Volume 6, Issue 5, May – 2021 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Robust Near-field Narrowband Beamformer Against

Steering Angle Mismatch and Distance Error Using
Diagonal Loading Technique
Rony Tota Dr. Md Selim Hossain
Dept. of Electrical and Electronic Engineering Dept. of Electrical and Electronic Engineering
Rajshahi University of Engineering and Technology Rajshahi University of Engineering and Technology
Rajshahi, Bangladesh Rajshahi, Bangladesh

Abstract:- Steering angle mismatch and distance error a broadband array require tapped delay line (TDL) filter in
severely degrade the performance of adaptive front of each element [2] that increases computational
beamformer. This digression is more acute in near-field complexity. So this paper investigates about narrowband
beamforming due to inconvenience of estimating signal’s processing.
accurate location especially the radial distance. As the
radial distance of signals vary in near-field which is The optimal beamformer which is also known as
constant in case of far-field. So robustness of near-field Minimum variance distortionless response (MVDR) steered
array against these errors is a must. Diagonal loading is the main beam in desired signal direction, cancelling
the most popular and easeful method to increase the interference solving a constrained optimization problem.
robustness of beamformer. This paper presents three These constrained assumes desired signal position and
robust techniques such as Fixed Diagonal Loading (FDL), direction. If any mismatch occurs that is if the signal directs
Optimal Diagonal Loading (ODL) and Variable Diagonal from slightly different position or direction then performance
Loading (VDL) for near-field narrowband beamformer of adaptive beamformer is hampered [3-5]. So a robust
and compares their performance with the optimal adaptive beamformer is requisite to remove these look
beamformer. From the investigation it’s observed that the direction disparity and distance error.
proposed robust techniques show superior performance
than conventional with high directivity in the desired Many existing near-field research has done in [6-9]
signal direction, low side lobe level and sharp interference emphasizing only on the optimization of array response for
and false signal rejection capability. some certain noise and interference environment. Robust and
efficient antenna array processor against look direction error is
Keywords:- (Beamforming, Near-field, Antenna-array, discussed in [10] for far-field signals. Various far-field pattern
MVDR, FDL, ODL, VDL) synthesis techniques for robust broadband array processor are
mentioned in [11]. Diagonal loading is the most used robust
I. INTRODUCTION beamforming method and several diagonal techniques for far-
field array processor are applied in [12]. These various loading
Beamforming nowadays are being the most emerging method can be transferred for near-field array processor.
topic in signal processing scenario as its providing
comprehensive application in the era of RADAR, SONAR, So the aim of this paper is to design a robust near-field
Air traffic control, Wireless and Satellite communication etc. narrowband FDL, ODL and VDL based array processor and to
Most of the literature about beamforming has done for far- compare the array response pattern of these robust techniques
field case. In far-field signal’s radial distance is assumed at with the adaptive beamformer. Simulation results show that
infinite distance so a far-field signal arrives at array are any steering angle and distance disparity harshly demote the
considered as plane wave front and that assumption simplify performance of optimal beamformer and ODL and VDL
the analysis [1]. In near-field case signal is located at a limited robust techniques restore the optimum performance. This
distance that is near-field steering vector is a function of radial paper also investigates that how look direction disparity and
distance, elevation and azimuthal angle. So signal received by distance error affects on array signal power and SINR
near-field array is not a plane wave front rather it spherical. A performance. Analysis shows that proposed ODL offers least
severe performance degradation occurs if a near-field signal is signal cancellation rate at look direction error but in the
analyzed as plane wave. This paper investigates near field presence of distance error proposed VDL performs better than
beamformer assuming source signal as spherical wave front. ODL. VDL based robust beamformer maintains almost
constant SINR over the array processor than robust FDL and
Narrowband signal has higher sensitivity and longer ODL method.
range than the broadband. Interference effect on narrowband
signal is also lower than wideband. In some application such
as in mobile telephony, sound recording, microphone array
etc. narrowband is more preferable than wideband [1]. Besides

