Upload
buinguyet
View
216
Download
1
Embed Size (px)
Citation preview
Ph.D Thesis ; Rajashree J W ; 2011 Page 12
CHAPTER 2
LITERATURE REVIEW AND APPROACH TO THE PROBLEM
In this chapter a review of the technical literature on solving the antenna array thinning
problem is carried out. This is followed by a brief description of the problem definition
and the investigative approach taken up for the study.
Background information about EAA and related mathematical relationships are given in
Annexure A.
2.1 LITERATURE REVIEW
2.1.1 THINNING PROCESS An EAA consists of a large number of discrete antenna elements placed in a regular grid.
In many communication and Radar systems these antenna elements are in a regular linear
or planar grid. Characterization of these antenna arrays is generally done through their
Radiation Patterns. Typical metrics include Radiation Gain, Side lobe level (SLL), Half-
Power Beam width (HPBW), Directivity etc. These aspects are discussed in Appendix A.
The process of ‘Thinning’ involves reducing total number of active elements in the
antenna array without causing major degradation in system performance as measured
through these metrics. By thinning, periodicity in the array is broken and hence it can
also be considered as a subset of non-uniformly spaced array. Historically, a vast
majority of research efforts in this context were concentrated on non-uniformly spaced
arrays during 1960s. The major focus then was to find antenna element positions or inter
element spacings using various techniques.
Most of these methods were deterministic in nature, using measurable and quantifiable
knowledge of the array physics to determine an optimal result. The user develops cost
function criteria based upon array parameters of interest to exploit knowledge of the
deterministic qualities of the array. Typical deterministic optimization algorithms applied
to the synthesis problem include Newton’s Method, Simplex Method, LMS, Conjugate
Ph.D Thesis ; Rajashree J W ; 2011 Page 13
Gradient, as well as many other constrained and unconstrained linear/non-linear
programs.
The trends found in the literature related to thinning in antenna arrays can be covered
under the following heads
a) Analytical methods
b) Numerical algorithms
c) Stochastic Search
d) Others
2.1.2 ANALYTICAL METHODS
Unz [2.1] studied a linear array with general arbitrarily distributed elements. The far zone
pattern was expressed using Jacobi expansion. The approach used a matrix form solution
which requires the manipulation of matrices of the order of number of elements. It
involves expressing the radiation pattern in a series expansion, truncating the expansion
and finding the solution of the matrix to find the desired spacing. He found that the
additional degree of freedom created by the random distribution of elements allowed him
to achieve the same performance as of an equally spaced array with fewer elements. The
complexity of this formulation, has limited its practical applications.
An analysis of various directional arrays with variable spacings was made by D D King,
R F Packard and R K Thomas [2.2]. It was an empirical approach to select a set of
spacings according to a specified law such as logarithmic spacing. Non-monotonically
increasing spacings were also investigated and found that they seem to offer promise of a
reasonable radiation pattern. It was found, fewer elements were required for a desired
beam width compared to an equally spaced array. Lacking better method during 1960’s,
this approach gave a better start to the design of unequally spaced arrays. However, it
provided little guidance for new designs other than those studied earlier.
Another approach using Poisson’s Sum formula and the introduction of a new role to the
source function was proposed for unequally spaced array antennas by Ishimaru [2.3].
This is done by converting the total radiation from the set of radiators from series
summation to an integral and applying the change of variables of integration using source
Ph.D Thesis ; Rajashree J W ; 2011 Page 14
function and source number function. By this, the linear array is replaced by a series of
continuous source distribution, each of which has such a phase variation that the peak
occurs at different angles. Reduction of side lobes and suppression of grating lobes are
then attempted based on one of the conventional techniques such as Taylor.
Sandler [2.4] tried to find the equivalence between equally and unequally spaced arrays.
Lo [2.5] used a probabilistic approach to analyze large randomly spaced arrays. It was
found that sidelobe level is closely linked with number of elements in the array but
weakly connected to the aperture size whereas directivity was related to aperture size.
