Received May 8, 2019, accepted May 11, 2019, date of publication May 15, 2019, date of current version

June 4, 2019.
Digital Object Identifier 10.1109/ACCESS.2019.2916995

A Multi-User Detection Scheme Based on Polar

Code Construction in Downlink Underwater
Acoustic OFDM Communication System
GANG QIAO, SIYU XING , AND FENG ZHOU
Acoustic Science and Technology Laboratory, Harbin Engineering University, Harbin 150001, China
Key Laboratory of Marine Information Acquisition and Security (Harbin Engineering University), Ministry of Industry and Information Technology,
Harbin 150001, China
College of Underwater Acoustic Engineering, Harbin Engineering University, Harbin 150001, China
Corresponding author: Feng Zhou (zhoufeng@hrbeu.edu.cn)
This work was supported in part by the National Natural Science Foundation of China under Grant 61431004, Grant G051917003,
Grant 61571151, Grant 61501134, Grant 11774074, Grant 61601136, and Grant 61771152, and in part by the National Key Research and
Development Program of China under Grant 2017YFC0305702 and Grant 2018YFC0308500.

ABSTRACT In this paper, we propose a novel polar code construction scheme, suitable for the under-
water acoustic (UWA) multi-user communication system using orthogonal frequency division multiplex-
ing (OFDM). We use the polar codes because the code rate of a polar code can be arbitrarily designed.
Moreover, the polar codes can also be used to create a nested code structure, divided into several subsets
(codebook). We exploit this feature of the polar codes and place the information bits in the codebook to avoid
superimposition of all the users. This helps to eliminate the influence of the power allocation coefficient and
make full use of the power of the transducer at the transmitted node. Furthermore, the proposed scheme
maximizes the utilization efficiency of the OFDM subcarrier resource for each user by avoiding null pilot
tones that are usually required for channel equalization in the traditional schemes. The proposed scheme also
does not depend on the channel state information (CSI) as prior information as each user can decode their
own data independently by using their own codebook without successive interference cancellation (SIC).
The simulation and tank experiment results demonstrate that the proposed scheme can significantly guarantee
the reliability of all the users, and reduces the complexity of the receiver, thus making it most suitable for the
downlink UWA multi-user communication systems.

INDEX TERMS Underwater acoustic communication (UAC), orthogonal frequency division multiplex-
ing (OFDM), multi-user detection, polar code construction.

I. INTRODUCTION multipath, strong Doppler, strong noise and other complex

The shallow underwater acoustic (UWA) channel is com-
monly regarded as the most difficult wireless communica-
tion channel because of its complex environments, such as
low carrier frequency, strong multi-path interferences and
high noise level. The shallow seas with a depth less than
200 meters are normally near the mainland, which makes
them the most extensive and densely populated areas of
human ocean development, shipping, etc. However, the sound
propagation in the shallow water is susceptible to the com-
plex effects of the sea surface and multiple reflections of
the sea floor and the sea surface, not mention the shipping,
underwater engineering and other types of human industry.
Thus, the shallow seawater acoustic channel exhibits strong
The associate editor coordinating the review of this manuscript and
approving it for publication was Qilian Liang.
2169-3536
2019 IEEE. Translations and content mining are permitted for academic research only.
VOLUME 7, 2019 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.
random transmission characteristics, making the reliable and
high-rate underwater communication much more difficult
than the wireless communication. The reliability of shallow
seawater acoustic communication has always been a great
challenge worldwide. Hence, the reliable and high-rate under-
water communication system remains a worldwide challenge
because of the unique nature of the underwater environment.
Since, orthogonal frequency division multiplexing (OFDM)
is an attractive technique for high-rate data transmission
due to the minimizing effect over frequency-selective fading
channels [1]. Therefore, OFDM has been extensively used
in high-rate underwater acoustic communication (UAC)
systems [2]–[4].
In recent decades, underwater acoustic sensor networks
have been considered as a promising approach for explor-
ing and monitoring the ocean. Hence, designing reliable
65973