IJISRT21MAY824 www.ijisrt.com 991

Volume 6, Issue 5, May – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
II. NEAR-FIELD SYSTEM MODEL Here nl(t) denotes total noise of lth element and E[nk(t)
nl(t)] = 0 for k≠l and E[nk(t) nl(t)] = n2 for k=l means that
A signal is considered as near-field signal if the radial noise of each element is uncorrelated with another also this
distance of signal from the reference be r < 2D2/where D noise signal is uncorrelated with source signal. Here n2
denotes maximum dimension of antenna known as antenna denotes noise power.
aperture and  be the wavelength of signal [13]. Let consider a
linear antenna array with L elements and distance of each In this paper array response pattern, signal power and
element from another be d also consider the origin of the SINR are investigated at two point of view. Firstly power
coordinate system shown in Fig. 1 be the time reference. pattern, signal power and SINR are analyzed in the presence
of look direction error that is steering angle is varied keeping
z radial distance of signal as constant shown in Fig. 2. Secondly
the same parameters are observed against distance error that is
ith Source the radial distance of signal is changed while the signals
(ri, 𝜃i, 𝜑i)
𝑥i -𝑥l steering angle is constant like that of Fig. 3
lth
element 𝜃i 𝑥i
y
𝑥l
𝜑i 𝑦

𝑥 r In
Fig.1: Near field coordinate system. rs Interference n
Signal source
Let the focal point of a ith near field signal is (ri i i ), here rI2 Interference 2
ri denotes radial distance, i is the angle of elevation and i be Interference 1
the azimuth angle. Assuming that the signal from that focal rI1
point directs towards the array and time taken by this spherical 𝜃s= 𝜃I1= 𝜃I2= ……= 𝜃In
wave front to reach the array and measured from the lth
x
element to origin is given by [14] 1 2 3 …. l …. L
il (ri i i ) = (xi - xl - xi / c (1) d
Where Fig. 3: Distance discrimination of signals while angle of
xi = ri (sinicos i + sinisin i + cosi (2) arrival is constant.

be the position vector of ith signal source. Similarly xl Total array correlation matrix R of this near-field processor
is given by
denotes position vector of lth element and c is the speed of
R=Ps S0H S0 + Pi SiH Si +n2I (5)
propagation of spherical wave front. For L elements antenna
array let signal induced on array is x1(t), x2(t)…………xL(t). In
Here Ps be the power of the signal of interest, S0 denotes
vector notation induced signal x(t) of array is given by
look direction steering vector, Si is the steering vector at the
interference direction, Pi be the interference power and I
x(t) = [x1(t), x2(t)…………xL(t)]T (3)
denotes identity matrix.
If the distance between lth element and ith source be ril
III. BEAMFORMING TECHNIQUES
then induced signal at lth element is given by [15]
Techniques of beamformer defines the performance of
xl(t) = ri * si (t+il) / ril + nl(t) (4)
beamforming array processor. Conventional beamformer
y
shows maximum response in the look direction but can’t
detect and nullify interference sources. Optimal array
processor has the ability to sense and cancel out these
Interference n
r In directional interferences but at steering vector mismatch it
Interference 2
rI2 provides worst performance. It considers actual signal of
Interference 1 interest (SOI) as interference and create null in that direction.
rI1
Robust FDL, ODL and VDL can recover these problems. In
Signal source
rs this section Optimal, FDL, ODL and VDL techniques are
𝜃In
described shortly.
𝜃I2 rs = rI1= rI2= ……= rIn
𝜃I1 A. Optimal Beamformer
𝜃s
x
1 2 3 …. l …. L Optimal or MVDR beamformer maximizes output SINR
d
without the knowledge of interference power and direction
Fig. 2: Direction of arrival (DOA) of signals keeping radial with only knowing SOI direction [16]. The weight of optimal
distance as constant. beamformer is selected as

IJISRT21MAY824 www.ijisrt.com 992

Volume 6, Issue 5, May – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
w=R-1 S0 / S0H R-1 S0 (6) IV. PERFORMANCE EVALUATION

Here R denotes total array correlation matrix and weight Array power pattern of near-field narrowband based
w is the solution of the following optimization problem optimal, robust FDL, ODL, and VDL beamformer against
minimize wH R w signals incidence angle and radial distance of signals are
w analyzed in this section. Output power and SINR variation
Subject to wH S0 = 1 against DOA and distance error are also shown.

This constraints minimizes noise and interference power,

keeping unity response at SOI direction. Minimization of
noise and interference maximizes output SNR.

B. Fixed Diagonal Loading Method

Optimal beamformer can’t provide precise performance
at steering vector errors. FDL techniques recover the problem
of this optimal beamformer by updating the array correlation
matrix diagonally with a fixed value

RFDL = R+I (7)


denotes FDL factor and the value of this constant is fixed
to 10n2. To determine this value is a challenging task which
is discussed in [17].