Design technique used by Mafett[2.6], Willey[2.7], Ogg [2.8] ,and Allen[2.9] for
unequally spaced arrays are based on density tapering, or space tapering, in which the
density of equal-amplitude elements is made proportional to the amplitude of the aperture
illumination of a conventionally designed array. The primary reason for employing
density tapering is to achieve low side lobes without the need for amplitude tapering. The
density tapered technique, when applied to large arrays, seems to give satisfactory results.
In a space tapered array that is derived from a reference array using an illumination taper
which has a maximum in the centre of the array, the minimum active element spacing
will occur at the centre. As the number of elements increases, the minimum spacing
decreases; conversely, as the number of elements decreases, the minimum spacing
increases.
The minimum allowable spacing will be determined principally by the physical size of
the element and, perhaps, by the effects of mutual coupling between elements. The
corresponding maximum number of elements will provide the best approximation to the
desired illumination function. This method is based on the fact that the illumination taper
can be expressed as a function of its moments.
An analytical iterative method was suggested by Harrington[2.10]. This approach starts
with an a priori reasonable set of element spacings by some means; these spacings are
later perturbed about their initial value and effects studied. The side lobes can be reduced
in height to approximately 2/N times the main lobe level, where N is the number of
Ph.D Thesis ; Rajashree J W ; 2011 Page 15
elements, with main beam width remaining essentially the same as for the uniform array.
However, this method seems to be applicable only to control side lobes close to the main
beam.
A second iterative method reported by Andreasen [2.11] differs from the first in that it
does not start with a a priori set of spacings; it builds up the design one element at a time.
One of the arrays designed using this method is reported to have 21 elements and is 76
wavelengths long when used as a broadside array. The 3-db beam width is 0.74 degree,
the side lobe level -7.4 db. Such an approach is probably not suitable for applications
such as radar.
The method of synthesizing a space tapered linear array involves approximation of a
desired current distribution over the aperture. For an antenna possessing a continuous
aperture, the far-field pattern can be calculated from a knowledge of the currents in the
aperture. The side lobe structure and beam width of the radiation pattern are controlled by
the use of an appropriate illumination function which tapers the aperture current density
[2.12]. In amplitude tapered array, the current sheet of the continuous aperture is
approximated by discrete current sources, the elements. The illumination taper is
obtained by varying the relative amplitudes of the elements in the array. For arrays, in
which the elements are closely spaced (about 0.5 wavelength), there is little difference
between the patterns formed by the continuous aperture and by the discrete array
aperture. The statistically tapered arrays are useful when the number of elements is large
and when it is not practical to employ an amplitude taper to achieve low side lobes.
2.1.3 NUMERICAL METHODS
A major milestone paper in applying numerical technique in the form of dynamic
programming to unequally spaced arrays was done by M. Skolnik, G. Nemhauser, and J.
Sherman, III [2.13].
In this approach, the single N dimensional optimization problem is converted into a
sequence of N one-dimensional optimization problems. Thus instead of examining mN
cases for the Brute–Force approach, approximately (N-1) m2cases only need to be
examined, where N elements of the array can occupy any one of the m possible locations
Ph.D Thesis ; Rajashree J W ; 2011 Page 16
in the aperture. The design of the complete array is built up from successive designs of
partial arrays. In general, this method will not give an optimal solution for the array, since
the inter-element dependence is not fully accounted for in the procedure. Nevertheless,
the technique, even with this shortcoming provided the much needed base for the design
of one class of antennas.
Klimczak, William N [2.14] proposed deterministically thinned aperture phased antenna
array. According to this study the grating lobes are suppressed and minimized when
unequal sized elements are employed in the array. By varying the size of the radiating
elements, the positions of the elements will not be periodic and the spacing between
adjacent elements in general will not be equal. Byon Kun Chang et al [2.15] proposed
minimax-maxmini algorithm to obtain uniform side lobe in the sense of Dolph-
Chebyshev array pattern. The minimax-maxmini algorithm is solved iteratively by the
revised simplex method in linear programming. The same authors, yet in another paper
[2.16], proposed weighted least square method for optimum thinned antenna arrays. In
this paper, a weighted least square approach is proposed to find an optimum beam pattern
in a Dolph-Chebyshev sense. The square of the error between a thinned array pattern and
a desired pattern is minimized in such a way that some sections of the array visual range
were emphasized than other ones to find an optimum beam pattern. It was shown that an
exponential weighting function is appropriate for achieving a uniform side lobe. Also, it
was found that the minimum mean square error does not necessarily yield an optimum
beam pattern based on convergence characteristics.