and efficient data communication protocol is necessary for
reliable data transmission in underwater applications. Some
high-energy efficient schemes with low memory requirement
were proposed to repalace the traditional ones, such as the
link quality-aware queue-based spectral clustering routing
protocol and Quality-of-Service aware evolutionary routing
protocol [5]–[6].
With the developing of the UAC network, multi-user
communication technique draws much more interest than
the single-user, and numerous methods of multi-user
detection have been proposed [7]–[9]. These methods
can be sorted into two typically types: the orthogonal
approaches and the non-orthogonal approaches. The orthog-
onal approaches can be achieved by time-division multi-
ple access (TDMA) [10], frequency-division multiple access
(FDMA) [11], and orthogonal frequency division multi-
ple access (OFDMA) [12]–[14], while the non-orthogonal
approach mainly consists of code division multiple access
(CDMA) [15]–[17], interleave division multiple access
(IDMA) [18]–[20], and a recent popular scheme called the
superposition coding [21]–[23].
In [24], CDMA and OFDM are combined to realize multi-
user UWA communication by adopting the time reversal mir-
ror technique. The proposed technique takes the advantages
of both CDMA and OFDM, which effectively suppressed
multi-path interference and reduced the error rate. However,
due to the usage of the spreading code, there exists a trans-
mitted rate loss of the UWA OFDM communication sys-
tem. A superposition coding suitable for the downlink UWA
OFDM system is proposed in [25]. By applying superposition
coding, the source node transmits the superimposed signal of
the two users. The two users utilize different constellation
mapping mode, and then, the signal of both users is added
with a power allocation coefficient, which mainly determined
the bit error rate (BER) performance of the two user’s system.
At the receiver, User A decodes its own data in the presence of
interference from User B’s signal, while User B with a higher
signal-to-noise ratio (SNR) performs successive interference
cancellation (SIC). User B firstly decodes the data of User A,
then subtracts the codeword of User A from the received
signal to better decode its own data. The scheme performs
well when the power difference between the two users is large
enough and require the channel state information (CSI). Thus,
the complexity of the receiver is high which limits its use for
many applications.
The polar codes were invented by Erdal Arikan [26], based
on a novel concept called channel polarization. As soon as
polar codes were invented, they became an active research
subject due to the fact that polar codes showed promising
results with increasing capacity. Polar codes have now
been extended to a number of applications and generaliza-
tions [27]–[31], and it is proved that the polar code has
better performance than other existing channel coding meth-
ods. The advantage of polar code is the explicit construction
and very low complexity of encoding and decoding. Fur-
thermore, the code rate of a polar code can be arbitrarily
65974
G. Qiao et al.: Multi-User Detection Scheme Based on Polar Code Construction
designed, which makes it more flexible adapting to various
application scenarios, such as security coding on the physical
layer [32]–[34].
In this paper, we propose a novel polar code construction
scheme suitable for multi-user communication and especially
the UWA OFDM communication system. Polar codes offer
the flexibility to adjust the code rate on the fly and this
advantage is exploited in the proposed scheme. Moreover,
the polar codes can also be used to create a nested code
structure, divided into several subsets where each subset carry
the information bits and is called the codebook, for different
users. The codebook is usually known to all the users and
the transmitting node. The transmitted data of each user is
already placed at different information bits in every codeword
after conducting polar encoding, thus, there is no need to
superimpose the signal of all the users, which can eliminate
the influence of the power allocation coefficient. Meanwhile,
the novel polar encoding scheme is also a kind of channel
coding scheme, the following process of the UWA OFDM
communication system remains the same as single user ones,
resulting in no need for CSI and reserved null pilot subcarriers
for each user. Different users can decode their own data inde-
pendently by using their own codebook without SIC, which
reduces the complexity of the receiver. The proposed scheme
can significantly enhance the quality and the frequency effi-
ciency of the OFDM communication system, thus making it
most suitable for UWA multi-user communication.
The rest of the paper is organized as follows. In Section II,
we introduce the principle of the original polar code,
including the channel polarization, polar code construction,
the encoder and decoder of polar code. In Section III,
we present a novel polar construction and encoder scheme
which is designed for multi-user downlink UWA OFDM
communication system, and also introduce the construction
of the system model. The simulation and experimental results
were given in Section IV and conclusions are contained in
Section V.
II. PRINCIPLE OF THE ORIGINAL POLAR CODE
A. CHANNEL POLARIZATION AND ORIGINAL
POLAR CODE CONSTRUCTION
The key idea of the channel polarization can be expressed
as a kind of channel equivalent transformation, and mainly
consists of two parts: channel combining phase and channel
splitting phase.
Considering a binary input discrete memoryless symmetric
(BI-DMS) channel W , which can be described as
W : X → Y, (1)
where X denotes the input data symbol and Y represents the
output, the transition probability W (y|x), x ∈ X and y ∈ Y .
Normally, X = {0, 1}.
The channel polarization can manufacture N polarized
channel out of N identical independent copies of BI-DMS
channels. The first step is to transform the same indepen-
dent channels into two types of synthesized channels with
VOLUME 7, 2019
G. Qiao et al.: Multi-User Detection Scheme Based on Polar Code Construction
slightly varying reliabilities: the good bit-channels and the
bad bit-channels. Then the significant difference, which is
good ones get better and the bad get worse, can be shown
when the polarization transformation is applied on the result-
ing channels recursively. When N → ∞, the channel can be
polarized into the channel either extremely noisy or noiseless
without capacity loss. The symmetric capacity of bit-channels
after the channel polarization when N = 2048 is given
in Fig.1.
FIGURE 1. The symmetric capacity of bit channels after channel
polarization when N=2048.
After the channel polarization, a rate of transmission close
to capacity can be easily achieved by transmitting over only
the good bit-channels, while distributing frozen bits to the
extremely noisy ones. A polar code may be specified by
(N , K , F), where N represents the code length, which needs
to be at any finite length of 2n . K is defined as the number
of good bit-channels, in other words, the length of the infor-
mation bits in a codeword. Thus, the code rate (CR) can be
calculated as R = K /N . F is a set of N − K integer indices
known as the frozen bit locations from {0, 1, . . . , N − 1}.
Polar code construction is a selection of the K bit-channels
out of N , which defines a polar code. In order to conduct this
selection, a ranking algorithm for the bit-channels is needed.
Normally, the measurement information of the polar code
construction is the Bhattacharyya parameter, which is defined
as Equation (2)
1 Xp
B(W ) = W (y|0)W (y|1). (2)
y∈Y
In order to avoid underflow and overflow, the log-domain
calculations of the Bhattacharyya parameter are needed.
Then, a polar code construction can be achieved by the
following process and the framework of construction of the
original polar code is shown in Fig. 2.
1) Calculate the logarithm of the bit-channel
Bhattacharyya parameter Z (WNi ) = log10 [B(WNi )],
2) Sort Z (WNi ) with the ascending order, one marked array
flag is used to record the index of the bit-channel
while the other marked array value is set to record
the corresponding Z (WNi ).
3) The top K values in the marked array flagare the
indexes of the good bit-channels, which carries the
VOLUME 7, 2019
FIGURE 2. The framework of construction of the original polar code.
information bits, and the rest N − K values are
the indexes of the frozen bits. σ is the K th value in the
marked array value, representing the threshold of the
logarithm of the bit-channel Bhattacharyya parameter
with the code rate R = K /N .
4) Finally, the construction of the codebook C of the polar
code with the code rate R = K /N is completed.
B. THE ENCODER OF THE ORIGINAL POLAR CODE
The encoding operation of an (N , K , F) polar code for a
vector information bit u = (u1 , u2 , . . . , uK ), whose length
1
is K , can be obtained as follows. Define n = log2 (N ) and
⊗n
F = F ⊗ · · · ⊗ F (n copies of F) be the n-fold Kronecker
 
