322 IEEE COMMUNICATIONS LETTERS, VOL. 21, NO. 2, FEBRUARY 2017
An Eigendecomposition-Based Approach to Blind Beamforming in a
Multipath Environment
Liang Zhang, Bin Liao, Senior Member, IEEE, Lei Huang, and Chongtao Guo
Abstract—This letter devises a new approach to blind
beamforming for multipath coherent signal reception. It is shown
that in the scenarios where the interferences are moderately
stronger than the signal, namely, the interference-to-signal ratio
is mildly high, the composite steering vector (CSV) of multipath
coherent signals can be estimated via the eigendecomposition of
a reduced covariance matrix, which subtracts the components of
interferences and noise from the array covariance matrix. The
proposed eigendecomposition-based blind beamformer (EBB) is
thereby obtained by substituting the estimate of CSV into the
minimum variance distortionless response beamformer. It is also
shown that the proposed EBB weight vector is actually equal to
the principal eigenvector of the reduced covariance matrix. The
performance of the proposed EBB is demonstrated by numerical
simulations.
Index Terms— Blind beamforming, multipath propagation,
coherent signals, eigendecomposition.
I. INTRODUCTION
B
EAMFORMING is known to be a spatial filtering tech-
nique which utilizes sensor arrays to enhance the desired
signal and to suppress those unwanted interference and noise.
Great success has been achieved in wireless communication
by taking advantage of beamforming techniques with antenna
arrays [1]. Given the knowledge of propagation parameters
such as the array response and direction-of-arrival (DOA) of
the desired signal, the optimal beamformer which achieves the
maximum signal-to-interference-plus-noise ratio (SINR) can
be designed by minimizing the output power subject to a con-
straint of unity array response at the DOA of the desired signal.
This leads to the well-known minimum variance distortionless
response (MVDR) beamformer [2].
In scenarios where the knowledge of transmission channel,
or content of the desired signal or array response is unknown,
it is more desirable to design the so-called blind beamformer
by exploiting various structural properties of the problem.
For instance, by utilizing the cyclostationarity property of
the desired signal, a class of blind beamforming approaches,
known as spectral self-coherence restoral (SCORE), was pro-
posed in [3]. On this basis, in [4], additional sparse constraint
on the beampattern was employed to suppress array gains
in the whole spatial angular range for sidelobe and interfer-
ence suppression. Moreover, by making use of the structural
Manuscript received August 22, 2016; revised October 23, 2016; accepted
November 4, 2016. Date of publication November 8, 2016; date of current
version February 9, 2017. This work was supported in part by the National
Natural Science Foundation of China under Grants 61401284 and 61601307,
in part by the Foundation of Shenzhen under Grants JCYJ20140418091413566
and JCYJ20160422102022017. The associate editor coordinating the review
of this letter and approving it for publication was D. Calin. (Corresponding
author: Bin Liao.)
The authors are with the College of Information Engineering, Shenzhen Uni-
versity, Shenzhen 518060, China (e-mail: zhangliang19840425@gmail.com;
binliao@szu.edu.cn; lhuang@szu.edu.cn; ctguo@szu.edu.cn).
Digital Object Identifier 10.1109/LCOMM.2016.2626365
information in higher-order statistics (fourth order cumulants),
a least squares approach to blind beamforming was presented
in [5]. Additionally, the property of the constant modulus of
many communication signals has also been widely utilized for
blind beamforming [6].
Besides the insufficiency of knowledge of propagation para-
meters in communications, the high correlation or coherence
among the signals and/or interferences due to multipath prop-
agation is also an issue of great importance, as the ignorance
of signal coherence could result in considerable beamforming
performance deterioration. In the past several decades, many
efforts have been devoted to the investigation of this problem
with the aid of, say, split-polarity transformation [7], spatial
smoothing [8], diversity gain [9] and sequential blind con-
stant modulus algorithm [10]. More recently, a decorrelation-
based blind beamformer (DBB) is proposed for the multipath
environment where the desired signal rather than interfer-
ence consists of coherent components [11]. In principle, the
DBB algorithm adopts the spatial smoothing scheme to decor-
relate the signals. Thus, this method is restricted to some
specific array geometries and its performance may be limited
due to the reduced effective array aperture. This motives us to
design a new blind beamformer with improved performance.
Particularly, in this letter an eigendecomposition-based blind
beamformer (EBB) for multipath coherent signal reception is
devised. Under the condition that each interference is mildly
stronger than the desired signal, the interference covariance
matrix can be well estimated via the eigendecomposition of
the array covariance matrix. Thus, the composite steering
vector (CSV) of the multipath signals is estimated (subject to
a scalar) as the principal eigenvector of the reduced covariance
matrix which eliminates the components of interference and
noise. As a result, the EBB weight vector can be obtained
by substituting the CSV estimate into the MVDR beam-
former. Moreover, it is shown that the resulting beamformer
is also equivalent to the principal eigenvector of the reduced
covariance matrix. Hence, the proposed EBB can be simply
obtained via eigendecomposition. Examples are carried out
to illustrate the improvements of the proposed method over
existing approaches.
II. P
ROBLEM STATEMENT
Assume that P coherent signals caused by multipath
propagation and Q uncorrelated interferences impinge on an
M-element array with DOAs {θ
p
}
P
p=1
and {φ
q
}
Q
q=1
, respec-
tively, where Q + 1 < M. The output of the array at the
nth time instant is expressed as
x(n) =
P
p=1
h(θ
p
)s(n)α
p
+
Q
q=1
h(φ
q
)s
q
(n) + e(n)
= ds(n) + i(n) + e(n) (1)
1558-2558 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.