -1, the preamble interference has minimum interference weights is considered. Fig. 5 gives the
proposed preamble structure.
magnitude 1-j 2(δ − β ) . When at an even subcarrier and the
2

VOLUME XX, 2017 9

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2864221, IEEE Access

1 1 1 1 1 1 1 1 1 -1 -1 -1
1 -1 1 1 -1 1 1 -1 1 -1 1 -1
1 1 1 1 1 1 1 1 1 -1 -1 -1
1 -1 1 1 -1 1 1 -1 1 -1 1 -1
1 1 1 1 1 1 1 1 1 -1 -1 -1
1 -1 1 1 -1 1 1 -1 1 -1 1 -1
1 1 1 1 1 1 1 1 1 -1 -1 -1
1 -1 1 1 -1 1 1 -1 1 -1 1 -1

antenna 1 antenna 2
FIGURE 5. Proposed preamble structure for a 2x2 MIMO-FBMC/OQAM system, with N =8 .

TABLE 2. COMPARISONS OF THE ADDITIONAL INTERFERENCE POWER IN ANTENNA 1

The third symbol
Subcarrier 1 Subcarrier 2 Subcarrier 3 Subcarrier 4
Method

1+j (2δ -β ) 1+j 2( β − δ ) -1+j 2(γ + δ +β ) 1+j 2(γ + δ + β )

2 2 2 2
ICM[16]
(1.03308761) (1.01860496) (5.56933376) (5.56933376)

1-j (2γ +β + 2δ ) 1 + j 2β 1 + j 2(δ − β − γ ) 1 + j 2( β + γ )

2 2 2 2
NPS
(3.30985249) (1.40525956) (0.6766940) (3.68107876)

1 + jβ 1 + j (4δ − 2γ ) 1 + j (4δ − 2γ )
2 2 2

Proposed 1
(1.10131489) (5.00480144) (5.00480144)

The fourth symbol

Subcarrier 1 Subcarrier 2 Subcarrier 3 Subcarrier 4
Method

1+j (2δ -β ) 1+j 2( β − δ ) 1+j 2(γ − β -δ ) 1 + j 2( β + δ − γ )

2 2 2 2
ICM[16]
(1.03308767) (1.01860496) (0.981504) (1.018496)

j 2γ +β） 1 + j 2(γ − β ) 1+j 2γ 1 + j 2( β − γ )

2 2 2 2
1+（
NPS
(2.74) (1.13264164) (1.00160064) (1.01860496)

1 + jβ 1 + j (4δ − 2γ ) 1 + j (4δ − 2γ )
2 2 2
Proposed 1
(1.10131489) (0.99999986) (0.99999986)

In antenna 1 and antenna 2, we can calculate the
interference weights for the second and fifth symbols in
frame. The preamble interference at an even subcarrier has
the same magnitude with -1+j 4δ , and the preamble
interference at an odd subcarrier has the same magnitude
with 1-j 4δ . The additional interference from the third and
fourth symbols can also be calculated. In antenna 1, the
preamble interference in the third subcarrier at an even
subcarrier has magnitude 1 + j (2γ + 4δ ) ; the preamble
interference at the first subcarrier has magnitude 1 + j β ; the
preamble interference at the third subcarrier has magnitude 1.
And the preamble interference in the fourth subcarrier at an
even subcarrier has magnitude 1 + j (4δ − 2γ ) ; the preamble
interference at the first subcarrier has magnitude with 1 + j β ;
the preamble interference at the third subcarrier has
magnitude 1. Similarly, the additional interference weights in
antenna 2 can also be calculated. The preamble interference
in the third subcarrier at an even subcarrier has magnitude 1;
the preamble interference at the first subcarrier has
magnitude 1+j (2γ + 2δ +β ) ; the preamble interference at the
third subcarrier has magnitude 1+j (2γ + 4δ ) . And the
preamble interference in the fourth subcarrier at an even
subcarrier has magnitude 1; the preamble interference at the
first subcarrier has magnitude with 1+j (2δ +β -2γ ) ; the
preamble interference at the third subcarrier has
magnitude 1+j (4δ − 2γ ) .
Table 2 and Table 3 show the comparison of the additional
preamble interference power of Ref. [16] ICM, NPS and
proposed method in antenna 1 and antenna 2, respectively.
We can find that the interference from the third and fourth
symbols in the proposed method is partly less than NPS and
ICM methods. Moreover, compared with other two preamble
structures, the proposed structure is less affected by the
fourth symbol interference in the two antennas, especially in
antenna 2. This means that the proposed structure has smaller
additional interference effects.