1 1 1
product of Arikan’s standard polarizing kernel F = .
01
Then a codeword of Polar code can be generated as
x = G · u = (F⊗n )F C · u, (3)
where F C is the complementary set of the frozen bits loca-
tions F, which is usually known as the information bits
1
locations. G =(F⊗n )F C is the generator matrix of polar code.
Meanwhile, there is an alternative form of this polar code as
x = F⊗n d, (4)
where d ∈ {0, 1}N is such that dF = 0 and dF C = u.
The information bits are assigned to the polarized channels
with the lower error probabilities, while the frozen bits are
assigned to the remaining less reliable channels. The frozen
bits generally take a fixed value of zeros or a fixed vector
which is known at both the encoder and the decoder.
C. THE DECODER OF THE ORIGINAL POLAR CODE
In this paper, the Successive Cancellation (SC) decoding
technique is selected as the decoder for polar code. Due to
the regular structure of polar codes, the SC algorithm can be
regarded as a soft/hard information-sending algorithm over
the trellis of polar code. The trellis has n stages and N levels.
The hard messages passed in the trellis are the estimation
bits corresponding to the variable nodes si,j , where i and j are
the stage and level index in the trellis, respectively, 1 ≤ i ≤
n + 1, 1 ≤ j ≤ N . The hard messages update rule can be
expressed as Equation (5)
 
j−1

i,j+2i−1 , mod 2 = 0
s ⊗ s
i,j
Si,j = 2i−1 (5)
si,j , otherwise.

where ⊗ is the modulo-2 operation and b·c indicates the floor
function.
The soft messages corresponding to these bits are the log-
likelihood ratios (LLR) values denoted as Li,j = L(si,j ).
For example, the trellis bit values on the first stage are the
65975
 G. Qiao et al.: Multi-User Detection Scheme Based on Polar Code Construction

estimations of the source block s1,j = ûj , the correspond-

ing soft messages are L1,j = L(ûi,j ). The received signal
after bit-reversal permutation is yj , its corresponding LLR is
denoted by
FIGURE 3. The framework of the proposed polar code construction
P(yj |1) designed for the multi-user detection scheme in the UWA OFDM
Ln+1,j = log . (6) communication system.
P(yj |0)
The soft messages update rule can be expressed as
Equation (7)
Li,j

Li+1,j Li+1,j+2i−1
2 tanh−1 [tanh( ) · tanh( )],




 2  2
= j−1 (7)
i−1
mod 2 = 0


 2
i,j−2i−1 )(Li+1,j−2i−1 ) + Li+1,j ,

(1 − 2s otherwise
where i = 1, 2, . . . , n + 1, j = 1, 2, . . . N , tanh(·) denotes the
hyperbolic tangent function, and the other notations are the
same with Equation (6).
FIGURE 4. An illustration of polar coding for the multi-user detection
The bit decision rule of the SC decoder is described as scheme. The message bits of different users located at the different good
Equation (8) bit-channels index for each user, and the frozen bits located at the noisy
 bit-channels.

0, if i ∈ F

ûi = 0, if i ∈ F C & L1,j ≥ 0 .
 each user, respectively. The CR for each user is Ruser =
1, otherwise