C. Robust Optimal Diagonal Loading Technique

ODL performs better than FDL in near-field array Fig. 4: Array power pattern with 10 look direction error when
elements no. of the array = 50, signal power = 1.0, assumed signals
processor against direction of arrival and distance error. Array
position (r, ) = (11800, 900), position of accurate signal (r, )
correlation matrix for ODL is given by [18]
= (11810, 900), interference no. = 4, power of each interference =
10, interference position: (11200, 900), (11400, 900), (111200,
RODL = R+I (8) 900), (111400, 900), noise power = 0.01.
.
The constant for ODL computed using SOI power, noise
power, norm of steering vector with and without considering
steering vector mismatch and defined as

n2+ PsS0Sac (9)

Where  S0  and  Sac  are the norm of steering vector

without and with considering the DOA and Distance error.
The distortion bound of steering vector,  is given by [19]

 = max ( S0 - Sac  ) (10)

D. Robust Variable Diagonal Loading

Correlation matrix for VDL is defined as

RVDL = R+R-1*I (11)

VDL correlation matrix is updated using inverse of Fig. 5: Output signal power variation against look direction error
when elements no. of the array = 50, signal power = 1.0, interference
original correlation matrix that provides superior weight no. = 4, power of each interference = 10, interference position:
adaption capabilities against steering vector error. The (11200, 900), (11400, 900), (111200, 900), (111400, 900),
constant is calculated [20] using (9) and (10). noise power = 0.01.

IJISRT21MAY824 www.ijisrt.com 993

Volume 6, Issue 5, May – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Fig. 4 shows power pattern of MVDR and three robust Table 1: Output signal strength variation against DOA angle
FDL, ODL and VDL beamformer with respect to signals error of conventional MVDR and proposed robust FDL, ODL
incidence angle. MVDR is unable to overcome look direction and VDL techniques
error it takes accurate look direction as interference i.e. create Beamform Output signal strength variation in dB for
null at exact signal direction. FDL can’t recover the problem ing several DOA error
of optimal beamformer rather it acts like that of MVDR. ODL techniques Witho 0.5o 1o 1.5o 2o
and VDL provide robustness against look direction error i.e. if ut dispari dispari dispari dispari
signals position is slightly deviated from its accurate position error ty ty ty ty
ODL and VDL don’t take it as error rather these show MVDR 0 -4.477 -11.28 -16.78 -21.07
maximum array response in accurate signal direction. While FDL 0 -0.468 -1.802 -3.633 -5.642
providing robustness VDL performs better than ODL as ODL ODL 0 0.0382 0.0689 0.0905 0.102
can’t generate null at exact interference direction (e.g. fourth VDL 0 0.0593 0.1203 0.1827 0.2457
interference positioned at 1400 in Fig. 4) but VDL has almost
done.

Fig. 5 shows output signal strength variation of four

beamformer against signals steering angle error. MVDR
which is not robust shows worst performance. For the
conventional MVDR beamformer the output signal decays at a
large rate with increasing the look direction error. It is
observed from this figure that the proposed robust techniques
provides less signal cancellation. Fig. 6 clarifies that ODL
based robust beamformer has less signal power variation i.e.
less signal cancellation rate. Table 1 displays a comparison of
signal strength variation for different beamforming techniques.
Fig.7 expresses output SINR change with look direction error.
Conventional MVDR beamformer has sharp SINR changes
with error in incidence angle. VDL performs best among these
three robust beamformer in terms of SINR variation. Fig. 7: Output SINR variation against look direction error when
elements no. of the array = 50, signal power = 1.0, interference no. =
4, power of each interference = 10, interference position: (11200,
900), (11400, 900), (111200, 900), (111400, 900), noise power
= 0.01.

Near-field beamforming array processor is not only the

function of signals incidence angle but also of signals radial
distance. This section discusses the performance of
conventional optimal beamformer against signals radial
distance error and how the problems of this optimal
beamformer can be overcome using various robust techniques.

20

0
Beam pattern (dB) ----->

-20
Assumed
signal
-40 distance
Fig. 6: Output signal power variation of proposed ODL and proposed
VDL against look direction error when elements no. of the array = -60
50, signal power = 1.0, interference no. = 4, power of each Accurate signal
distance
interference = 10, interference position: (11200, 900), (11400, -80
900), (111200, 900), (111400, 900), noise power = 0.01. MVDR
-100 Robust FDL
Robust ODL
Robust VDL
-120
5 10 15 20 25 30 35 40 45
Radial Distance (λ) ----->
Fig. 8: Array power pattern with 1 look direction error when
elements no. of the array = 70, signal power = 1.0, assumed signals
position (r, ) = (20500, 900), position of accurate signal (r, )
= (21500, 900), interference no. = 4, power of each interference =

IJISRT21MAY824 www.ijisrt.com 994

Volume 6, Issue 5, May – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
10, interference position: (10500, 900), (15500, 900), (30500, 900), (15500, 900), (30500, 900), (40500, 900), noise power =
900), (40500, 900), noise power = 0.01. 0.01.