VOLUME XX, 2017 9

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2864221, IEEE Access

TABLE 3. COMPARISONS OF THE ADDITIONAL INTERFERENCE POWER IN ANTENNA 2

The third symbol

Method Subcarrier 1 Subcarrier 2 Subcarrier 3 Subcarrier 4

1+j (2γ -β ) 1+j 2( β − γ -δ ) -1+j 2(δ +β ) 1 + j 2( β + δ )

2 2 2 2
ICM[16]
(1.46580625) (0.25281264) (2.29231424) (2.29231424)

1 − jβ 1 + j 2( β + γ ) 1 + j 2(δ − β ) 1 + j 2β
2 2 2 2
NPS
(0.89868511) (3.68107876) (0.98139504) (1.40525956)

Proposed
1+j (2γ + 2δ +β ) 1+j (2γ + 4δ )
2 2

1 1
(4.30985249) (5.00480144)

The fourth symbol

Subcarrier 1 Subcarrier 2 Subcarrier 3 Subcarrier 4
Method

-1+j (2γ +β ) 1 + j 2(γ + β − δ ) 1-j 2(δ +β ) 1 + j 2( β + δ )

2 2 2 2
ICM[16]
(2.74) (2.29322384) (0.29231424) (2.29231424)

j β -2δ）
1+ （ -1+j 2 β 1 + j 2β
2 2 2
NPS 1
(0.96691239) (1.40525956) (1.40525956)

1+j (2δ +β -2γ ) 1+j (4δ − 2γ )

2 2
Proposed 1 1
(0.96676671) (0.99999984)
proposed scheme are discussed. In the simulations, the
IV. SIMULATION RESULTS channel estimation performance is also evaluated with
In this section, different preamble-based channel different parameters, including the prototype filter, the
estimators for a MIMO-FBMC/OQAM system are simulated
number of subcarriers, and the number of
and compared. The PAPR, BER and MSE performance of
the preamble schemes (IAM, ICM and NPS) versus the
transmit/receive antennas.
TABLE 4. PROFILES OF THE TWO CHANNELS
Pedestrian A channel Path 1 Path 2 Path 3 Path 4

Delay (ns) 0 110 190 110

Relative power (dB) 0 -9.7 -19.2 -22.8

Extended Pedestrian A Path 1 Path 2 Path 3 Path 4 Path 5 Path 6 Path 7

channel
Delay (ns) 0 30 70 90 110 190 410

Relative power (dB) 0 -1.0 -2.0 -3.0 -8.0 -17.2 -20.8

 H−H ˆ 
2 Table 4 shows the channel profiles of the two channels with
The MSE, E   is plotted with respect to the 10 MHz sampling frequency.
 H2 
 
  A. PAPR COMPARISON
signal to noise ratio (SNR). The simulation channel models For fair comparison, the transmission power is kept equal
are 4-tap pedestrian A (PA) and 7-tap extended pedestrian A for all preamble methods. Fig. 6 shows the comparison
(EPA) channels with low spatial correlations. We take between the conventional preambles and the proposed
modulation as 4QAM. The number of subcarrier is N=256. preamble. The plots show the squared magnitude of the
The square root raised cosine (SRRC) filter is adopted in modulated preambles at the second antenna. Table 5 shows
FBMC/OQAM system, the roll off factor of the filter equals the values of PAPR for the six preambles.
to one, and the length of filter is 4T0 . ZF equalizer is used. It is obvious that ICM in [17] has the best PAPR
performance in the six preambles, and NPS provides the

VOLUME XX, 2017 9

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2864221, IEEE Access

worst PAPR of 188.41 dB. It should be noted that the following simulations, SRRC filter is adopted in
proposed preamble has approximately the same PAPR value FBMC/OQAM systems.
as Ref. [16] ICM. The PAPR of the proposed preamble can
be acceptable. -1.1 SRRC
TABLE 5 10
IOTA
PAPR COMPARISON FOR THE FIVE PREAMBLE METHODS EGF