III. PRINCIPLE OF THE PROPOSED MULTI-USER
DETECTION SCHEME BASED ON POLAR CODE
As polar codes can be employed to create a nested code
structure, subsets of the set of good bits-channels for the
original polar code are defined to act as the information bits
for different users. In other words, all the users share the
good bit-channels of the original polar code with a certain
code rate R. Normally, the set of good bit-channels of the
original polar code and all the subsets are known to the
sending node and all the users. The subsets are defined as
the codebook for each user, then the CR of each user together
makes the R. In the following section of this paper, the number
of multi-user is set to two, and the CR for each user is equal,
Ruser = R/2. Since R must be less than 1, Ruser can only
approach 1/2 infinitely.
A. POLAR CODE CONSTRUCTION AND ENCODER FOR
THE PROPOSED MULTI-USER DETECTION SCHEME
Assuming that the User A’s channel is better than User B’s,
the polar code construction for each user is achieved by the
following process.
Firstly, calculating the logarithm of the bit-channel
Bhattacharyya parameter Z (WNi ), sorting Z (WNi ) with the
ascending order, and recording marked array flag and value
is identical to the original polar code construction.
Secondly, top KuserA = K /2 values in the marked array
flag are selected as the indexes of the good bit-channels for
User A, and the next top KuserB = K /2 values are that
for User B. Then, the rest N − K values are the indexes
of the frozen bits. σ1 and σ2 represents the thresholds of
(8) the logarithm of the bit-channel Bhattacharyya parameter for
R/2 = K /2N .
Finally, the construction of the codebook of the polar code
for each user is completed. The framework of the proposed
polar code construction designed for the multi-user detection
scheme in the downlink UWA OFDM communication system
is shown in Fig. 3.
Then the codeword of the proposed polar code can be
generated as
x = (F⊗n )FuserA · uuserA + (F⊗n )FuserB · uuserB , (9)
where FuserA and Fuser2 denote the information bits of the
two users, that satisfies FuserA ∪ FuserB = F C , the transmitted
data for each user is uuserA = (u1 , u2 , . . . , uK /2 ) and uuserB =
(u01 , u02 , . . . , u0K /2 ).
The alternative form of this polar code can be given as
x = F⊗n d, (10)
where d can be composed of dF = 0, dFuserA = uuserA and
dFuserB = uuserB .
The illustration of polar coding for the multi-user detection
scheme is shown in Fig. 4.
B. SYSTEM MODEL OF THE UWA OFDM
COMMUNICATION ADOPTED THE
MULTI-USER DETECTION SCHEME
BASED ON POLAR COEDS
After the proposed polar code encoding, the transmitted data
of each user is already placed at different information bits in
every codeword, thus, there is no need to superimpose the two
users’ signal. As the novel polar encoding scheme is also a
scheme of channel coding, the codewords can be processed

65976 VOLUME 7, 2019

G. Qiao et al.: Multi-User Detection Scheme Based on Polar Code Construction
FIGURE 5. The framework of the UWA OFDM system which adopted the proposed multi-user detection scheme based on a novel polar code
construction technique.
as the normal UWA OFDM communication system, such
as constellation mapping and allocated to the subcarriers.
The polar codewords after constellation mapping are XD ,
the OFDM symbol with the comb pilot XP in the frequency
domain can be represented as
X (k) = aXD (k) + bXP (k),
where
a = 1, b = 0 / Pom
k∈
a = 0, b = 1 k ∈ Pom .
Po is the position of the comb pilot tones, and all users
can share the same Po and XP . There is no need to reserve
null subcarriers for each user to obtain the sampling value
of the underwater acoustic channel frequency response ĤP ,
each user can maximize the utilization efficiency of OFDM
subcarrier resource.
The communication rate Scom can be expressed as follow
Scom =
NOFDM · (1P − Nuser ) · NBits_subcarrier · CR · Fs
, ûi = 0, if i ∈ Fuseri & L1,j ≥ 0 .
1P · NOFDM _with_CP · κ
(13)
where NOFDM and NOFDM _with_CP represent the number of
the total subcarriers in an OFDM symbol and that in an
OFDM symbol with the cyclic prefix, NBits_subcarrier is the
number of the bits on each subcarrier, and Nuser denotes the
number of different users. 1P is defined as the pilot spacing,
CR has been pre-defined as the code rate, Fs is known as the
sampling frequency, and κ is the up-sampling factor.
As there is no need for the proposed scheme to reserve
null subcarriers for each user, the Nuser for the proposed
scheme can be fixed to 1. Thus, the communication rate
can be significantly increased with the increasing number of
different users.
Subsequently, IFFT is taken to generate the OFDM signals
x in the time domain. The transmitted signal x experiences
different channels while reaching each user’s receiver. The
framework of the UWA OFDM system which adopted the
proposed multi-user detection scheme based on a novel polar
code construction technique is shown in Fig. 5.
VOLUME 7, 2019
Take one OFDM symbol as an example, after taking
the fast Fourier transform (FFT) operation at the receiver,
the received signal in the frequency domain can be given as
Yi (k) = X (k)Hi (k) + Wi (k), (14)
ith
where Hi (k) denotes the frequency response of the user’s
(11) underwater acoustic channel and Wi (k) represents the white
Gaussian noise in the frequency domain. As all users can
share the same Po and XP , the estimation of Hi (k) also
share the same way. Denote Ĥi (k) as the estimation of Hi (k),
(12) YP is the value of the pilot tones, the sampling value of
the underwater acoustic channel frequency response ĤiP
can be achieved by ĤiP = YP /XP and Ĥi (k) should be
obtained by interpolation scheme or other channel estimation
scheme, such as compressed sensing. After the constellation
de-mapping, the received data need to be sent to the polar
decoder, and the bit decision rule of the SC decoder for the
ith user can be described as Equation (15)

0, if i ∈ F

(15)

