Diagnosis of Faulty Sensors in Antenna Array using Hybrid Differential Evolution based Compressed Sensing Technique

International Journal of Electrical and Computer Engineering (IJECE)
Vol. 7, No. 2, April 2017, pp. 961~966
ISSN: 2088-8708, DOI: 10.11591/ijece.v7i2.pp961-966  961
Journal homepage: http://iaesjournal.com/online/index.php/IJECE
Diagnosis of Faulty Sensors in Antenna Array using Hybrid
Differential Evolution based Compressed Sensing Technique
Shafqat Ullah Khan1
, M. K. A. Rahim2
, I. M. Qureshi3
, N. A. Murad4
1,2,4
Advanced RF & Microwave Research Group, Department of Communication Engineering,
Faculty of Electrical Engineering, Universiti Teknologi Malaysia, 81310 Skudai, Johor, Malaysia
3
Departement of Electrical Engineering, Air University Islamabad, Pakistan
Article Info ABSTRACT
Article history:
Received Dec 15, 2016
Revised Mar 12, 2017
Accepted Mar 28, 2017
In this work, differential evolution based compressive sensing technique for
detection of faulty sensors in linear arrays has been presented. This algorithm
starts from taking the linear measurements of the power pattern generated by
the array under test. The difference between the collected compressive
measurements and measured healthy array field pattern is minimized using a
hybrid differential evolution (DE). In the proposed method, the slow
convergence of DE based compressed sensing technique is accelerated with
the help of parallel coordinate decent algorithm (PCD). The combination of
DE with PCD makes the minimization faster and precise. Simulation results
validate the performance to detect faulty sensors from a small number of
measurements.
Keyword:
Algorithm
Antenna measurement
Differntial evolution
Fault detection
Parallel coordinate decent Copyright © 2017 Institute of Advanced Engineering and Science.
All rights reserved.
Corresponding Author:
M. K. A. Rahim ,
Advanced RF & Microwave Research Group,
Department of Communication Engineering,
Faculty of Electrical Engineering,
Universiti Teknologi Malaysia,
81310 Skudai, Johor, Malaysia.
Email: mkamal@fke.utm.my
1. INTRODUCTION
Antenna arrays are commonly used in radars, microwave and satellites [1]. The array consist of a
radiating elements. Therefore, the failure chance of one or more sensors increases as the array size increases.
As a result, the pattern generated by the antenna is considerably damaged. Therefore, it is important to design
an efficceint algorithm for the detection of defective elements before correction the patterns. The failure of
sensor disturbs the pattern of the antenna array in terms of sidelobes and nulls [1], [2]. In normal scenario, the
antenna array are located on the ground, the issue can be easily resolved by changing the defective elements
but difficult to replace in case of satellite communications. In literature, there are numbers of available
techniques for array detection [3], [4]. Although, these algorithms are expensive computationally because of
the fact that these techniques require the same measurements as the number of sensors deployed in the
antenna array.
In antenna array detection, the target is to diagnose the defective elements. Compressed sensing
(CS) algorithm have benn developed for the detection of defective elements in the antenna community [5].
CS algorithms reduces number of linear measurements for the detection of defective elements in antenna
array. The sparse vector is the difference of excitation weights of original antenna array and actual array [5].
In CS, sparsity in signals allows to under sample the signal i.e., below the Nyquist sampling criterion and
small number of linear measurements in the signal consists the whole information compared with its
dimensions, so the accurate diagnosis from a small number of linear measurements is possible [6]. The CS

 ISSN: 2088-8708
IJECE Vol. 7, No. 2, April 2017 : 961 – 966
962
has applications in the field of magnetic resonance Imaging [7]. The measurement matrix should follow the
restricted isometry property to avoid information in the signal from damaged. For the detection of the
defective elements with high probability, the required number of linear measurements should be, where m
number of linear measurements are required from an antenna array of length having the number of defective
elements [5].
In this paper, a differential evolution based compressive sensing technique for detection of faulty
sensors in linear arrays has been presented. This algorithms works from taking the linear measurements of
power pattern generated by actual signal. The difference between the measured original reference antenna
array field pattern and collected measurement is minimized using a hybrid Differential Evolution (DE) from
compressive measurements. In the proposed method, the slow convergence of DE is accelerated with the help
of Parallel Coordinate Decent algorithm (PCD). The combination of DE with PCD makes the minimization
faster and precise. Simulation results of proposed compressed sensing based hybrid DE with PCD validate
the detection of the faulty sensors performance from a less number of linear measurements. The rest of the
article is organized as. The proposed solution is discussed in section 2. Section 3 states the numerical
simulations, while section 4 concludes proposed work and recommended future direction.
2. RESEARCH METHOD
Consider an antenna array consist of N number of radiating elements whose power patterns is
given by [8]
 