Minimax is an optimization method which does not require derivative information; it can
be applied to nonlinear problems for which analytical solutions are not easily possible
[2.17,2.18]. The method also lends to a situation which has sure-convergence properties.
The main objective in this method is to minimize the maximum error and to maximize
the minimum error with respect to the desired result. Success in applying the method to
antenna synthesis is based on how well the residues, which represent the error between
the actual and desired pattern at the sample points are chosen and calculated. Schjer-
Jacobsen and Madsen[2.19] was successful in applying this technique to different types
of antenna synthesis problems including optimization of Dolph-Cbebyshev arrays by
Ph.D Thesis ; Rajashree J W ; 2011 Page 17
spacing variation and to find the element positions in non-uniformly spaced linear arrays
with uniform excitation that produce minimized (equal) sidelobe levels. The iterative
method for minimization is based on successive linear/ approximately linear
approximations to the nonlinear residuals.
Various programming approaches have been attempted for minimizing the maximum
error α and maximizing the minimum error β as a multi-objective linear programming
problem. The simplest approach is to minimize α and maximize β alternatively, subject to
the inequality constraints and updating the spacings at each iteration.
In Adaptive goal programming approach, the problem is converted into solving a multi-
objective linear programming problem with a fixed goal [2.20]. Since a fixed goal for the
sidelobe level is generally not easy to determine, an adaptive version of the goal
programming is taken. In this approach, the goals were set for α and β iteratively based
on the difference between α and β. If the goals for α and β are g1(k) and g2(k)
respectively, the methodology minimized the total error of α, β with respect to their goals,
where k is iteration index.
Chang et al [2.21] has compared the conventional minimax algorithm with other
approaches for a 51 element array with the array length that of 101-element equi-spaced
linear array of half-wavelength element spacing. It is shown that the minimax- algorithm
with adaptive goal programming performs best with respect to sidelobe. The maximum
sidelobe level is shown to be -23.5 dB. It is found that the sidelobe performance is much
closer to the Dolph-Chebyshev pattern than results obtained with the conventional
minimax algorithm. For array steered to 30 degrees from normal, sidelobes achieved are
about -19 dB.
2.1.4 STOCHASTIC SEARCH AND OTHER APPROACHES
Search based on stochastic approach is through probability rules, working in an “oriented
random” manner. These methods use only information from the objective function, not
requiring knowledge of its derivatives or possible discontinuities [2.22,2.23].
Ph.D Thesis ; Rajashree J W ; 2011 Page 18
These techniques gained popularity with the recent progress in computer systems, since
they require simultaneous working through a large number of solutions of the proposed
problem. That is necessary in order to let the method explore properly all regions from
the search domain containing the optimal solution. Since these techniques work with
probability rules, it is less likely that they will converge to a local minimum.
Some of the significant Stochastic algorithms useful for solving optimization problems
are Simulated annealing [2.24], Tabu search [2.25], Neural networks[2.26], Particle
Swarm Optimization [2.27], Ant Colony Optimization [2.28] and Evolutionary
algorithms.
Out of these, Ant Colony Organization (ACO), Particle Swarm Optimization (PSO) and
Evolutionary algorithms have been used for solving the antenna array thinning problem.
All the three algorithms derive their inspiration from nature and can be classified under
NBSA.