1, otherwise

Considering each user has its own codebook and there is no
cross among these codebooks, each user can obtain their own
output data by conducting SC decoding scheme according to
their codebooks.
IV. NUMERICAL SIMULATION AND
TANK EXPERIMENT RESULTS
A. SIMULATION PARAMETERS AND RESULTS
Simulations are performed to evaluate the performance of
this novel multi-user detection scheme. The main parameters
of the polar coding UWA OFDM communication system
are given in Table 1. Meanwhile, a sparse shallow water
acoustic channel, generated by a channel simulation soft-
ware is adopted in the simulations. It is assumed that the
distance between the transducer and the hydrophone is about
5700 meters, and the transducer is located at 8 meters under
the water surface. The hydrophones of the two users are
located at 8 meters and 15 meters under the water surface,
respectively. The velocity gradient distribution is set as a
65977

TABLE 1. Simulation parameters of the underwater acoustic OFDM
system.
FIGURE 6. The CIR of the simulation channel for both the two users.
surface channel, and the average depth of the sea is set to
55 meters. The impulse responses of the simulation channel
for both users are illustrated in Fig. 6. It can be obviously
seen that the direct sound intensity is higher than any other
multipath.
The original polar coding scheme given in [21] is chosen as
a control scheme. As discussed in Section III, the code rate of
each user is Ruser = R/2. The code rate of the original scheme
is set to Roriginal = Ruser = R/2 and Roriginal = 2∗ Ruser = R,
to compare the system BER performance of the polar code
designed for multi-user detection scheme and the original
one. Comparisons of the overall BER performance between
the proposed multi-user detection scheme and original polar
code with different code rate is given in Fig. 7. We set 10−7
our reference as BER is negligible and can be counted as
no error. In Fig. 7 (a), the code rates of the two users are
7/16 and 1/4, and the code rate of the original polar code
is 7/8, 7/16, 1/2, and 1/4, while the code rates in Fig. 7 (b)
are 3/8 and 5/16. The dash curves represent the BER of the
proposed scheme and control scheme with the lower code
rate, whilst the solid ones are the higher ones.
It can be seen that the BER performance of User A is
about 1 dB better than that of User B, despite the code
65978
G. Qiao et al.: Multi-User Detection Scheme Based on Polar Code Construction
FIGURE 7. Comparisons of the overall BER performance between the
proposed multi-user detection scheme and original polar code with
different code rate. (a): Code Rate are 7/16 and 1/4 for the two users and
corresponding Code Rate for the Original Polar code and control scheme.
(b): Code Rate are 3/8 and 5/16 for the two users and corresponding
Code Rate for the Original Polar code and control scheme.
rate difference. It is because that the good bit-channels is
prior assigned to User A when constructing the proposed
polar code. The BER performance of both users with the
code rate Ruser = R/2 are better than the original polar
code with code rate Roriginal = 2∗ Ruser = R, mainly due
to the length of the information bits in one polar codeword of
the two users is shorter than that of the original polar code.
Therefore, the error correction capability of the proposed
polar coding scheme for the multi-user system becomes better
when compared with the original one. With the decreasing of
the code rate, the difference between them decreased from
2 dB to 1 dB.
Although the codebook for each user is independent,
the LLR value may be affected by the good bit-channels of
the other user when updating the LLR in the SC decoder.
It is indicated that the BER performance of each user is worse
than the original polar code with the same code rate. However,
the difference decreases with the decrease in code rate. The
difference in code rate 7/16 is about 6 dB, while it decreases
to 2 dB when the code rate is 1/4. When the code rate of the
multi-user system is 7/16, which is close to 1/2, the BER
of the multi-user finally become zeros at the condition of
Eb/N 0 = 17dB, and both of the two users can guaran-
tee the reliable communication. The code rate determines
both the efficiency of data transmission and the system BER
VOLUME 7, 2019
G. Qiao et al.: Multi-User Detection Scheme Based on Polar Code Construction

FIGURE 8. The figures restored by the received data after decoding in these simulations with different CRs and
Eb/N0 conditions. (a): CR is 1/4 for the two users and corresponding CR for the Original Polar code and control
scheme under the condition of Eb/N0=9dB. (b): CR is 5/16 for the two users and corresponding CR for the
Original Polar code and control scheme under the condition of Eb/N0=11dB. (c): CR is 3/8 for the two users and
corresponding CR for the Original Polar code and control scheme under the condition of Eb/N0=13dB. (d): CR is
7/16 for the two users and corresponding CR for the Original Polar code and control scheme under the condition
of Eb/N0=16dB.