ICM[16]
IAM-C E-IAM-C NPS Proposed
ICM[17] 10
-1.3

140.09
PAPR 104.81 230.91 188.41 140.1
52.43

BER
-1.5
10
IAM-C
20
Amplitude

10
-1.7
10
0
0 200 400 600 800 1000 1200
Time
E-IAM-C
-1.9
30 10
Amplitude

0 1 2 3 4 5 6 7 8 9 10
20
SNR(dB)
10
0 FIGURE 7. BER performance of the proposed methods with three
0 200 400 600 800 1000 1200 different prototype filters.
Time
ICM[16] 0
20 10
Amplitude

IAM-C
10 E-IAM-C
ICM[16]
0 ICM[17]
0 200 400 600 800 1000 1200
Time NPS
(a) Proposed

ICM[17]
BER

-1
20 10
Amplitude

10

0
0 200 400 600 800 1000 1200
Time
NPS
20
Amplitude

10
-2
10
0 0 1 2 3 4 5 6 7 8 9 10
0 200 400 600 800 1000 1200
SNR(dB)
Time
Proposed FIGURE 8. BER performance of the preamble methods in PA channel.
20
Amplitude

Then, Fig. 8 and Fig. 9 show the BER and MSE

10
performance of the proposed preamble and conventional
0
0 200 400 600 800 1000 1200 preambles under PA channel for 2x2 MIMO-FBMC/OQAM
Time systems, respectively. It can be seen that, in both the BER
and MSE cases, the proposed preamble method outperforms
(b)
FIGURE 6. Squared magnitude of the modulated preambles at the the other preamble schemes over the entire SNR range
second antenna, and (a) two IAMs and ICM[16], (b) ICM[17], NPS and considered. The performance of the six preamble methods
proposed preambles.
degrades in higher SNRs. Ref. [16] ICM can provide better
performance than the Ref. [17] ICM. This is consistent with
B. CHANNEL ESTIMATION PERFORMANCE
the simulation result in SISO systems [18]. We can also find
At first, we evaluate the proposed structure with different
that NPS can only provide better BER and MSE performance
prototype filters, including square root raised cosine (SRRC)
than the Ref. [16] ICM in the SNR range of 0-7. At the
filter, extend Gaussian function (EGF) filter and isotropic
higher SNRs, there is an error floor for NPS. The NPS
orthogonal transform algorithm (IOTA) filter. Fig. 7 shows
method in MIMO systems cannot provide the same superior
the BER performance of the proposed method with three
performance as it in SISO systems. This is because of the fact
different prototype filters for 2x2 MIMO-FBMC/OQAM
that only part of the additional interference power in NPS is
systems. As shown in Fig. 7, the FBMC/OQAM system by
less than Ref. [16] ICM method. Table 2 and Table 3 show
utilizing EGF filter has the worst BER performance, and the
the detailed additional interference power of the three
system by utilizing SRRC and IOTA filters can obtain
methods. For NPS, the additional interference from the
similar BER performance. SRRC filter can provide better
fourth symbol in antenna 1 is more than Ref. [16] ICM
BER performance than IOTA in higher SNRs. In the
method.

VOLUME XX, 2017 9

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2864221, IEEE Access

1
methods have better channel estimation performance in less
10
IAM-C multipath PA scenario. NPS still only provide better BER
E-IAM-C and MSE performance than Ref. [16] ICM in the lower SNRs.
0 ICM[16]
10
ICM[17] The proposed method can provide slightly better BER
NPS performance than NPS and Ref. [16] ICM methods.
Proposed
-1 0
10 10
MSE

-2
10
IAM-C
-1 E-IAM-C
10
-3 ICM[16]
10
ICM[17]
NPS

MSE
Proposed
-4
10
0 1 2 3 4 5 6 7 8 9 10
-2
SNR(dB) 10

FIGURE 9. MSE performance of the preamble methods in PA channel.

In Fig. 8, the Ref. [16] ICM, NPS and proposed preamble
method gradually reach an error floor. This phenomenon is
due to domination of errors from inter-symbol interference -3
10
0 1 2 3 4 5 6 7 8 9 10
and interference cancellation residual. It can be seen from Fig. SNR(dB)
9, compared with the conventional five preamble methods, FIGURE 11. MSE performance of the preamble methods in EPA channel.
the channel estimation MSE performance is effectively From Fig. 11, the proposed method can provide a best
improved by the proposed method. This is due to the MSE performance in the six methods. We can find that the
utilization of symmetry pattern to cancel interference and MSE performance gains of the proposed method compared to
consideration of the additional interference effect in the the other methods are evidently more than that shown in Fig.
middle of the preamble structure. Most of interference power 9. However, with the increasing of multipath, the MSE error
from the third and the fourth symbols are reduced with the floor effect is more obvious. The degradation on the
proposed method. Therefore, the proposed method can proposed method is more significant in EPA channel
provide a significant improvement MSE performance. scenario.
-1
0 10
10
IAM-C 8x8,512
E-IAM-C 4x4,512
ICM[16] 2x2,512
ICM[17] 8x8,256
NPS 4x4,256
-2
Proposed 10 2x2,256
MSE
BER

