Hindawi

Wireless Communications and Mobile Computing

Volume 2021, Article ID 9983615, 9 pages
https://doi.org/10.1155/2021/9983615

Research Article
Coherent Target Direction-of-Arrival Estimation for Coprime
Arrays: From Spatial Smoothing Perspective

Dongming Wu , Fangzheng Liu, Zhihui Li, and Zhenzhong Han

National University of Defense Technology, Hefei 230000, China

Correspondence should be addressed to Dongming Wu; wudongming163@163.com

Received 5 March 2021; Accepted 1 July 2021; Published 19 July 2021

Academic Editor: Luis Castedo

Copyright © 2021 Dongming Wu et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

In this paper, we investigate the issue of direction-of-arrival (DOA) estimation of multiple signals in coprime arrays. An algorithm
based on multiple signal classification (MUSIC) and forward and backward spatial smoothing (FBSS) is used for DOA estimation of
this signal caused by multipath and interference. The large distance between adjacent elements of each subarray in the coprime
arrays will bring phase ambiguity issues. According to the feature of the coprime number, the ambiguity problem can be
eliminated. The correct DOA estimation can be obtained by searching for the common peak of the spatial spectrum and finding
the overlapping peaks in the MUSIC spectrum of the two subarrays. For the rank deficit problem caused by the coherent signal,
the FBSS algorithm is used for signal preprocessing before the MUSIC algorithm. Theoretical analysis and simulation results
show that the algorithm can effectively solve the rank deficiency and phase ambiguity problems caused by coherent signals and
sparse arrays in the coprime arrays.

1. Introduction beamforming technology [9]. In radar target detection,

DOA estimation is the basis for achieving high-precision
Array signal processing is a branch of the signal processing direction finding [10]. Therefore, it is of great significance
field and is widely used in radar, sonar, satellite, wireless to study how to improve the accuracy of DOA estimation.
communications, seismology, and other fields [1, 2]. Array The performance of DOA estimation is determined by the
signal processing is based on a group of spatially arranged resolution, accuracy, the number of distinguishable targets,
array antennas to process the signal [3]. The purpose of array etc. [11]. In response to these aspects, corresponding theoret-
signal processing is to enhance useful target signals, suppress ical and applied research has been carried out at home and
noise, and obtain signal spatial information. Compared with abroad, which has enabled the rapid development of DOA
a single antenna, the use of an antenna array has outstanding estimation theory [12–14].
advantages in terms of spatial resolution, receiving sensitivity, The past researches have proposed a large number of
and anti-interference [4]. Thus, array signal processing has DOA estimation algorithms for different array models, such
made rapid progress in research and engineering applications as the uniform linear array, L-shaped linear array, and uni-
in the past 30 years [5]. form circular array [15–17]. In the traditional array structure,
DOA estimation of space signal is a basic problem in the angle ambiguity is avoided by setting the spacing of array
array signal processing. DOA estimation is to estimate the elements no more than half wavelength. However, when the
direction of arrival of the signal by receiving the target echo frequency of the received signal is high, too small array
data through the array antenna in the noise or interference element spacing will cause larger mutual coupling, and the
environment. And it is a kind of direction-finding technique physical array layout is difficult to achieve. At the same time,
[6, 7]. In wireless communication, accurate DOA estimation high resolution means a larger array aperture, and more
of the signal source can improve communication quality [8]. physical array elements will further increase the system cost
And it can improve physical layer security combined with and complexity. Sparse arrays can overcome the structural
2 Wireless Communications and Mobile Computing