performance, thus, it should be selected carefully to balance

the reliability and effectiveness in practical applications.
To illustrate the performance of the proposed scheme more
intuitive, a Grayscale figure is chosen as the transmitted data.
The figures reconstructed at the receiver after polar decoding
are shown in Figure 8. The sub-figure (a), (b), (c) and (d) rep-
resents the transmitted results of different CRs at different
Eb/N0 conditions respectively. The value of Eb/N0 for each
CR is selected as the same value when BER of User A is close
to being counted as no error. FIGURE 9. The schematic diagram of the location of the transducers in
The system BER values in these figures are the results this experimental site.
which are obtained only in one simulation respectively, while
the curves in Fig. 7 are the average value of many circulations. seen in Fig. 8 (c) and (d). Both the two users can guarantee
Moreover, the total bits of the transmitted data in the simula- reliable communication at certain Eb/N0 conditions.
tions are at different magnitude. Thus, the BER data may not
be the same value we got in Fig. 7. From Fig. 8 (a) and (b), B. TANK EXPERIMENT RESULTS
it can be found that there is no error in both the original To evaluate the feasibility of the proposed multi-user detec-
polar code and the User A with code rate 1/2 and 5/16. The tion scheme for UWA OFDM communication system based
BER of the original polar code with code rate 1/2 and 5/8 are on a novel polar coding scheme, a field experiment was con-
0.0011 and 0.0068, respectively, while the BER of User B is ducted at the channel tank operated by Harbin Engineering
at the magnitude of 10−4 . With the increase of CR, the BER University, China and the schematic diagram of the exper-
of User A becomes at the magnitude of 10−5 , which can be imental site is shown in Fig.9. The length of the tank is