1
2 1
cos sin
2
N
i n i
n
n
ArrayFactor w kd 

   
   
  
 (1)
where wn is excitation weight of elements, the wave number is k , while the distance between the antenna
element is d . The defective power pattern of the actual array is given by
 
1
2 1
cos sin
2
N
m i n i
n
n m
n
ArrayFactor a kd 


   
   
  
 (2)
where anis nth excitation of the actual array. In (2), an is given by,
0
n
n
with probability
w otherwise
a

 

(3)
where 1  is fraction of defective elements. The refecene and decetive pattern with
16 35N and sidelobeslevel dB  are shown in Figure 1.The difference power pattern between the
original and the actual array is given [5] by
     i i m ip ArrayFactor ArrayFactor    (4)
or it can be follows as
 
1
2 1
cos sin
2
N
i n i
n
n
p x kd 

   
   
  
 (5)
where  , 1,2,3,...,ip i K  and nx is actual array vector is given by
n n nx w a 

IJECE ISSN: 2088-8708 
Diagnosis of Faulty Sensors in Antenna Array Using Hybrid Differential Evolution .... (Shafqat Ullah Khan)
963
To find the diagnosis problem in the antenna array, estimate the difference vector nx . In real
situation, failures are very small as compared to the antenna sensors, thus the difference vector becomes
sparse. The problem of diagnosis of defective elements can be reshaped in the frame of sparsness. The vector
pis given to find the ol -norm which follows the equation
p-Ex=r (6)
   
   
1 1exp sin exp sin
exp sin exp sinK K
jnkd jNkd
jnkd jNkd
 
 
 
 
  
 
  
E
where E is the measurement matrix [9]. If Eis square matrix and invertible then through matrix inversion
the unique solution can be found. In practical environment, the matrix Eis ill-posed which is
underdetermined. Thus to find the sparsest solution. Although, it has non-convex formulation as it involve
N
S
 
 
 
 
exhaustive searches for the defective sensors.
2
2
ˆ argminx
x p Ex  (7)
2
02
ˆ argminx
x p Ex subjectto x S   (8)
The 1l -norm is convex and provides sparsity in the proposed solution, while l -norm is not tractable
nor convex. Therefore replace l -norm by 1l -norm as follows:
 2
12
ˆ argmin
x
x p Ex x  (9)
Now to developed hybrid DE which uses the mutation operator and crossover procedures of the
conventional algorithm. DE is a nature inspired evolutionary technique and was introduced by Storn and
Price to solve real problems [10]. The differential evolution is stochastic based searche technique. Although
the earlier convergence ofdifferntial evolution results in a higher possibility of searching near a local
optimum. The algorithm is based on an operator called mutation, which adds an amount get by the difference
of two randomly selected chromosomes of the population. Finding the difference between two randomly
selected chromosomes from the current population, in fact the algorithm calculating the gradient in that
region and this algorithm is an efficient way to self adaption the mutation operator. The iterative algorithm
called PCD is used to adjust the second good solution in the current population when the fitness of the best
chromosome remains the same during the iteration. This will help us the convergence issue related with the
NP problem of (8). The update equation of PCD is given by the following expression,
 1 1k k s kx x e x    (10)
where the constant  is estimated through line search. The initial value of proposed solution can be either an
estimate of the least square solution or a zero vector. The term es is computed as follows.
  1
1 1
T T
s k ke x diag E E E r

 
 
  
 
  (11)
Here k kr p Ex  is residue and  represents the shrinkage operator
 
0 if u
u u
u if u
u


 



 
   

 

(12)

