![](https://csdnimg.cn/release/download_crawler_static/15735832/bg1.jpg)
Short communication
Robust adaptive beamforming with random steering vector mismatch
Bin Liao
a
, Chongtao Guo
a
, Lei Huang
a,
n
, Qiang Li
a
, Guisheng Liao
b
, H.C. So
c
a
College of Information Engineering, Shenzhen University, Shenzhen 518060, China
b
National Laboratory of Radar Signal Processing, Xidian University, Xian 710071, China
c
Department of Electronic Engineering, City University of Hong Kong, Hong Kong
article info
Article history:
Received 3 May 2016
Accepted 1 June 2016
Available online 5 June 2016
Keywords:
Robust minimum variance beamforming
Steering vector mismatch
Semidefinite programming
abstract
In this paper, random steering vector mismatches in sensor arrays are considered and probability con-
straints are imposed for designing a robust minimum variance beamformer (RMVB). To solve the re-
sultant design problem, a Bernstein-type inequality for stochastic processes of quadratic forms of
Gaussian variables is employed to transform the probabilistic constraint to a deterministic form. With the
use of convex optimization techniques, the deterministic problem is reformulated to a semidefinite
programming (SDP) problem which can be efficiently solved. In order to overcome the degradation
caused by the presence of the signal-of-interest (SOI) in the training snapshots, two methods with dif-
ferent application conditions to interference-plus-noise covariance matrix (INCM) construction are also
introduced. Additionally, the uncertainty of the sample covariance matrix is taken into account to im-
prove the robustness when the INCM-based approaches are not feasible. Numerical examples are pre-
sented to demonstrate the performances of the proposed robust beamformers in different scenarios.
& 2016 Elsevier B.V. All rights reserved.
1. Introduction
Adaptive beamforming is a fundamental technique for direc-
tional signal transmission and reception. In practice, sensor arrays
suffer from various imperfections such as look direction mismatch,
sensor position perturbation, and local scattering. It is known that
traditional beamforming methods, e.g., Capon beamformer, are
quite sensitive to the mismatch between the actual steering vector
(SV) and nominal one that caused by the array imperfections.
Thus, robust beamforming approaches which are capable of
amending this drawback have attracted extensive attention [1–10].
According to the statistical characteristics, the mismatch can be
modeled to be either deterministic or stochastic in general.
For deterministic SV and covariance matrix mismatches, a
quadratic constraint on the Euclidean norm of the beamformer
weight vector or the mismatch vector/matrix can be imposed to
strengthen the robustness [1–3]. On the other hand, the SV mis-
match may be a random process in certain applications [4–6].In
such cases, it is desirable to maintain the beamformer distortion-
less response only for operational conditions which occur with a
sufficiently high probability rather than for all operational condi-
tions that correspond to the uncertainty set. This leads to the
probability-constrained beamforming [6].
It is shown in [6] that the triangle and Chebyshev's inequalities
can be used to simplify the constraint in the probability-con-
strained beamforming problem. Actually, it is possible to tackle
such a problem by other simplification schemes, which has not
been addressed yet. Towards this end, a Bernstein-type inequality
is first proposed to simplify the probabilistic constraint. Further-
more, motivated by the fact that the presence of SOI in the training
data may cause significant performance degradation [7–10],two
strategies for interference-plus-noise covariance matrix (INCM)
construction are introduced. The first one constructs the INCM in
the absence of interference SV mismatches, whereas the second
takes these mismatches into account in the scenario with strong
interferences. When the INCM-based methods are not applicable,
the array sample covariance matrix (SCM) uncertainty is taken
into account for robustness improvement. Simulation results il-
lustrate that the proposed beamformers can provide satisfactory
performance.
2. Problem formulation
Let us consider an array with M sensor elements receiving
multiple far-field narrowband signals. The array observation at the
time instant k can be written as
θ()= () ( )+ ()+ () ()ksk k kxain 1
00
where
(
k
0
denotes the waveform of the SOI,
θ(
0
corresponds to
the SV with direction-of-arrival (DOA)
θ
0
,
θ()=
∑
()(
=
ksk
a
q
Q
qq
1
is
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/sigpro
Signal Processing
http://dx.doi.org/10.1016/j.sigpro.2016.06.001
0165-1684/& 2016 Elsevier B.V. All rights reserved.
n
Corresponding author.
E-mail address: lhuang@szu.edu.cn (L. Huang).
Signal Processing 129 (2016) 190–194