VOLUME 7, 2019 65979


about 25 meters and the average depth of the tank is about
5 meters. During this experiment, the two hydrophones were
located on the same vertical line, and they were 1.5 meters
and 2.5 meters under the tank surface. The distance between
the two receivers is larger than 10 times of the wavelength,
which makes the two users experience the independent chan-
nels. The horizontal distance between the transducer and the
receiver was 8.35 meters and the depth of the transducer was
2 meters under the tank surface. The parameters of the UWA
OFDM system remains the same as it was in the simulations,
and the code rate is set as 7/16 for both the two users, while
that for the original polar code are 7/16 and 7/8.
FIGURE 10. The real Channel Impulse Response of the tank channels.
Before any transmission, a linear frequency modu-
lated (LFM) signal is placed as the probe signal for the
purpose of symbol synchronization. Meanwhile, the estima-
tion of the CIR for each user can be obtained by matched
filtering the received LFM data with the transmitted LFM
signal. The real tank CIR of the two users are illustrated in
Fig.10. It is noticed that the CIR has seven paths, the one
with the maximum power is the direct sound, and the next
three maxima are made up by the reflexed sound from the
surface, reflexed sound from the bottom, reflexed sound from
both the surface and the bottom. The power of these second
reflexed sounds is much lower than the four paths mentioned
above, which appears during the time period of 0.03∼0.04s.
The CIR of User B got the direct sound and one of the reflexed
sound almost at the same time, that may affect the system
BER performance.
In the tank experiment, the original data comes from a
grayscale image, and the total bits of the image is at the
magnitude of 106 . The figures restored by the received data
after decoding are shown in Fig. 11. The received SNR is
about 9.79 dB. It can be found that there is no error in the
original polar code with code rate 7/16, while the BER of
the original polar code with code rate 7/8 is 0.0015. This is
consistent with the relationship between the polar code rate
and the system BER performance, the larger the code rate,
the worse the performance. The BER of the two users are at
the magnitude of 10−5 and 10−4 respectively, both of them
can guarantee reliable communication.
65980
G. Qiao et al.: Multi-User Detection Scheme Based on Polar Code Construction
FIGURE 11. The figures restored by the received data after decoding in
this field experiment.
V. CONCLUSION
This paper proposed a novel polar code construction scheme
which is designed suitable for the multi-user communication
and embedded it in the downlink UWA OFDM communica-
tion system. Due to the character that the code rate of the
polar code can be designed arbitrarily and the polar codes
can be used to create a nested code structure, a novel polar
code construction scheme suitable for multi-user downlink
UWA OFDM system is proposed. The transmitted data of
each user can be located at different information bits in every
codeword during the polar encoding. Thus, there is no need
to superimpose the signal, which can eliminate the influence
of the power allocation coefficient and make full use of the
power of the transducer at the sending node. Meanwhile,
the CSI as prior information is no longer needed and all users
share the same pilot tones and comb pilot vector, which makes
the utilization efficiency of OFDM subcarrier resource for
each user maximized. At the receiver, each user can simply
obtain its output data independently by decoding according
to the own codebook without SIC, which reduces the com-
plexity of the receiver for each user. Simulation and tank
experiment results demonstrate that the proposed multi-user
detection scheme based on the novel polar code construction
technique can significantly guarantee the reliability of all the
users, thus making it suitable for the downlink UWA OFDM
communication system.
REFERENCES
[1] R. Van Nee and R. Prasad, OFDM for Wireless Multimedia Communica-
tions. Norwood, MA, USA: Artech House, 2000.
[2] F. Zhou, Q. Wang, D. Nie, and G. Qiao, ‘‘DE-Sync: A Doppler-enhanced
time synchronization for mobile underwater sensor networks,’’ Sensors,
vol. 18, no. 6, pp. 1710-1–1710-15, 2018.
[3] G. Qiao, S. Z. Liu, F. Zhou, and Z. Sun, ‘‘Experimental study of long-
range shallow water acoustic communication based on OFDM-modem,’’
Adv. Mater. Res., vols. 546–547, pp. 1308–1313, Jul. 2012.
[4] W. Wang, G. Qiao, Y. Wang, S. Xing, and S. Liu, ‘‘Decision feedback
estimation of multiple input/multiple output orthogonal frequency division
multiplexing channel based on punching technique via UWA shallow sea,’’
Acta Armamentarii, vol. 34, no. 9, pp. 1116–1124, 2013.
VOLUME 7, 2019
G. Qiao et al.: Multi-User Detection Scheme Based on Polar Code Construction
[5] M. Faheem, G. Tuna, and V. C. Gungor, ‘‘LRP: Link quality-aware queue-
based spectral clustering routing protocol for underwater acoustic sensor
networks,’’ Int. J. Commun. Syst., vol. 30, no. 12, pp. e3257-1–e3257-17,
2017.
[6] M. Faheem, G. Tuna, and V. C. Gungor, ‘‘QERP: Quality-of-service (QoS)
aware evolutionary routing protocol for underwater wireless sensor net-
works,’’ IEEE Syst. J., vol. 12, no. 3, pp. 2066–2073, Sep. 2018.
[7] Y.-L. Yin, F. Zhou, G. Qiao, and S.-Z. Liu, ‘‘Orthogonal multicarrier M -ary
cycle shift keying spread spectrum underwater acoustic communication,’’
Acta Phys. Sinica, vol. 62, no. 22, 2013, Art. no. 224302.
[8] J. Yin, G. Yang, D. Huang, L. Jin, and Q. Guo, ‘‘Blind adaptive multi-user
detection for under-ice acoustic communications with mobile interfering
users,’’ J. Acoust. Soc. Amer., vol. 141, no. 1, pp. EL70–EL75, 2017.
[9] L.-X. Guo, X. Han, J.-W. Yin, and X.-S. Yu, ‘‘Underwater acoustic com-
munication by a single-vector sensor: Performance comparison using three
different algorithms,’’ Shock Vib., vol. 2018, Nov. 2018, Art. no. 2510378.
[10] N. Morozs, P. Mitchell, and Y. V. Zakharov, ‘‘TDA-MAC: TDMA without
clock synchronization in underwater acoustic networks,’’ IEEE Access,
vol. 6, pp. 1091–1108, 2018.
[11] X. Cheng, F. Qu, and L. Yang, ‘‘Single carrier FDMA over underwater
acoustic channels,’’ in Proc. CHINACOM Conf., Harbin, China, Aug. 2011,
pp. 1052–1057.
[12] K. Tu, T. M. Duman, M. Stojanovic, and J. G. Proakis, ‘‘OFDMA for
underwater acoustic communications,’’ in Proc. 49th Annu. Allerton Conf.
Commun., Control, Comput., Sep. 2011, pp. 633–638.
[13] Y. Zhang et al., ‘‘Adaptive OFDMA for downlink underwater acoustic
communications,’’ in Proc. MTS/IEEE OCEANS Conf., Sep. 2014, pp. 1–5.