limitations of traditional arrays by increasing the array ele- In this paper, the MUSIC algorithm and the FBSS algo-
ment spacing [18] and thus have been developed and widely rithm are combined to estimate the DOA of coherent signals
used, such as the Minimum Redundancy Array (MRA) [19, based on the coprime matrix model under the condition of
20], Nested Array (NA), and coprime array (CPA) [21–23]. multipath and interference, and the formulas to solve the sig-
The coprime formation is composed of two subarrays, nal coherence and angle ambiguity under the coprime matrix
and the spacing between the subarrays is mutually prime are given. Finally, the DOA estimation method for coherent
[24]. Compared with the traditional uniform array, the signals is simulated, and the simulation results show the
element spacing of the coprime array is greater than half a effectiveness of the method.
wavelength. The increase in the element spacing brings the The remainder is given as follows: Section 2 outlines the
advantages of an increase in the array aperture and a signifi- basic array signal model of the coprime array. In Section 3,
cant reduction in the mutual coupling effect between the proposed method for coherent target DOA estimation
elements and significantly improves the estimation accuracy based on coprime arrays is presented, and the problem of
and resolution [25]. The DOA of two uniform subarrays of phase ambiguity and rank deficiency is discussed together
the coprime array is estimated, respectively. According to with its elimination method. Numerical simulations and con-
the relatively prime characteristics of the element spacing of clusions are presented in Sections 4 and 5, respectively.
the two subarrays, it is proved that the DOA estimation
results of the two subarrays are unique [26, 27]. The coprime Notations. Throughout the paper, we use the lowercase
array which does not reduce the array aperture of the original (uppercase) boldface symbols to represent vectors (matrices).
array is simple to implement, and the estimation accuracy is ð⋅ÞT and ð⋅ÞH denote the transpose and the conjugate trans-
greatly improved compared with the uniform array with the pose, respectively. I N denotes N × N identity matrix, diag ð⋅Þ
same number of antennas [28–30]. denotes the diagonal matrix operator, and E½⋅ denotes the
DOA estimation algorithms mainly include traditional expectation operator.
beamforming, subspace algorithm, and maximum likelihood
estimation [31–33]. Among them, the beamforming method 2. Array Signal Model
has larger error and low resolution; the maximum likelihood
algorithm uses the probability distribution of the signal and The coprime array is a sparse array constructed by using the
adopts the high-dimensional search method, which has a property of a coprime number. It is composed of two uni-
large amount of computation. The subspace algorithm uses form linear arrays. Assuming that the number of subarray
the orthogonality of signal and noise subspace to realize angle elements is M and N, the spacing between two subarrays is
estimation, which requires less computation but cannot Nλ/2 and Mλ/2, respectively, where M and N are coprime
process coherent signals [34]. Generally, the minimum res- integers and λ represents the wavelength of the received sig-
olution that can be achieved under a certain array length is nal. The first element of the two subarrays coincides, which is
called the Rayleigh Resolution Limit, and the method that also called the reference element. The coprime array contains
exceeds the Rayleigh Resolution Limit is called the superre- a total of elements, and the positions of M + N − 1 elements,
solution algorithm. Multiple signal classification (MUSIC) and the positions of the elements are
proposed in 1979 and estimating signal parameters via
rotational invariance techniques (ESPRIT) proposed in 


1986 belong to subspace algorithm and are also early clas- λ λ
d = Nm ∪ Mn , ð1Þ
sical superresolution methods [35, 36]. No matter the 2 2
MUSIC algorithm or ESPRIT algorithm, it is necessary for
the array element to receive the uncorrelated signal. At this
time, the covariance matrix of the source is a full rank where 0 ≤ m ≤ M − 1, 0 ≤ n ≤ N − 1:
matrix, so that the covariance matrix of the signal can be Figure 1(a) is a schematic diagram of the structure of a
eigendecomposed and the signal subspace and noise sub- coprime array with M + N − 1 elements. For the convenience
space can be distinguished. of analysis, the coprime matrix is divided into two subarrays,
Most signals are coherent signals in the actual application in which the black dot represents subarray 1 and the hollow
environment because of the multipath effect and complex dot represents subarray 2, as shown in Figure 1(b). In fact, the
transmission channel [37, 38]. For early DOA estimation two subarrays are in a straight line and share the first element.
algorithms such as MUSIC and ESPRIT, they are all based It is assumed that there are far-field narrow-band signals
on subspace for DOA estimation. When the received signal from different directions in the space, the incident angle is
is correlated, the eigenvector corresponding to the source sig- θk , k = 1, 2, 3, ⋯, K, and the output noise of each element is
nal cannot be obtained by decomposing the subspace eigen- a complex Gaussian distribution with zero mean value, which
values. Therefore, DOA estimation of coherent source are independent of each other and have the same average
signals has always been a difficult problem, which is also power σ2 . The output of the mth element can be expressed as
the focus of spectral estimation. In order to distinguish
coherent signals accurately, the spatial smoothing method,
singular value decomposition method (SVD method), matrix K
decomposition method (MD method), and Toeplitz method xm ðt Þ = 〠 aðθk Þsk ðt Þ + nm ðt Þ: ð2Þ
are developed [39–41]. k=1
Wireless Communications and Mobile Computing 3

N𝜆/2 N𝜆/2
Co-prime array
M+N–1
M𝜆/2 M𝜆/2
(a) An example of coprime array
N𝜆/2
Subarray 1
1 2 3 4 M–1
1 2 3 N–1
Subarray 2

M𝜆/2
(b) Two coprime uniform linear subarrays

Figure 1: Basic structure of coprime linear array.