Óscar Quevedo-Teruel, et al [2.29] have applied ACO technique to successful thinning
of 100 element linear and 200 element planar arrays. In their approach they use the array
with all elements “on” as the initial value and take advantage of the entire structure in
which there is only one nest from which all ants depart. For a 100 element linear array
they obtained a Sidelobe level (SLL) of -20.52 dB with 20 elements optimally selected to
be “off”. In case of planar array the SLL obtained was -25.76 and -25.67 dB in the two
orthogonal planes.
S. Mosca and M. Ciattaglia [2.30] has applied ACO for array thinning in combination
with the classical combinatorial Knapsack problem. By this technique they achieved
sidelobe levels of -18 dB for a 30 x 30 element planar array with Taylor distribution with
a thinning factor of 0.556.
In a detailed article dealing with antenna design, Nanbo Jin and Yahya Rahmat-Samii has
analyzed various derivatives of PSO including real-number, binary, single-objective and
Multi-objective optimizations for antenna design[2.31,2.32]. Their simulation study
involved various linear arrays from 10 elements to 200 elements and they were able to
Ph.D Thesis ; Rajashree J W ; 2011 Page 19
establish a pareto-front to trade off between peak SL and antenna elements “turned on”.
Results obtained from simulation were also compared with some experimental antenna
arrays.
Using Simple Genetic Algorithm (SGA) as a tool for optimisation in electromagnetic
problems was studied initially by Rahmat-Samii, Johnson, Weile, Haupt, DH Werner and
others [2.33-2.37]. Typically for a 400 element planar array of isotropic sources, a
thinning factor of 0.4 having a sidelobe level of -22.6 db could be achieved. An alternate
way of obtaining element positions in a non-uniform array is to perturb the element
positions within a uniformly spaced array [2.38]. GA has been used to find the new
positions of a linear array with 2N +1 elements by using a side lobe level cost function.
The same principle has been extended to scanning radar arrays as well as other
applications.
Synthesis of thinned antenna arrays using Real-Coded GA [2.39-2.41] and Hybrid GA
i.e., combining both Boolean and real valued variables [2.42, 2.43] have been explored
for specific cases. When the thinned array has variable amplitude as a parameter for
element excitation, SGA with binary coding shall not be adequate. Using a binary coding
technique may not only give rise to quantization noise but also shall require time
consuming coding / decoding procedures. In order to effectively address this problem by
means of a GA-based procedure, a hybrid coding is used. The chromosome assumes this
structure
; … . . … . ; … . . … . (2.1)
where M is the number of active elements, is a Boolean value indicating the state
(turned on or off) of the kth array element, ωk is the corresponding excitation coefficient,
and N is the number of intervals in which the array length has been discretized.
According to the adopted representation, suitable genetic operators have to be defined in
order to obtain admissible solutions and possibly enhance the convergence process.
In spite of all the good features of GA in solving a multi-objective problem, a GA-based
procedure is fairly slow to “fine tune” the optimum solution after locating an appropriate
Ph.D Thesis ; Rajashree J W ; 2011 Page 20
region (called attraction basin) in the solution space. According to [2.44], SGA must be
customized for each application in order to give optimal results.
Other techniques such as Iterative Fourier technique [2.45], alternating projection
technique [2.46], Fractal technique [2.47,2.48], Matrix pencil method [2.49,2.50] have
also been attempted for thinning antenna arrays in some form. Wang, Fang, and Chow
[2.51] propose a combined approach including sub array amplitude weights, random sub
array and the random displacing rows to reduce the Grating lobes in a thinned phased
array.
2.1.5 CONCLUSIONS BASED ON THE LITERATURE REVIEW:
Following conclusions are drawn from the literature review conducted on the subject
• The topic of thinned array synthesis is a current research topic and many
researchers are working in newer areas.
• Synthesis of thinned arrays is a multi-dimensional, multi-objective problem
involving nonlinear functions. Though classical analytical methods were being
used in earlier years, better results are possible with the advent of fast computing
machines and newer techniques.
• Stochastic optimization techniques are better suited for solving this
problem.However these stochastic techniques are fairly slow in converging to
optimum results. There is also a need for fine tuning the parameters and
customize the approach for each application individually.