-1
10

-3
10

-2 -4
10 10
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
SNR(dB) SNR(dB)
FIGURE 12. MSE performance of the proposed method with different
FIGURE 10. BER performance of the preamble methods in EPA channel. number of subcarriers and different number of antennas.
Fig. 10 and Fig. 11 depict the BER and MSE performance Fig. 12 shows the MSE performance of the proposed
of the preambles for 2x2 MIMO-FBMC/OQAM systems in channel estimation method with different number of
the Extended PA channel scenario, which has more multipath subcarriers and antennas. PA channel is adopted in the
than PA scenario as shown in Table 4. Results show an simulation. The proposed method exhibits MSE floor at high
improved performance of the proposed method compared to SNR region for the six curves. This is due to domination of
the other five methods, for both BER and MSE errors from inter-symbol interference and interference
measurements. cancellation residual. The curve for
Increasing the number of delay paths to 7, it can be { Nt × N r =2 × 2, N =256} can provide the best MSE
observed from Fig. 10 that the entire BER performance is performance in the six curves. It can be found that increasing
worse than that shown in Fig. 8. Evidently, the six preamble the number of antennas and subcarriers degrades the MSE

VOLUME XX, 2017 9

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2018.2864221, IEEE Access

performance. However, it should be also noted that [13] E, Kofidis, D. Katselis, “Improved interference approximation
method for preamble-based channel estimation in FBMC/OQAM,” in
increasing the number of antennas will make the pilot Proc. IEEE 19th European Signal Process. Conf., Sept. 2011, pp.1603-
sequence longer, which will reduce the system spectrum 1607.
efficiency. [14] E, Kofidis, “Preamble-based Estimation of Highly Frequency
Selective Channels in FBMC/OQAM Systems,” IEEE Trans. Signal
From the simulation results, it can be verified that the Process., vol. 65, no. 7, pp.1855-1868, Apr. 2017.
proposed method provides performance improvement in the [15] S. W. Kang, K. H. Chang, “A novel channel estimation scheme for
two channel scenarios considered. The simulation results are OFDM/OQAM-IOTA system,” ETRI Journal, vol.29, no. 4, pp.430-
consistent with the data analysis in Table 2 and Table 3. 436, Apr. 2007.
[16] T. W. Yoon, S. Tm, S. H. Hwang, H. J. Choi, “Pilot Structure for
When under a fewer delay paths channel, the superiority is High Data Rate in OFDM/OQAM-IOTA System,” in Proc. IEEE 68th
reflected more obviously. Vehicular Technology Conf., Sept. 2008, pp.21-24.
[17] S. Hu, G. Wu, T. Li, Y. Xiao, et al., “Preamble design with ICI
cancellation for channel estimation in OFDM/OQAM system,” IEICE
V. CONCLUSIONS
Transactions on Communications, vol. E93-B, no. 1, pp. 211-214, Jan.
In this paper, we have proposed a novel preamble structure 2010.
for channel estimation in MIMO-FBMC/OQAM systems. [18] H. Wang, W. C. Du, L. W. Xu, “Novel Preamble Design for Channel
The conventional extended preamble schemes (IAM, ICM Estimation in FBMC/OQAM Systems,” KSII Transactions on Internet
and Information Systems, vol. 10, no. 8, pp.3672-3688, Aug. 2016.
and NPS) are reviewed. The proposed method is compared [19] A. I. Perez-Neira, et al., “MIMO Signal Processing in Offset-QAM
with other five preamble schemes under the PA and EPA Based Filter Bank Multicarrier Systems,” IEEE Trans. Signal Process.,
channels. Simulation results demonstrate that the proposed vol. 64, no. 21, pp.5733-5761, Nov. 2016.