Fig. 8 compares the beampattern of various Table 2: Output signal strength variation against distance error
beamformer. Conventional MVDR can’t differentiate the of conventional MVDR and proposed robust FDL, ODL and
assumed and accurate signal’s distance i.e. when incidence VDL techniques
signal is slightly deviated from its accurate distance MVDR Beamforming Output signal strength variation in dB for several distance
techniques error
beamformer can’t provide maximum response rather it takes
Without 0.1 0.2 0.3 0.4
this deviation as interference and create null at accurate error disparity disparity disparity disparity
position.
MVDR 0 -18.37 -29.59 -36.47 -41.40
FDL 0 -4.41 -11.25 -16.86 -21.28
Output signal strength and output SINR variation ODL 0 -0.1162 -0.2321 -0.3474 -0.4622
against signals radial distance are shown at Fig. 9 and Fig. 11 VDL 0 0.0215 0.0493 0.0834 0.1237
respectively. Robust ODL and VDL performs better than
MVDR and FDL method. Fig. 10 clarify the performance of 40
robust ODL and VDL. VDL shows less signal power
variation compared to ODL and VDL display positive power 30
error in according with the distance error. Table 2 compares
the signal strength of various beamformer in the presence of MVDR
20

SINR (dB) ----->

distance error. One can observes from Fig.11 that the Robust FDL
proposed VDL based robust beamformer has comparatively Robust ODL
Robust VDL
slower SINR degradation property. 10

0
0
Output Signal Strength(dB) ----->

-5
-10
-10

-15 -20
-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
-20
Distance Error (λ ) ----->
-25 Fig. 11: Output SINR variation against look direction error when
MVDR
Robust FDL
elements no. of the array = 70, signal power = 1.0, interference no. =
-30 Robust ODL 4, power of each interference = 10, interference position: (10500,
Robust VDL
900), (15500, 900), (30500, 900), (40500, 900), noise power =
-35
-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.01.
Distance Error (λ ) ----->
Fig. 9: Output signal strength variation against look direction error V. CONCLUSION
when elements no. of the array = 70, signal power = 1.0, interference
no. = 4, power of each interference = 10, interference position: In this paper three robust beamforming techniques are
(10500, 900), (15500, 900), (30500, 900), (40500, 900), noise discussed to remove the problems of conventional MVDR
power = 0.01. beamformer as severe performance degradation of MVDR
beamformer occurs in the presence of steering angle mismatch
0.1 and distance error. MVDR beamforming array processor
provides maximum array response in the desired signal
Output Signal Strength(dB) ----->

0
direction creating null at each interference position but this
-0.1 performance don’t sustain if the signals position is displaced
from its previous position. MVDR takes the signal
-0.2 displacement as interference signal i.e. provides null at
-0.3
accurate signals position. This paper proposes three robust
beamforming techniques FDL, ODL and VDL. Fig. 4 and Fig.
-0.4 8 elucidates that FDL can’t provide maximum radiation
pattern at accurate signal position but it has less signal and
-0.5 Robust ODL
Robust VDL
SINR cancellation rate than MVDR. ODL and VDL
-0.6 techniques can sense any slight steering angle variation or any
-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 slight radial distance changes but don’t receive it as
Distance Error (λ ) ----->
interference or false signal rather continues to maintain
Fig. 10: Comparison of ODL and VDL robust techniques while maximum array response in accurate signal direction. Fig.4
measuring output signal strength against signals distance error when
elements no. of the array = 70, signal power = 1.0, interference no. =
confirms that at look direction error VDL performs better than
4, power of each interference = 10, interference position: (10500, ODL as ODL don’t indicate null at exact interference
direction. Table 1 shows that at 20 error output signal power