[14] L. Ma, S.-Z. Liu, and G. Qiao, ‘‘Sparse channel estimation and pilot
optimization for underwater acoustic orthogonal frequency division mul-
tiple access uplink communications,’’ Acta Phys. Sinica, vol. 64, no. 15,
pp. 154304-1–154304-10, 2015.
[15] M. Stojanovic and L. Freitag, ‘‘Multichannel detection for wideband
underwater acoustic CDMA communications,’’ IEEE J. Ocean. Eng.,
vol. 31, no. 3, pp. 685–695, Jul. 2006.
[16] T. C. Yang and W.-B. Yang, ‘‘Interference suppression for code-division
multiple-access communications in an underwater acoustic channel,’’
J. Acoust. Soc. Amer., vol. 126, no. 1, pp. 220–228, Jul. 2009.
[17] J. Yin, P. Du, G. Yang, and H. Zhou, ‘‘Space-division multiple access
for CDMA multiuser underwater acoustic communications,’’ J. Syst. Eng.
Electron., vol. 26, no. 6, pp. 1184–1190, 2015.
[18] J. Dang, F. Qu, Z. Zhang, and L. Yang, ‘‘Experimental results on
OFDM-IDMA communications with carrier frequency offsets,’’ in Proc.
MTS/IEEE OCEANS Conf., May 2012, pp. 1–5.
[19] S. Aliesawi, C. C. Tsimenidis, B. S. Sharif, and M. Johnston, ‘‘Contin-
uous pilot based adaptive estimation for IDMA systems on underwater
acoustic channels,’’ in Proc. IEEE Int. Conf. Acoust., Speech, Signal
Process. (ICASSP), May 2011, pp. 3476–3479.
[20] J. Dang, F. Qu, Z. Zhang, and L. Yang, ‘‘OFDM-IDMA communications
over underwater acoustic channels,’’ in Proc. Mil. Commun. Conf., 2011,
pp. 418–423.
[21] A. Zafar, M. Shaqfeh, M.-S. Alouini, and H. Alnuweiri, ‘‘On multiple users
scheduling using superposition coding over Rayleigh fading channels,’’
IEEE Commun. Lett., vol. 17, no. 4, pp. 733–736, Apr. 2013.
[22] Z. Ding, Z. Yang, P. Fan, and H. V. Poor, ‘‘On the performance of
nonorthogonal multiple access in 5G systems with randomly deployed
users,’’ IEEE Signal Process. Lett., vol. 21, no. 12, pp. 1501–1505,
Dec. 2014.
[23] S. Vanka, S. Srinivasa, Z. Gong, P. Vizi, K. Stamatiou, and M. Haenggi,
‘‘Superposition coding strategies: Design and experimental evaluation,’’
IEEE Trans. Wireless Commun., vol. 11, no. 7, pp. 2628–2639, Jul. 2012.
[24] Y. Wang, J. Yin, Z. Pan, and P. Du, ‘‘Application of OFDM-CDMA
in multi-user underwater acoustic communication based on time rever-
sal mirror,’’ in Proc. 3rd Int. Conf. Pioneering Comput. Scientists, Eng.
Educators (ICPCSEE), 2017, pp. 325–334.
[25] L. Ma, S. Zhou, G. Qiao, S. Liu, and F. Zhou, ‘‘Superposition coding
for downlink underwater acoustic OFDM,’’ IEEE J. Ocean. Eng., vol. 42,
no. 1, pp. 175–187, Jan. 2017.
[26] E. Arikan, ‘‘Channel polarization: A method for constructing capacity-
achieving codes for symmetric binary-input memoryless channels,’’ IEEE
Trans. Inf. Theory, vol. 55, no. 7, pp. 3051–3073, Jul. 2009.
[27] R. Mori and T. Tanaka, ‘‘Source and channel polarization over finite fields
and Reed–Solomon matrices,’’ IEEE Trans. Inf. Theory, vol. 60, no. 5,
pp. 2720–2736, May 2014.
VOLUME 7, 2019
[28] P. Akuon and H. Xu, ‘‘Polar coded spatial modulation,’’ IET Commun.,
vol. 8, no. 9, pp. 1459–1466, Jun. 2014.
[29] S. Korada, E. Şaşoğlu, and R. Urbanke, ‘‘Polar codes: Characterization of
exponent, bounds, and constructions,’’ IEEE Trans. Inf. Theory, vol. 56,
no. 12, pp. 6253–6264, Dec. 2010.
[30] W. Wang, L. Li, and K. Niu, ‘‘An efficient construction of polar codes
based on the general partial order,’’ EURASIP J. Wireless Commun. Netw.,
vol. 2019, no. 12, pp. 1–12, 2019.
[31] M. Hanif and M. Ardakani, ‘‘Polar codes: Bounds on Bhattacharyya
parameters and their applications,’’ IEEE Trans. Commun., vol. 66, no. 12,
pp. 5927–5937, Dec. 2018.
[32] H. Mahdavifar and A. Vardy, ‘‘Achieving the secrecy capacity of wiretap
channels using polar codes,’’ IEEE Trans. Inf. Theory, vol. 57, no. 10,
pp. 6428–6443, Oct. 2011.
[33] H. Wang, N. Li, Z. Han, and X. Tao, ‘‘Polar coding for the wiretap channel
with shared key,’’ IEEE Trans. Inf. Forensics Security, vol. 13, no. 6,
pp. 1351–1360, Jun. 2018.
[34] R. A. Chou and A. Yener, ‘‘Polar coding for the multiple access wiretap
channel via rate-splitting and cooperative jamming,’’ IEEE Trans. Inf.
Theory, vol. 64, no. 12, pp. 7903–7921, Dec. 2018.
GANG QIAO received the bachelor’s degree,
in 1996, the master’s degree, in 1999, and the Ph.D.
degree from the College of Underwater Acous-
tic Engineering, Harbin Engineering University,
China, in 2004. Since 1999, he has been with
the College of Underwater Acoustic Engineering,
Harbin Engineering University, China, where he is
currently a Professor and the Associate Dean of
the College of Underwater Acoustic Engineering.
He has already published over 80 papers and holds
seven national invention patents. His current research interests include under-
water communication and networks, detection, and positioning of under-
water targets, and the sonar designed for small carriers. He is a member
of the Acoustical Society of China, a member of the Youth Federation of
Heilongjiang Province, and the Vice Chairman of the Robotics Society of
Heilongjiang Province. He received the National Award for the Outstanding
Scientific and Technological Work, and the Science and Technology Award
for Young Talents in Heilongjiang Province.
SIYU XING received the B.S. and M.S.
degrees from the Harbin Engineering University,
in 2011 and 2014, respectively, where she is cur-
rently pursuing the Ph.D. degree with the College
of Underwater Acoustic Engineering. From 2016
to 2017, she has experienced a one-year study with
the School of Engineering, Newcastle University,
as a joint Ph.D. student, supported by the China
Scholarship Council. Her research focuses on the
physical layer techniques of underwater acoustic
communication, the key technology of underwater acoustic OFDM commu-
nication systems, such as the peak-to-average power ratio reduction, Doppler
estimation, and compensation.
FENG ZHOU received the B.S. and M.S. degrees
in communication and information systems from
Jilin University, China, in 2002 and 2005, respec-
tively, and the Ph.D. degree from the College of
Underwater Acoustic, Harbin Engineering Univer-
sity (HEU), where he is currently an Associate
Professor with the College of Underwater Acous-
tic, for the past 14 years. Throughout his career,
he has worked on primarily on underwater acoustic
communications. His current research interests are
underwater acoustic spread spectrum communications, underwater acous-
tic multi-carrier communications, localization in underwater networks, and
underwater acoustic modem development. He received the National Natural
Science Foundation Award to design efficient communication solutions for
underwater applications, in 2013 and 2015, respectively.
65981