If the first element is selected as the reference element, the to the estimation accuracy. Generally, the higher the degree
output of the subarray with M elements is of freedom is, the higher the positioning accuracy will be.

xM ðt Þ = AM ðθÞsðt Þ + nM ðt Þ, ð3Þ 3. DOA Estimation of Coherent Signals

where AM ðθÞ = ½aM ðθ1 Þ, aM ðθ2 Þ, ⋯, aM ðθK Þ. The steering Due to the interference effect of coherent signals, the number
vector of the Kth source is expressed as of subspaces processed by the ordinary DOA estimation
algorithm will be reduced and affect the direction-finding
h iT
accuracy. However, the FBSS algorithm does not appear in
aM ðθK Þ = 1, e−jNπ sin θK , ⋯, e−jN ðM−1Þπ sin θK : ð4Þ
such a situation. Based on this idea, we decompose the
coprime array into two uniform subarrays. For each subarray,
Source vector sðtÞ = ½s1 ðtÞ, s2 ðtÞ, ⋯, sK ðtÞT . Similarly, the the FBSS algorithm and the traditional MUSIC algorithm are
output of the subarray with N elements is combined to process the coherent signal. By analyzing the
DOA results of the two subarrays, the correct target angle
xN ðt Þ = AN ðθÞsðt Þ + nN ðt Þ, ð5Þ can be obtained, and the problem of rank deficiency caused
by phase ambiguity and coherent signal is solved.
where AN ðθÞ = ½aN ðθ1 Þ, aN ðθ2 Þ, ⋯, aN ðθK Þ. The steering
vector of the Kth source is expressed as 3.1. Spatial Smoothing on Subarrays. Coherent signals are
easily generated in signal transmission due to the complex
h iT space environment. The appearance of coherent sources
aN ðθK Þ = 1, e−jMπ sin θK , ⋯, e−jMðN−1Þπ sin θK : ð6Þ
may lead to serious degradation of DOA estimation perfor-
mance. In the traditional MUSIC algorithm based on subspace,
Because the noise is independent of each other, the noise the covariance matrix of the received data needs to be full rank,
and the signal are independent of each other, the covariance but the covariance matrix of the coherent source is not full
matrix of the noise is σ2 I, and the covariance matrix of the rank, the signal eigenvectors diverge into the noise subspace,
output of the two subarrays: and the singular value decomposition cannot completely dis-
    tinguish the signal subspace from the noise subspace, which
RM = E xM xM H = AM E ssH AM H + σ2 IM , AM Rss AM H + σ2 IM , leads to deterioration of DOA estimation performance.
The basic idea of the spatial smoothing algorithm is to
RN = AN Rss AN H + σ2 IN : divide the array into several overlapping subarrays and use
ð7Þ the covariance matrix of the received data of subarrays to
replace the original covariance matrix. By sacrificing a certain
In the DOA estimation based on the coprime array, the effective array aperture, the covariance matrix of the received
array aperture is greatly expanded by the construction of a data is restored to full rank, so as to achieve the preprocessing
virtual array model. At the same time, the ranks of covariance operation of decoherence.
matrices constructed by different methods are also different, The covariance matrix uses the autocorrelation relation-
but generally, the virtual array degree of freedom of the ship between signals to extract information. In practical
coprime array is far greater than that of the physical array. applications, the maximum likelihood function of the covari-
The degree of freedom is an important sign that the antenna ance matrix is usually calculated by selecting a large enough
array can estimate the number of targets or sources. The number of snapshots to approximate the ideal covariance
higher the degree of freedom is, the more sources the array matrix. In this case, the estimated covariance matrix of the
can estimate. Besides, the degree of freedom is proportional output data can be expressed as
4 Wireless Communications and Mobile Computing

1 L the most classic superresolution DOA estimation algorithm,

RM = 〠 x ðlÞxH
M ðl Þ, which obtains the cell direction by searching the spectrum
L l=1 M
ð8Þ peak in the spatial domain. Compared with multidimen-
1 L sional algorithms such as maximum likelihood (ML) and
R N 〠 x N ð l Þx H
N ð l Þ: weighted subspace fitting (WSF), the algorithm has less com-
L l=1 putation. The basic idea of the MUSIC algorithm is to eigen-
decompose the covariance matrix of the array output data to
As shown in Figure 2, consider the subarray with the obtain the signal subspace corresponding to the signal com-
number of M elements and the spacing of Nλ/2 in the ponent and the noise subspace orthogonal to the signal com-
coprime matrix. Under the forward space smoothing ponent and then use the orthogonality of the two subspaces
algorithm, the equidistant linear array is divided into L sub- to estimate the signal parameters.
arrays by sliding, and each subarray has n elements, where The covariance matrices of the two submatrices are
n = M − L + 1. eigendecomposed, respectively, to obtain
In this case, the output of the first forward subarray can
be expressed as RM = USM 〠 USM H + UNM 〠 UNM H ,
SM NM
f
xl ðt Þ = ½xl ðt Þ, xl+1 ðt Þ, ⋯, xl+n−1 ðt ÞT = AM Dl−1 sðt Þ + nl ðt Þ ð1 ≤ l ≤ LÞ,
ð14Þ
RM = USN 〠 USN + UNN 〠 UNN H :
H
ð9Þ SN NN