IJISRT21MAY824 www.ijisrt.com 995

Volume 6, Issue 5, May – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
variation of ODL and VDL are 0.1020 dB and 0.2457 dB [12]. M. S. Hossain, L. C. Godara and M. R. Islam, “Robust
respectively i.e. ODL based beamformer has least signal and efficient broadband beamforming algorithms in the
cancellation rate. But from Table 2 one can observe that presence of steering angle mismatch using variable
output power variation for ODL based beamformer is -0.4622 loading,” 2011 IEEE Third Latin-American Conference
dB and for VDL is 0.1237 dB at 0.4 disparity i.e. at on Communications, pp. 1-5, 2011, doi:
distance error VDL beamformer has little power change. Fig. 10.1109/LatinCOM.2011.6107392.
7 and Fig. 11 prove that VDL has higher capability to retain [13]. B. D. Steinberg, “Priciples of Apperture and Array
constant SINR over the array processor. System Design: Including Random and Adaptive
Arrays”. New York: Wiley, 1996.
REFERENCES [14]. M. R. Islam and M. Reed. “Analysis of near field
broadband PIC antenna array processor with orthogonal
[1]. P. Thomas, R. Verburgh, M. Catrysse, and D. interference beam.” 2013 IEEE International
Botteldooren , “Design of a microphone array for near- Conference on Acoustics, Speech and Signal
field conferencing applications”, Proceedings of Processing , pp. 4164-4168, 2013.
Meetings on Acoustics 30, 055001, 2017. [15]. M. R. Islam, L. C. Godara and M. S. Hossain, “A
[2]. O. L. Frost, “An algorithm for linearly constrained computationally efficient near field broadband
adaptive array processing,” Proc. IEEE, vol. 60, no. 8, beamformer,” 2011 IEEE Third Latin-American
pp. 926–935, Aug. 1972. Conference on Communications, Belem, Brazil, pp.1-5,
[3]. D. H. Johnson and D. E. Dudgeon, “Array Signal 2011.
Processing: Concepts and Techniques”, NJ: Prentice [16]. G.Gan, Lu, and Z. Yi. “Automatic computation of
Hall, 1993. diagonal loading factor for robust adaptive
[4]. B. D. Van Veen and K. M. Buckley, “Beamforming: a beamforming based on Gaussian distribution,” AEU-
versatile approach to spatial filtering,” ASSP International Journal of Electronics and
Magazine, IEEE, vol. 5, pp. 4-24,1988. Communications , Vol. 67, No. 7, 570-573, 2013.
[5]. A. B. Garshmen, “Robust adaptive beamforming in [17]. G. Yuri, Nechaev, and P. Ilia, “Probability of false
sensor arrays,” Aeu Intl. J. Electronics, peaks occurring via circular and concentric antenna
Communications, vol. 53, no. 6, pp. 305-314, Dec. arrays DOA estimation,” 39th International Conference
1999. on Telecommunications and Signal Processing (TSP),
[6]. M. R. Islam, L. C. Godara and M. S. Hossain, “A IEEE , 2016.
computationally efficient near field broadband [18]. M. F. Reza, and M. S. Hossain. “Performance
beamformer,” in Communications (LATINCOM), 2011 investigation of robust concentric circular antenna array
IEEE Latin-American Conference on, Brazil, pp. 1-5, beamformer in the presence of look direction disparity,"
Oct., 2011. AEU-International Journal of Electronics and
[7]. E. Fisher and B. Rafaely, “Near field spherical Communications, Vol. 82, 52-57, 2017.
microphone array processing with radial filtering.” [19]. J. R. Lin, Q. Peng, and H. Shao, “On diagonal loading
IEEE Trans. Audio, Speech,Language Process, vol. 19, for robust adaptive beamforming based on worst-case
no. 2, pp. 256–265, Feb. 2011. performance optimization,” ETRI J., vol. 29, no. 1, pp.
[8]. P. Chen, X. Tian and Y. Chen, “Optimization of the 50–58, Feb. 2007.
Digital Near-Field Beamforming for Underwater 3-D [20]. J. Gu and P. Wolfe, “Robust adaptive beamforming
Sonar Imaging System,” IEEE Transactions on using variable loading,” in Proc. 4th IEEE Workshop
Instrumentation and Measurement, vol.59, no.2, pp.415- Sensor Array Multichannel Process, pp. 1–5, Jul. 2006.
424, Feb. 2010.
[9]. M. Palmese, and A. Trucco, “An Efficient Digital CZT
Beamforming Design for Near-Field 3-D Sonar
Imaging,” IEEE Journal of Oceanic Engineering,
vol.35, no.3, pp.584-594, July 2010.
[10]. M. S. Hossain, G. N. Milford, M. C. Reed and L. C.
Godara, “Efficient Robust Broadband Antenna Array
Processor in the Presence of Look Direction Errors,” in
IEEE Transactions on Antennas and Propagation, vol.
61, no. 2, pp. 718-727, Feb. 2013, doi:
10.1109/TAP.2012.2225014.
[11]. M. S. Hossain, G. N. Milford, M. C. Reed and L. C.
Godara, “Robust Efficient Broadband Antenna Array
Pattern Synthesis Techniques,” in IEEE Transactions on
Antennas and Propagation, vol. 62, no. 9, pp. 4537-
4546, Sept. 2014, doi: 10.1109/TAP.2014.2332363

IJISRT21MAY824 www.ijisrt.com 996