• Most of the studies on thinning of antenna arrays are restricted to smaller arrays,
involving a maximum of a few hundreds of elements. There appears to be a
problem in scaling the approach for handling larger arrays.
2.1.6 ADDITIONAL OBSERVATIONS ABOUT NBSA
As mentioned earlier, NBSA draw inspiration from nature’s ability to optimise. They are
not deterministic and thus form a subgroup under stochastic algorithms.
Ph.D Thesis ; Rajashree J W ; 2011 Page 21
Based on the literature study reported above, it was observed that NBSA have the
following special characteristics, which are useful for solving optimization problems
related to thinning of EAA,
• Gradient free (do not use derivative information).
• Effective for multi-objective functions.
• Robust leading to practical acceptable solution rather than the best solution.
• Largely independent of initial design/solution domain.
• Useful when dealing with problems, where the solution space is vast and unknown.
2.2 PROBLEM FOR INVESTIGATION
2.2.1 HYPOTHESES FOR INVESTIGATION Based on the above study and the conclusions drawn as above, it was proposed to take up
the following hypotheses for investigation for the present research work
1. The NBSA are well suited for tackling the optimization problem related to
thinning in antenna arrays.
2. It is possible to formulate a search algorithm by which NBSA can be used for
solving the thinning problems related to large arrays.
3. The approach can be extended so that the concept can be used for building
practical systems with large number of elements.
2.2.2 NEED TO INVESTIGATE OPTIMAL SEARCH STRATEGY
It was observed from the literature review that in spite of possessing many good features
for solving a problem of the nature of thinned array, NBSA needs to be fine tuned and
customised for an application. In fact Eiben [2.52] in his review paper observes optimal
parameters of these techniques vary not only from problem to problem but also from
stage to stage during the optimization process for a given problem. The paper sites nearly
140 references to support his view and suggests the need for optimal search strategies
based on appropriate choice of algorithm parameters for a given problem. To quote
Ph.D Thesis ; Rajashree J W ; 2011 Page 22
another researcher [2.53], “cost of determining the exact optimal GA parameters usually
exceeds the cost of solving the original problem!”
2.2.3 APPROACH FOR PRESENT INVESTIGATION
In view of this, the investigation was taken up in three phases.
Phase I: Technique for establishing Optimal Search Strategy
Phase II: Study on Scaling and Dynamic thinning
Phase III: Simulation Results and Discussions
These are discussed in the following chapters. Figure 2.1 shows the flow of the total
investigative procedure.
In Chapter 3 an optimal search strategy using NBSA is established for array thinning
problems. This is augmented with the simulated results presented for a benchmark
problem.
Chapter 4 discusses the methodology for scaling up the above strategy for large practical
systems involving few thousands of antenna elements. Search technique for exploring
large solution space and implications of thinning for dynamic cases are also discussed
here. In this connection zoning, geometric and dynamic symmetry, Bulk Array
Computation (BAC) and Thinning Algorithm based on Genetic Search (TAGS) have
been evolved as a part of the present investigation.
Corresponding simulation results are presented in Chapter 5.
Lastly Chapter 6 concludes with the summary of current research work and directions for
future work.
Ph
h.D Thesis ; R
CIn
CLiter
Chapter-
CScaling and
CResults
C
ajashree J W ;
Chapter 1ntroduction
Chapter 2rature Review
3 Optimal SeStrategy
Chapter 4d Dynamic Th
Chapter 5and Discussi
Chapter 6
2011
Figure 2
w
earch
hinning
ions
2.1 Thesis O
•Optimizatio• Introductio•Objectives a
•Thinning of•Conclusions•Hypotheses
• General M•Customizat• Observatio
•Issues conce•Resolving d•Summary o
• Study on E•Study on Im•Application
•Summary•Conclusion•Future Dire
rganization
on and NBSA on to Present iand motivatio
f EAA -A liters based on the
s for investigat
ethodology oftion of NBSA tons and conclu
erning scalingdynamic thinnof the study
Establishing semprovement ton of TAGS to l
ns ections
Techniquesinvestigative w
on for Present
rature Reviewe literature revtion and prop
f NBSA Technto benchmarkusions
g and dynamicning issues
earch Strategyo Search stratlarge EAA’s
Pa
workwork
wviewosed approach
niques k Problem
c thinning.