where AM = ½aM ðθ1 Þ, aM ðθ2 Þ, ⋯, aM ðθK Þ is the n ∗ K Among them, matrices USM ∈ ℂM∗K and USN ∈ ℂN∗K are
dimension direction matrix and aM ðθÞ is the n-dimension the signal subspaces formed by the eigenvectors corresponding
guidance vector. D = diag ðe jð2πNd/λÞ sin θ1 , e jð2πNd/λÞ sin θ2 , ⋯, to K large eigenvalues in R ̂ M and R ̂ N , matrices ∑SM ∈ ℂK∗K
e jð2πNd/λÞ sin θK Þ is a rotation-invariant matrix between and ∑SN ∈ ℂ K∗K
are the diagonal matrices formed by K large
subarrays. ̂
eigenvalues in RM and R ̂ N , matrices UNM ∈ ℂM∗ðM−KÞ and
The covariance matrix of the lth forward submatrix can UNN ∈ ℂ M∗ðN−KÞ
are the noise subspaces formed by the eigen-
be expressed as vectors corresponding to M − K and N − K small eigenvalues
h i  ̂ M and R
in R ̂ N , and the matrices ∑NM and ∑NN are diagonal
H
f f f
Rl = E xl ðt Þxl ðt ÞH = AM Dl−1 RS Dl−1 AH
M + σ I: ð10Þ
2 matrices composed of M − K and N − K small eigenvalues in
̂ M and R
R ̂ N , respectively (these small eigenvalues are equal,
which is the noise power σ2 ).
The forward spatial smoothing covariance matrix is
Under ideal conditions, the signal subspace US and the
defined as
noise subspace UN are orthogonal to each other, so the array
flow pattern vector aH ðθÞ corresponding to the signal sub-
1 L f space is also orthogonal to the noise subspace UN , namely,
Rl = 〠R : ð11Þ
L l=1 l
aH ðθÞUN = 0: ð15Þ
Similarly, if the subarray is divided from the last element
of the array, the covariance matrix of backward spatial In practice, the steering vector and the noise subspace can-
smoothing can be obtained as follows: not be completely orthogonal due to the existence of other
noises. Usually, the minimum optimization search process is
1 L b used to find the minimum value to realize the direction-of-
Rb = 〠R : ð12Þ arrival estimation. This process can be expressed as
L l=1 l

θMUSIC = argθ min aH ðθÞU ̂H

̂ NU N aðθÞ: ð16Þ
Because the backward smooth array is the conjugate
reverse order of the forward smooth array, the relation
between Rb and Rl is the conjugate reverse order invariant. Based on the orthogonality between the signal subspace
Although the one-way smoothing algorithm can solve the and the noise subspace, the spectral function of the MUSIC
problem of coherent signals, it sacrifices more array aperture. space power spectrum of the two subarrays can be
The FBSS algorithm can increase the number of estimable expressed as
cells by simultaneously performing forward and backward
1
smoothing. The covariance matrix is the average of forward PMUSIC M = ,
smoothing and backward smoothing covariance matrices M ðθÞUM UM aM ðθÞ
aH ̂ ̂H
ð17Þ
1
1 PMUSIC = :
R f b = R f + Rb : ð13Þ N
̂NU
a N ð θ ÞU ̂H
N aN ðθÞ
H
2

3.2. DOA Estimation of Subarrays. The MUSIC algorithm is Among them, the value θ range is generally ð−π/2, π/2Þ.
Wireless Communications and Mobile Computing 5

Forward spatial smoothing

Subarray L
Subarray 2
Subarray 1
1 2 ... N N+1 M–N+1 ... M

Subarray 1
Subarray L–1
Subarray L

Backward spatial smoothing

Figure 2: Schematic diagram of FBSS algorithm.