 ISSN: 2088-8708
964
3. RESULTS AND ANALYSIS
In this simulation, consider a Chebyshev antenna array of 16 number of elements with element
separation is used as test antenna. The pattern denotes -35 dB sidelobe level and nulls directed at desired
angles as depicted in Figure 1. To check the results the power pattern is sampled in the interval of 10 degrees,
19 samples were taken from the damage power pattern. To check the validity of the proposed method, used
the Matlab as a programming tool. At the instant, let us consider that the 3rd and 6th elements in the array
become damage. Now to detect the location of faulty sensors, PCD algorithm is applied. After applying the
PCD algorithm, the number and the location of defective elements is diagnosed. The blue (square) denote the
Chebyshev original array weights, magenta (circle) shows defective elements and red (cross) the diagnosed
fault which is shown in Figure 2. Similarly the same fault location is diagnosed with DE and hybrid DE with
PCD. After applying the proposed DE hybridized PCD, the diagnosed fault is depicted in Figure 3 and
Figure 4. By using the hybrid DE, location and number of defective elements precisely diagnosed than PCD
and DE alone.
The mean square error (MSE) is computed for PCD, DE and hybrid DE. From Figure 5, it obvious
that hybrid DE along PCD outperforms than DE and PCD alone.
Figure 1. The original array and defective array with
3rd
, 6th
elements
Figure 2. Diagnosis of defective elements with PCD
algorithm
Figure 3. Diagnosis of defective elements with DE
algorithm
Figure 4. Diagnosis of defective elements with hybrid
DE with PCD algorithm
0 20 40 60 80 100 120 140 160 180
-120
-100
-80
-60
-40
-20
0
Angle (Degrees)
Far-FieldPattern(dB)
Original array
Defected array
0 2 4 6 8 10 12 14 16
0
0.2
0.4
0.6
0.8
1
Number of Element
NormalizedArrayWeights
Original
Defective
PCD
0 2 4 6 8 10 12 14 16
0
0.2
0.4
0.6
0.8
1
Number of Element
Original
Defective
DE
0 2 4 6 8 10 12 14 16
0
0.2
0.4
0.6
0.8
1
Number of Element
Original
Defective
Hybrid DE

IJECE ISSN: 2088-8708 
Diagnosis of Faulty Sensors in Antenna Array Using Hybrid Differential Evolution .... (Shafqat Ullah Khan)
965
Figure 5. Mean square error of PCD, DE and hybrid DE
4. CONCLUSION
A compressed sensing based array detection algorithm have been developed. The array detection
problem is formulated using a DE, PCD algorithm and further DE hybridized with PCD algorithm. These
algorithms are designed for the detection of defective antenna elements and the numerical simulation confirm
that the proposed hybrid algorithm gives accurate detection of defective sensors in antenna array with a small
number of measurements. The slow and early convergence of DE is prohibited by hybridizing with PCD
algorithm. The hybrid DE-PCD algorithm perform the diagnosis of defective elements more correctly then
PCD and DE alone. This algorithm can be extended to circular arrays.
ACKNOWLEDGEMENTS
The authors thank the Research Management Centre (RMC), Ministry of Higher Education
(MOHE) for supporting the research work; School of Postgraduate Studies (SPS), Communication
Engineering Department, Faculty of Electrical Engineering (FKE), Universiti Teknologi Malaysia (UTM)
Johor Bahru under grant number 12H09 and 03E20.
REFERENCES
[1] S. U. Khan, et al., “Null Placement and Sidelobe Suppression in Failed Array Using Symmetrical Element Failure
Technique and Hybrid Heuristic Computation,” Progress In Electromagnetics Research B, vol. 52, pp. 165-184,
2013.
[2] S. U. Khan, et al., “Correction of Faulty Sensors in Phased Array Radars Using Symmetrical Sensor Failure
Technique and Cultural Algorithm with Differential Evolution,” The Scientific World Journal, vol. 2014, 2014.
[3] J. A. Rodríguez, et al., “Rapid method for finding faulty elements in antenna arrays using far field pattern samples,”
IEEE Trans. Antennas Propag., vol/issue: 57(6), pp. 1679-1683, 2009.
[4] S. U. Khan, et al., “Application of firefly algorithm to fault finding in linear arrays antenna,” World Applied
Sciences Journal, vol/issue: 26(2), pp. 232-238, 2013.
[5] M. D. Migliore, “A compressed sensing approach for array diagnosis from a small set of near field measurements,”
IEEE Trans. Antennas Propag., vol/issue: 59(6), pp. 2127-2133, 2011.
[6] M. Elad, “Sparse and redundant representations: from theory to applications in signal and image processing,”
Springer, 2010.
[7] J. A. Shah, et al., “A modified POCS-based reconstruction method for compressively sampled MR imaging,”
International Journal of Imaging System and Technology, John Wiley & Sons, vol/issue: 24(3), pp. 203-207, 2014.
[8] R. J. Mailloux, “Phased Array Antenna Handbook,” 2nd
ed, Norwood, MA, Artech House, 2005.
[9] S. Ji, et al., “Bayesian compressive sensing,” IEEE Trans. Signal Process, vol/issue: 56(6), pp. 2346-2356, 2008.
[10] R. Storn and K. Price, “Differential evolution: a simple and efficient adaptive scheme for global optimization over
continuous spaces, Technical Report TR-95-012,” International Computer Science Institute, Berkeley, USA, 1995.