ytegy
age 23
h
Ph.D Thesis ; Rajashree J W ; 2011 Page 24
REFERENCES
[2.1] H. Unz, “Linear arrays with arbitrarily distributed elements,” IRE Trans. on Antenna and Propagation, Vol. AP-8, pp 222-223, March 1960
[2.2] D. D. King, R. F. Packard and R. K. Thomas, “Unequally- spaced broad-band antenna arrays,” IRE Trans. on Antenna and Propagation, Vol. AP-8, pp 380-384, July 1960
[2.3] A. Ishimaru, “ Theory of unequally-spaced arrays,” IRE Trans. On Antennas and Propagation, Vol. AP-10, pp 691-702, November 1962.
[2.4] S.S. Sandler, “Some equivalences between equally and unequally spaced arrays,” IRE Trans. on Antenna and Propagation, Vol. AP-8, pp 496-500, September 1960
[2.5] Y.T.Lo, “A space weighted antenna array” ,IRE International convention Record, Vol. 10,Pr.1,pp.191-195,1962.
[2.6] A.L. Mafett, “Array factors with non-uniform spacing parameter,” IRE Trans. on Antenna and Propagation, Vol. AP-10, pp 131-136, March 1962
[2.7] R.E. Willey, “Space tapering of linear and planar arrays,” IRE Trans. on Antenna and Propagation, Vol. AP-8, pp 369-377, July 1962
[2.8] F.C. Ogg, “Steerable array radars,” IRE Trans. on Military Electronics, Vol. MIL-5, pp80-94, April 1961
[2.9] J.L. Allen, “Beam Shaping by Omission of Elements (‘Space Tapering’ or Density Tapering’),” Phased Array Radar Studies, M.I.T. Lincoln Laboratory, Lexington, Mass., Tech. Rept. No.236; pp 299-305, November 1961
[2.10] R. F. Harrington, “Sidelobe reduction by non-uniform element spacing,” IRE Trans.On Antenna ans Propagation, Vol. AP-9, pp 187-192, March 1961.
[2.11] M. G. Andreasen, “Linear arrays with variable inter-element spacings,” IRE Trans. On Antenna and Propagation, Vol. AP-10, pp 137-143, March 1962.
[2.12] M. Skolnik, J. Sherman, III, and F. Ogg, “Statistically designed density-tapered array,” IEEE Trans. Antennas Propag., vol. 12, no. 4, pp.408–417, Jul. 1964.
[2.13] M.I. Skolnik, G. Nemhauser, and J. W. Sherman, “Dynamic Programming Applied to Unequally Spaced Arrays,” IRE Trans. on Antenna and Propagation, Vol. AP-10, pp 35-43, January 1964.
[2.14] Klimczak, William N “Deterministic thinned aperture phased antenna array”, EP 0 315 689B1.
[2.15] Byon Kun Chang, Xinyu Ma, H Brian “Minimax-maxminiAlgorithm: A new approach to optimization of the thinned antenna arrays”, IEEE Trans. Antennas and Propagation, pp no 514-517, 1994
[2.16] Xinyu Ma and Byong Kun Chang, “ Weighted least square method for optimum thinned antenna arrays,” 1997 AP-S International Symposium Digest., pp 2232-2235
[2.17] K. Madsen, H. Schjer-Jacobsen and J.Voldby “Minimax solution of non-linear equations without calculating derivatives,” Mathematical Programming Study 3rd Ed, M.Balinski and. P. Wolfe , North Holland Publishing co., Amsterdam, 1975.
[2.18] K. Madsen, H. Schjer-Jacobsen and J.Voldby “Automated minimax design of networks,” IEEE Trans. on Circuits and Systems , Vol. CAS-22, pp 791-796, October1975.