The accurate DOA estimate can be obtained by searching 1, ⋯, M − 1. Combined with (19), the condition of phase
for the coincident peaks of the two subarray spectral func- ambiguity is obtained as follows:
tions of the coprime array.
2PM 2PN
3.3. Ambiguity Elimination. The calculation formula of phase = : ð21Þ
defuzzification for incoherent sources is given in Reference N M
[31]. For coherent sources, if there is phase ambiguity after
using the spatial smoothing algorithm for the subarray with After simplification, we can get NPN = MPM . Since M and
M elements, it can be seen from (13) that the steering vector N are relatively prime, it cannot make the equation hold in the
between the real angle θk and the blurred angle θk should be range of value; that is to say, θk does not exist and there is no
equal, that is, angle ambiguity. Therefore, the unique DOA estimation can
be determined by using the spatial smoothing algorithm and
 MUSIC algorithm, respectively, for the subarrays of the
AM ðθk Þ = AM θk , coprime array, and then finding the overlapped peaks in the
  ð18Þ two groups of spectrum.
exp ð−jNπ sin ðθk ÞÞ = exp −jNπ sin θk :
3.4. Complexity Analysis. The spatial smoothing algorithm,
SVD algorithm, and Toeplitz algorithm can process coherent
signals well, and the computational complexity of these three
After simplification, we get decoherence algorithms increases gradually. At present, the
spatial smoothing algorithm has the least amount of compu-
 2P tation; that is, the time of DOA processing is the shortest. In
sin ðθk Þ − sin θk = M , ð19Þ addition, spatial smoothing technology is also more mature,
N
which is a more practical algorithm for processing coherent
signals. The uniform linear array with M elements can distin-
where PM is a nonzero integer, θk , θk ∈ ð−π/2, π/2Þ. For any θk guish 2M/3 coherent targets by using the spatial smoothing
and θk , it must satisfy jsin ðθk Þ − sin ðθk Þj < 2, that is, j2P/Nj algorithm. And the virtual element number of the coprime
< 2. The value range of PM can be −ðN − 1Þ, −ðN − 2Þ, ⋯, array with two subarray elements M and N is OðMNÞ. There-
−1, 1, ⋯, N − 1; there are 2ðN − 1Þ values in total. Consider fore, Oð2MN/3Þ coherent targets can be distinguished by the
coprime array with this algorithm. In the same case, only O
that θk and θk can be exchanged. In addition to the real
ð2ðM + N − 1Þ/3Þ coherent targets can be distinguished by
angle, there are N − 1 fuzzy angles. That is to say, for a
the uniform linear array with this algorithm.
single subarray M whose element spacing is Nλ/2 in the
coprime array, there must be phase ambiguity. There are
N peaks in the MUSIC spectrum using spatial smoothing,
4. Simulation Results
and the N − 1 peaks correspond to the fuzzy angle. In this section, we have carried out the corresponding simu-
In the same way, when considering the single subarray N lation analysis to prove the effectiveness of FBSS and MUSIC
whose element spacing is Mλ/2 in the coprime array, the algorithms for coherent signals under the coprime array
fuzzy angle needs to meet the requirement: model. In the simulation process, the number of elements
of two subarrays of the coprime array is M = 7 and N = 5,
 2P and the spacing between elements is 5λ/2 and 7λ/2, where
sin ðθk Þ − sin θk = N : ð20Þ λ is half wavelength. For a fair comparison, a 12 uniform lin-
M
ear array with half-wavelength spacing is also simulated with
the FBSS and MUSIC algorithms. Consider two coherent sig-
The value range of PN can be −ðM − 1Þ, −ðM − 2Þ, ⋯,−1, nals in the space, which are incident from 0° and 30° to the
6 Wireless Communications and Mobile Computing

–10

Amplitude (dB)
–20

–30

–40

–50

–60
–90 –60 –30 0 30 60 90
Degree

Figure 3: Spatial spectrum of DOA estimation simulation of two subarrays for coherent sources.

–10
Amplitude (dB)

–20

–30

–40

–50

–60

–70
–90 –60 –30 0 30 60 90
Degree

CA
ULA

Figure 4: Spatial spectrum of DOA estimation simulation of uniform linear array and coprime array.

coprime array, respectively, and the noise is Gaussian 14 peaks, 12 of which are fuzzy angles. However, the com-
white noise. The searching steps for all methods are set mon spectral peak formed by the two subarrays is only at 0°
to be 0:02° . and 30° , which proves the correctness of the algorithm for
DOA estimation of coherent signals.
4.1. Spatial Spectrum. We then show the spatial spectrum Under this condition, we further compare the DOA
using FBSS and MUSIC algorithms in Figure 3, where estimation spectrum of 11 elements uniform linear array
we assume the signal to noise (SNR) as 10 dB and snap- and coprime array. The specific results are shown in
shot n = 200. The red spectral line is the subarray spectrum Figure 4. Through the comparison of DOA estimation
with M = 7, and the blue spectral line is the subarray spec- spectrum peaks, we can intuitively find that two coherent
trum with N = 5. It can be seen from the previous derivation signals, whether uniform linear array or coprime array,
that phase ambiguity will be generated when using spatial can be well distinguished. But the coprime array is better
smoothing and MUSIC algorithm for a single subarray of than the uniform linear array in suppressing interference.
coprime array. For a subarray with an element spacing Because the number of virtual elements of the coprime
of Nλ/2, estimating a DOA will produce N − 1 ambiguity array is much larger than that of the uniform linear
angles. Therefore, the subarray with M = 7 has 10 peaks, 8 array, it has a higher degree of freedom and better esti-
of which are fuzzy angles. And the subarray with N = 5 has mation performance.
Wireless Communications and Mobile Computing 7