[20] E. Kofidis, D. Katselis, “Preamble-based channel estimation in
method can achieve better BER and MSE performance than MIMO-FBMC/OQAM systems,” in Proc. IEEE Int. Conf. Signal
conventional preamble methods. It has been verified that our Image Process., Nov. 2011, pp. 579-584.
proposed preamble method is an efficient channel estimation [21] E. Kofidis, “Preamble-based estimation of highly frequency selective
channels in MIMO-FBMC/OQAM systems,” in Proc. 21th Eur.
scheme for MIMO-FBMC/OQAM systems. Wireless Conf., Budapest, Hungary, 2015, pp. 1-6.
[22] S. Taheri, M. Ghoraishi, P. Xiao, “Overhead reduced preamble-based
REFERENCES channel estimation for MIMO-FBMC systems,” in Proc. Int. Wireless
[1] K. Berbra, M. Barkat, F. Gini, et al., “A fast spectrum sensing for CP- Commun. Mobile Comput. Conf., 2015, pp. 1435-1439.
OFDM cognitive radio based on adaptive thresholding,” Signal [23] S. Taheri, M. Ghoraishi, P. Xiao, et al., “Evaluation of Preamble
Process., vol. 128 , pp. 252–261, Nov. 2016. Based Channel Estimation for MIMO-FBMC systems,” ZTE
[2] B. F. Boroujeny, H. Moradi, “OFDM Inspired Waveforms for 5G,” Communications, vol.14, no. 4, pp.3-10, Oct. 2016.
IEEE Commun. Surveys & Tutorials, vol. 18, no. 4, pp.2474-2492, [24] S. Hu, Z. Liu, et al., “Training Sequence Design for Efficient Channel
Apr. 2017. Estimation in MIMO-FBMC Systems,” IEEE Access, vol. 5, pp.4747-
[3] D. Gregoratti, X. Mestre, “Uplink FBMC/OQAM-Based Multiple 4758, Apr. 2017.
Access Channel: Distortion Analysis Under Strong Frequency
Selectivity,” IEEE Trans. Signal Process., vol. 64, no. 16, pp.4260-
4272, Aug. 2016.
[4] W. Chuang, C. Kim, S. Choi, D. Hong, “Synchronization Sequence
Design for FBMC/OQAM Systems,” IEEE Trans. Wirel. Commun.,
vol. 15, no. 10, pp.7199-7211, Oct. 2016.
[5] J. Nadal, C. N. Abdel, A. Baghdadi, “Low-Complexity Pipelined
Architecture for FBMC/OQAM Transmitter,” IEEE Trans. Circuits
Syst. II Express Briefs, vol. 63, no. 1, pp.19-23, Jan. 2016.
[6] H. Wang, X. P. Wang, L. W. Xu, W. C. Du, “Hybrid PAPR Reduction
Scheme for FBMC/OQAM Systems Based on Multi Data Block PTS
and TR Methods,” IEEE Access, vol. 4, pp.4761-4768, Sept. 2016.
[7] D. Chen, D. Qu, T. Jiang, Y. He, “Prototype Filter Optimization to
Minimize Stopband Energy With NPR Constraint for Filter Bank
Multicarrier Modulation Systems,” IEEE Trans. Signal Process., vol.
61, no. 1, pp.159-169, Jan. 2013.
[8] D. Chen, X. G. Xia, T. Jiang, X. Gao, “Properties and Power Spectral
Densities of CP Based OQAM-OFDM Systems,” IEEE Trans. Signal
Process., vol. 63, no. 14, pp.3561-3575, July 2015.
[9] P. Siohan, C. Siclet, N. Lacaille, “Analysis and design of
OFDM/OQAM systems based on filterbank theory,” IEEE Trans.
Signal Process., vol. 50, no. 5, pp.1170-1183, May 2002.
[10] H. S. Sourck, Y. Wu, J. W. Bergmans, S. Sadri, and B. F. Boroujeny,
“Complexity and performance comparison of filter bank multicarrier
and OFDM in uplink of multicarrier multiple access networks,” IEEE
Trans. Signal Process., vol. 59, no. 4, pp.1907-1912, Apr. 2011.
[11] C. Lele, J. Javaudin, R. Legouable, et al., “Channel estimation
methods for preamble-based OFDM/OQAM modulations,” European
Transactions on Emerging Telecommunications, vol. 19, no. 7,
pp.741-March 2008.
[12] J. F. Du, S. Signell, “Novel Preamble-Based Channel Estimation for
OFDM/OQAM Systems,” in Proc. IEEE Int. Conf. Commun., Jun.
2009, pp.1-6.

VOLUME XX, 2017 9

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
View publication stats http://www.ieee.org/publications_standards/publications/rights/index.html for more information.