Ph.D Thesis ; Rajashree J W ; 2011 Page 25
[2.19] H. Schjer-Jacobsen and K. Madsen, “Synthesis of nonuniformy spaced arrays using a general nonlinear minimax optimization method,” IEEE Trans. on Antenna and Propagation , Vol. AP- , pp 501-506, July1976.
[2.20] S.I. Gass. “Linear Programming Methods and Applications”, fifth edition, McGraw-Hill. Inc., 1985
[2.21] Byon Kun Chang, Xinyu Ma, H Brian “Minimax-maxmini Algorithm: A new approach to optimization of the thinned antenna arrays”, IEEE Trans. Antennas and Propagation, pp no 514-517, 1994
[2.22] Z. Michalewicz and M. Michalewicz, “Evolutionary computation techniques and their applications,” in IEEE — International Conference on Intelligent Processing systems, Beijing, 1997, pp. 14–25.
[2.23] P. G. Alotto, C. Eranda, and B. Brandstaetter, “Stochastic algorithms in electromagnetic optimization,” IEEE Transactions on Magnetics, vol. 34, no. 5, pp. 3674–3684, Sept. 1998
[2.24] S. Kirkpatrick; C. D. Gelatt; M. P. Vecchi, “Optimization by Simulated Annealing”, Science, New Series, Vol ., No.4598, May 13, 1983, PP 671-680
[2.25] Glover F. “Future Paths for Integer Programming and Links to Artificial Intelligence” Computers and Operations Research 13, (1986), pp. 533-549
[2.26] J. J. Hopfield and D. W. Tank , "Neura" Computation of Decisions in Optimization Problems” , Biological Cybernetics, , 52, 141-152 (1985), Springer verlag
[2.27] J. Kennedy and R. Eberhart, “Particle swarm optimization,” in Proc. IEEE Int. Conf. Neural Networks, 1995, vol. 4, pp. 1942–1948.
[2.28] J.L. Deneubourg et al, “Probablistic behavior in ants: a strategy of errors”, J.theor.Biology, 105, 259,1983
[2.29] O. Quevedo-Teruel and E. Rajo-lglesias, “ Ant-colony optimization for array synthesis,” IEEE International symposium Antenna and Propagation Society 9-14 July2006,pp 3301-3304
[2.30] S. Mosca and M. Ciattaglia, “ Ant colony optimization to design thinned arrays,”2006 IEEE International symposium Antennas and Propagation Society pp.4675 -4678
[2.31] Nanbo Jin Rahmat-Samii, Y. “Advances in Particle Swarm Optimization for Antenna Designs: Real-Number, Binary, Single-Objective and Multiobjective Implementations” , IEEE Trans. Antennas Propag.,vol. 55, no. 3, pp.556–567, May. 2007.
[2.32] Rahmat-Sami, Y, and Nanbo Jin, “Particle Swarm Optimization (PSO) in Engineering Electromagnetics: A Nature-Inspired Evolutionary Algorithm”, International Conf on Electromagnetics in Advanced Applications, 2007 , ICEAA200717-21 Sept. 2007 pp177 – 182
[2.33] J. M. Johnson and Y. Rahmat-Samii, “Genetic Algorithms in Engineering Electromagnetics,” Antennas Propag. Mag., vol. 39, no. 4, pp.7–21, Aug. 1997
[2.34] D. S. Weile and E. Michielssen, “Genetic algorithm optimization applied to electromagnetics: A review,” IEEE Trans. Antennas Propag.,vol. 45, no. 3, pp.343–353, Mar. 1997.
[2.35] Randy L. Haupt, “ An introduction to Genetic Algorithms for electromagnetics,” IEEE Antennas Propag. Mag., vol. 37, no. 2, pp.7–15, April 1995.