2.5

RMSE (degree)
1.5

0.5

0
0 200 400 600 800 1000
Snapshot

CA
ULA
CRB

Figure 5: RMSE versus the number of snapshots.

2.5

2
RMSE (degree)

1.5

0.5

0
–5 0 5 10 15 20 25
SNR (dB)

CA
ULA
CRB

Figure 6: RMSE versus the SNR.

4.2. Root Mean Square Error (RMSE). In this simulation, we ðiÞ

and θk and θk are the estimate and real values of the k
study the RMSE performance of the two arrays under differ- th DOA for the nth trial, n = 1, 2, 3, ⋯, N. The targets
ent configurations. The root mean square error (RMSE) of are located at θ1 = 10° , θ2 = 20° , θ3 = 30° , and θ1 , θ2 are
the estimates is defined as the performance metric: coherent signals. For each simulation scenario, S = 500
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rounds of Monte-Carlo runs are conducted. The Cramer-
u
u 1 N K h i Rao bound (CRB) is plotted as a benchmark.
RMSE = t
ðiÞ
〠 〠 θk − θk , ð22Þ Figure 5 depicts the RMSEs of different configurations in
NK n=1 k=1
terms of SNR, where the number of snapshots is 200. In
Figure 6, we compare the RMSEs of the two arrays versus
where N 0 denotes the times of Monte-Carlo simulations the number of snapshots, where the SNR is set as 5 dB. It is
8 Wireless Communications and Mobile Computing