 ISSN: 2088-8708
966
BIOGRAPHIES OF AUTHORS
Shafqat Ullah Khan received MS and PhD degree in Electronic Engineering from International
Islamic University, Islamabad, Pakistan and ISRA University Islamabad campus in 2008 and
2015, respective-ly. He is currently a Post Doctorate Fellow at Department of Communication
Engineering, Faculty of Electrical Engineering, Universiti Teknologi Malaysia, with Advanced
RF & Microwave Research Group. His research work mainly focused on detection and
correction of faulty arrays in radar beam forming using evolutionary computational and
compressed sensing techniques.
M. K. A. Rahim was born in Alor Star Kedah Malaysia on 3rd November, 1964. He received
the B Eng degree in Electrical and Electronic Engineering from University of Strathclyde, UK in
1987. He obtained his Master Engineering from University of New South Wales, Australia in
1992. He graduated his PhD in 2003 from University of Birmingham, U.K., in the field of
Wideband Active Antenna. From 1992 to 1999, he was a lecturer at the Faculty of Electrical
Engineering, Universiti Teknologi Malaysia. From 2005 to 2007, he was a senior lecturer at the
Department of Communication Engineering, Faculty of Electrical Engineering, Universiti
Teknologi Malaysia. He is now a Professor at Universiti Teknologi, Malaysia. His research
interest includes the design of active and passive antennas, dielectric
Ijaz Mansoor Qureshi, did his BS in Avionics CAE PAF in 1976, MS in Electrical Engineering
from Middle East Technical Univer-sity Ankara in 1980. Finally he did his PhD in HEP from
University of Toronto, Canada. He is currently Professor at Department of Electrical
Engineering, Air University, Islamabad. His research activities are in Digital Signal Processing
and Soft Computing.
Noor Asniza Murad obtained her first degree in 2001 from Universiti Teknologi Malaysia
(UTM), Malaysia, with Honours, majoring in telecommunication engineering. Shortly after
graduated, she joined UTM as a tutor attached to the Department of Radio Communication
Engineering (RaCED) ,UTM. She received her MEng. in 2003 from the same university and
later has been appointed as a lecturer in April 2003. She obtained her Ph.D from UK in 2011 for
research on micromachined millimeterwave circuits under supervision of Professor Micheal
Lancaster. Her research interests include antenna design for RF and microwave communication
systems, millimeterwave circuits design, and antenna beamforming. Currently, Noor Asniza
Murad is a member of IEEE (MIEEE), Member of Antenna and Propagation (AP/MTT/EMC)
Malaysia Chapter, and a Senior Lecturer at Faculty of Electrical Engineering UTM.

Diagnosis of Faulty Sensors in Antenna Array using Hybrid Differential Evolution based Compressed Sensing Technique

More Related Content

What's hot (19)

Similar to Diagnosis of Faulty Sensors in Antenna Array using Hybrid Differential Evolution based Compressed Sensing Technique (20)

More from IJECEIAES (20)

Recently uploaded (20)

Diagnosis of Faulty Sensors in Antenna Array using Hybrid Differential Evolution based Compressed Sensing Technique