[2.36] Randy L. Haupt and Douglas H . Werner “Genetic Algorithms in Electromagnetics”, John Wiley & Sons.2007
Ph.D Thesis ; Rajashree J W ; 2011 Page 26
[2.37] R L Haupt “Thinned Arrays using genetic algorithms,” IEEE Trans on Antennas and propagation, VOl 42, No 7, pp 993-999,July 1994
[2.38] A Tennant, M M Dawoud and A P Anderson, “ Array pattern nulling by element position perturbations using genetic algorithm” IEEE Electronics Letters, vol 30,no 3, pp 174-176, Feb 1994
[2.39] Mazhar Tayel, Nour Ismail and Ashraf Talaat, “Optimization for the Relative Side Lobe Level of the NXN Ultrasound Phased Array Using Genetic Algorithm to Treat Large Intestine Tumores, ” Proc. AIML 06 International Conference, 13 - 15 June 2006, Sharm El Sheikh, Egypt
[2.40] G K Mahanti, N Pathak,and P Mahanti,”Synthesis of thinned linear antenna arrays with fixed sidelobe level using Real-Coded GA”, Progress in Electromagnetics Research, PIER 75,319-328
[2.41] Keen-Keong Yan and Yilong Lu, “Sidelobe Reduction in Array-Pattern Synthesis Using Genetic Algorithm,” IEEE Trans on Antennas and propagation, Voll 45, No 7, pp 1117-1122,July 1997
[2.42] D. S. Weile and E. Michielssen, “Integer Coded Pareto Genetic Algorithm Design of Constrained Antenna Arrays,” Electron. Lett., vol. 32, pp. 1744-1745, Sept. 1996
[2.43] Massimo Donelli, Salvatore Caorsi, Francesco De Natale, Matteo Pastorino, and Andrea Massa , “Linear Antenna Synthesis with a hybrid GA, “Technical Report DIT-04-070, University of Trento, Italy, August 2004
[2.44] Golino, G,” Improved genetic algorithm for the design of the optimal antenna division in sub-arrays: a multi-objective genetic algorithm,” IEEE International Radar Conference, 9-12 May 2005 ,pp 629 – 634
[2.45] Will P. M. N. Keizer, “Linear Array Thinning Using Iterative FFT Techniques,” IEEE Trans. Antennas and Propagation, Vol 56, No. 8, pp 2757-2760,August ,2008
[2.46] O. M. Bucci, D. D’Elia, G. Mazzarella, and G. Panariello, “Antenna pattern synthesis: A new general approach,” Proc. IEEE, vol. 82, pp. 358–371, Mar. 1994.
[2.47] Y. Kim and D. L. Jaggard, “The fractal random array,” Proc. IEEE, vol.74, pp. 1278–1280, Sep. 1986.
[2.48] J. N. Bogard and D. H.Werner, “Optimization of peano-gosper fractile arrays using genetic algorithms to reduce grating lobes during scanning,” in Proc. IEEE Int. Radar Confe., Arlington, VA, May 9–12, 2005, pp. 905–909.
[2.49] V. Pierro, V. Galdi, G. Castaldi, I. M. Pinto, and L. B. Felsen, “Radiation properties of planar antenna arrays based on certain categories of aperiodic tilings,” IEEE Trans. Antennas Propag., vol. 53, no. 2, pp. 635–644, Feb. 2005.
[2.50] Yanhui Liu, Zaiping Nie, and Qing Huo Liu, “Reducing the Number of Elements in a Linear Antenna Array by the Matrix Pencil Method,” IEEE Trans. Antennas Propag., vol. 56, no.9, pp. 2955-2962, September 2008.
[2.51] Hao Wang, Da-Gang Fang, and Y. Leonard Chow, “Grating Lobe Reduction in a Phased Array of Limited Scanning, “IEEE Trans. Antennas Propag., vol. 56, no.6, pp. 1581-1586, June 2008.
[2.52] AE Eiben, R Hinerding, and Z Michaelwicz,” Parameter control in evolutionary algorithms,” IEEE Trans. Evol. Compt., vol.3, pp.124-141, July. 1999.
[2.53] J. Lis, “Parallel genetic algorithm with the dynamic control parameter,” in Proc.3rd IEEE Conf. Evol.Comput.,1996, pp.324-329.