obvious that the performance of all these configurations [5] Z. Zheng, C. Yang, W. Q. Wang, and H. C. So, “Robust DOA
improves with the increase of the SNR and number of Estimation Against Mutual Coupling With Nested Array,” IEEE
snapshots. But the performance of the coprime array is Signal Processing Letters, vol. 27, no. 4, pp. 1360–1364, 2020.
better than that of the uniform linear array in any case. [6] Z. Ye and X. Xu, “DOA estimation by exploiting the symmetric
Even in the case of low snapshot number and low signal- configuration of uniform linear array,” IEEE Transactions on
to-noise ratio, the coprime array can also show good Antennas and Propagation, vol. 55, no. 12, pp. 3716–3720, 2007.
DOA estimation performance. [7] J. He, L. Li, and T. Shu, “Sparse nested arrays with spatially
Although the phase ambiguity of a single subarray of the spread square acoustic vector sensors for high accuracy under-
coprime array is caused by the large element spacing, accu- determined direction finding,” IEEE Transactions on Aero-
rate DOA estimation can be achieved by comparing the peak space and Electronic Systems, 2021.
values of the two subarrays. Through the simulation experi- [8] J. Wang, H. Xu, G. J. Leus, and G. A. Vandenbosch, “Experi-
ment, we can see that compared with the uniform linear mental Assessment of the Co-Array Concept for DoA Estima-
tion in Wireless Communications,” IEEE Transactions on
array, the coprime array using the spatial smoothing algo-
Antennas and Propagation, vol. 66, no. 6, pp. 3064–3075, 2018.
rithm greatly improves the resolution and reduces the com-
[9] K. Almidfa, G. V. Tsoulos, and A. Nix, “Performance evalua-
putational complexity.
tion of direction-of-arrival (DOA) estimation algorithms for
mobile communication systems,” in VTC2000-Spring. 2000
5. Conclusions IEEE 51st Vehicular Technology Conference Proceedings (Cat.
No.00CH37026), pp. 1055–1059, Tokyo, Japan, 2000.
In this paper, we use FBSS and MUSIC algorithms for DOA [10] X. Zhang, L. Xu, L. Xu, and D. Xu, “Direction of departure
estimation of coherent signals based on the structure of (DOD) and direction of arrival (DOA) estimation in MIMO
coprime arrays, where the spatial spectrum of each decom- radar with reduced-dimension MUSIC,” IEEE Communica-
posed subarray can generate spectral peaks at the actual tions Letters, vol. 14, no. 12, pp. 1161–1163, 2010.
DOAs and multiple ambiguous DOAs simultaneously. And [11] M. C. Dogan and J. M. Mendel, “Applications of cumulants to
we solve the phase ambiguity by finding the common spectral array processing .I. Aperture extension and array calibration,”
peaks in the spectrum of the two subarrays. Theoretical anal- IEEE Transactions on Signal Processing, vol. 43, no. 5,
pp. 1200–1216, 1995.
ysis and simulation results show that the algorithm can effec-
tively process DOA estimation of coherent signals, and the [12] J. He, L. Li, and T. Shu, “Sparse nested arrays with spatially
spread orthogonal dipoles: high accuracy passive direction
coprime array has better performance than the uniform lin-
finding with less mutual coupling,” IEEE Transactions on
ear array. However, some spatial degrees of freedom are Aerospace and Electronic Systems, p. 1, 2021.
sacrificed when using the spatial smoothing algorithm.
[13] G. He, M. Song, X. He, and Y. Hu, “GPS signal acquisition
How to increase the spatial degrees of freedom and improve
based on compressive sensing and modified greedy acquisition
the direction-finding accuracy under low snapshot numbers algorithm,” IEEE Access, vol. 7, pp. 40445–40453, 2019.
will be our further research direction.
[14] G. He, M. Song, S. Zhang, P. Song, and X. Shu, “Sparse
GLONASS signal acquisition based on compressive sensing
Data Availability and multiple measurement vectors,” Mathematical Problems
in Engineering, vol. 2020, Article ID 9654120, 11 pages, 2020.
Data will be made available on request. [15] Zhongfu Ye, Jisheng Dai, Xu Xu, and Xiaopei Wu, “DOA esti-
mation for uniform linear array with mutual coupling,” IEEE
Transactions on Aerospace & Electronic Systems, vol. 45,
Conflicts of Interest no. 1, pp. 280–288, 2009.
[16] N. Xi and L. Liping, “A computationally efficient subspace
The authors declare that they have no conflicts of interest. algorithm for 2-D DOA estimation with L-shaped array,” IEEE
Signal Processing Letters, vol. 21, no. 8, pp. 971–974, 2014.
References [17] J.-J. Fuchs, “On the application of the global matched filter to
DOA estimation with uniform circular arrays,” IEEE Transac-
[1] H. Krim and M. Viberg, “Two decades of array signal process- tions on Signal Processing, vol. 49, no. 4, pp. 702–709, 2001.
ing research: the parametric approach,” IEEE Signal Processing [18] J. He, Z. Zhang, T. Shu, and W. Yu, “Sparse nested array with
Magazine, vol. 13, no. 4, pp. 67–94, 1996. aperture extension for high accuracy angle estimation,” Signal
[2] J. Benesty, J. Chen, and Y. Huang, “Microphone array signal Processing, vol. 176, p. 107700, 2020.
processing,” Journal of the Acoustical Society of America, [19] D. A. Linebarger, I. H. Sudborough, and I. G. Tollis, “Differ-
vol. 125, no. 6, pp. 4097-4098, 2008. ence bases and sparse sensor arrays,” IEEE Transactions on
[3] J.-J. Jiang, F.-J. Duan, J. Chen, Y.-C. Li, and X.-N. Hua, “Mixed Information Theory, vol. 39, no. 2, pp. 716–721, 1993.
near-field and far-field sources localization using the uniform [20] A. Moffet, “Minimum-redundancy linear arrays,” IEEE Trans-
linear sensor array,” IEEE Sensors Journal, vol. 13, no. 8, actions on Antennas and Propagation, vol. 16, no. 2, pp. 172–
pp. 3136–3143, 2013. 175, 1968.
[4] X. Wang, M. Huang, and L. Wan, “Joint 2D-DOD and 2D- [21] J. Yang, G. Liao, and J. Li, “An efficient off-grid DOA estima-
DOA estimation for coprime EMVS–MIMO radar,” Circuits, tion approach for nested array signal processing by using
Systems, and Signal Processing, vol. 40, no. 6, pp. 2950–2966, sparse Bayesian learning strategies,” Signal Processing,
2021. vol. 128, pp. 110–122, 2016.
Wireless Communications and Mobile Computing 9

[22] P. Pal and P. P. Vaidyanathan, “Nested arrays: a novel [39] S. U. Pillai and B. H. Kwon, “Forward/backward spatial
approach to array processing with enhanced degrees of free- smoothing techniques for coherent signal identification,” IEEE
dom,” IEEE Transactions on Signal Processing, vol. 58, no. 8, Transactions on Acoustics Speech and Signal Processing,
pp. 4167–4181, 2010. vol. 37, no. 1, pp. 8–15, 1989.
[23] P. P. Vaidyanathan and P. Pal, “Sparse sensing with co-prime [40] Chongying Qi, Yongliang Wang, Yongshun Zhang, and Ying
samplers and arrays,” IEEE Transactions on Signal Processing, Han, “Spatial difference smoothing for DOA estimation of
vol. 59, no. 2, pp. 573–586, 2011. coherent signals,” IEEE Signal Processing Letters, vol. 12,
[24] P. P. Vaidyanathan and P. Pal, “Theory of sparse coprime no. 11, pp. 800–802, 2005.
sensing in multiple dimensions,” IEEE Transactions on Signal [41] C. Zhou, Z. Shi, Y. Gu, and X. Shen, “DECOM: DOA estima-
Processing, vol. 59, no. 8, pp. 3592–3608, 2011. tion with combined MUSIC for coprime array,” in 2013 Inter-
[25] Z. Weng and P. M. Djurić, “A search-free DOA estimation national Conference on Wireless Communications and Signal
algorithm for coprime arrays,” Digital Signal Processing, Processing, pp. 1–5, Hangzhou, China, 2013.
vol. 24, pp. 27–33, 2014.
[26] C. Zhou, Z. Shi, Y. Gu, and N. A. Goodman, “DOA estimation
by covariance matrix sparse reconstruction of coprime array,”
in 2015 IEEE International Conference on Acoustics, Speech
and Signal Processing (ICASSP), pp. 2369–2373, South
Brisbane, QLD, Australia, 2015.
[27] Z. Shi, C. Zhou, Y. Gu, N. A. Goodman, and F. Qu, “Source
estimation using coprime array: a sparse reconstruction per-
spective,” IEEE Sensors Journal, vol. 17, no. 3, pp. 755–765,
2017.
[28] P. Pakrooh, L. L. Scharf, and A. Pezeshki, “Modal analysis
using co-prime arrays,” IEEE Transactions on Signal Process-
ing, vol. 64, no. 9, pp. 2429–2442, 2016.
[29] Q. Wu, F. Sun, P. Lan, G. Ding, and X. Zhang, “Two-dimen-
sional direction-of-arrival estimation for co-prime planar
arrays: a partial spectral search approach,” IEEE Sensors Jour-
nal, vol. 16, no. 14, pp. 5660–5670, 2016.
[30] J. Shi, G. Hu, X. Zhang, F. Sun, W. Zheng, and Y. Xiao,
“Generalized co-prime MIMO radar for DOA estimation with
enhanced degrees of freedom,” IEEE sensors journal, vol. 18,
no. 3, pp. 1203–1212, 2018.
[31] E. Gonen and J. M. Mendel, “Applications of cumulants to
array processing. III. Blind beamforming for coherent signals,”
IEEE Transactions on Signal Processing, vol. 45, no. 9,
pp. 2252–2264, 1997.
[32] X. Wang, L. T. Yang, D. Meng, M. Dong, K. Ota, and H. Wang,
“Multi-UAV cooperative localization for marine targets based
on weighted subspace fitting in SAGIN environment,” EEE
Internet of Things Journal, 2021.
[33] F. Wen, J. Shi, and Z. Zhang, “Closed-form estimation algo-
rithm for EMVS-MIMO radar with arbitrary sensor geome-
try,” Signal Processing, vol. 186, p. 108117, 2021.
[34] C. L. Liu and P. P. Vaidyanathan, “Remarks on the spatial
smoothing step in coarray MUSIC,” IEEE Signal Processing
Letters, vol. 22, no. 9, pp. 1438–1442, 2015.
[35] R. Schmidt, “Multiple emitter location and signal parameter
estimation,” IEEE Transactions on Antennas & Propagation,
vol. 34, no. 3, pp. 276–280, 1986.
[36] P. Roy and T. Kailath, “ESPRIT-estimation of signal parame-
ters via rotational invariance techniques,” IEEE Transactions
on Acoustics, Speech, and Signal Processing, vol. 37, no. 7,
pp. 984–995, 1989.
[37] J. H. Cozzens and M. J. Sousa, “Source enumeration in a corre-
lated signal environment,” IEEE Transactions on Signal Pro-
cessing, vol. 42, no. 2, pp. 304–317, 1994.
[38] E. BouDaher, F. Ahmad, and M. G. Amin, “Sparsity-based
direction finding of coherent and uncorrelated targets using
active nonuniform arrays,” IEEE Signal Processing Letters,
vol. 22, no. 10, pp. 1628–1632, 2015.