Upload
nguyenthien
View
216
Download
0
Embed Size (px)
Citation preview
Research ArticleInvestigation on the Inversion of the AtmosphericDuct Using the Artificial Bee Colony Algorithm Based onOpposition-Based Learning
Chao Yang1 Jian-Ke Zhang1 and Li-Xin Guo2
1School of Science Xirsquoan University of Posts and Telecommunications Xirsquoan 710121 China2School of Physics and Optoelectronic Engineering Xidian University Xirsquoan 710071 China
Correspondence should be addressed to Chao Yang yang chaomail163com
Received 14 January 2016 Accepted 21 March 2016
Academic Editor Sotirios K Goudos
Copyright copy 2016 Chao Yang et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
The artificial bee colony (ABC) algorithm is a recently introduced optimization method in the research field of swarm intelligenceThis paper presents an improved ABC algorithm named as OGABC based on opposition-based learning (OBL) and global bestsearch equation to overcome the shortcomings of the slow convergence rate and sinking into local optima in the process ofinversion of atmospheric duct Taking the inversion of the surface duct using refractivity from clutter (RFC) technique as anexample to validate the performance of the proposed OGABC the inversion results are compared with those of the modifiedinvasive weed optimization (MIWO) and ABC The radar sea clutter power calculated by parabolic equation method using thesimulated and measured refractivity profile is utilized to carry out the inversion of the surface duct respectively The comparativeinvestigation results indicate that the performance of OGABC is superior to that of MIWO and ABC in terms of stability accuracyand convergence rate during the process of inversion
1 Introduction
The lower atmospheric duct commonly encountered inmarine boundary layer is an abnormal electromagnetic envi-ronment due to the sharp variations of atmospheric tempera-ture and humidity above the sea surface In the ducting envi-ronment the performance of radar system and communica-tion system can be significantly changed such as the maxi-mumoperation range creation of radar holes where the radaris practically blind and strengthened sea surface clutter [1 2]Therefore it is of great importance to infer the atmosphericduct owing to its considerable effect on the radar and com-munication system that are designed to work under standardatmospheric conditionswith a typical slope of 0118M-unitss[3]
In general the atmospheric duct is represented by themodified refractivity profile The traditional methods ofdetermining the atmospheric duct include radiosondes rock-etsondes microwave refractometers and lidar Nevertheless
the traditional measurement methods have the drawbacks ofhigh cost and containing many restrictive factors RecentlyRFC technique [4 5] has been a promising method to inferthe atmospheric duct It uses the propagation characteristicsof radar sea clutter signal to infer the modified refractiv-ity profile information of atmospheric duct And the RFCtechnique has the advantages of simple devices and easyimplementation
Inversion of atmosphere duct from the RFC techniquehas been an important research subject over the past severaldecades owing to its important applications in radar systemand communication system More attention is paid to thestudy of inversion model and optimization model in RFCtechnique The detailed procedures of RFC technique aregiven by Gerstoft et al [4] and the inversion of the rangedependent and independent atmospheric duct using RFCtechnique is implemented by genetic algorithm Karimianet al [5] published a review paper on the latest researchdevelopments and the direction of future work to be done
Hindawi Publishing CorporationInternational Journal of Antennas and PropagationVolume 2016 Article ID 2749035 10 pageshttpdxdoiorg10115520162749035
2 International Journal of Antennas and Propagation
about RFC technique Zhao et al [6 7] derive the inversiontheoretical framework of adjoint method from parabolicequation model and the feasibility of the adjoint method isvalidated by numerical simulations As is known to all esti-mation of atmosphere duct using RFC technique is an inverseproblem Taking the nonlinear relation between the forwardpropagation model and atmospheric duct parameters intoconsideration the investigation on the optimization modelswith high performance is one of the most important researchtopics in the field of inversion of the atmospheric ductfrom radar sea clutter For instance the least square supportvector machine the particle swarm optimization (PSO)the simulated annealing algorithm and the ABC algorithmhave been applied to infer the atmospheric duct using RFCtechnique [8ndash11]
The ABC algorithm [12] is one of the most recentlyproposed swarm intelligence algorithms which simulatesthe intelligent behavior of honeybee swarm In ABC theoptimization procedures are implemented by simulating theintelligent foraging behavior of a honeybee swarm to shareinformation of bees for the purpose of finding the optimalsolution Currently the ABC algorithm has been applied tothe design of antenna and electromagnetic devices [13ndash15]In addition the ABC algorithm has been used to infer theatmospheric duct from RFC technique [11] and the compar-ative study results demonstrate that the performance of ABCis superior to that of the PSO for the inversion of atmosphericduct However the ABC also has the drawbacks of easilyfalling into local optima and slower convergence rate Toovercome this issue the improvedABChas been proposed byupdating the search equation to enhance its optimization per-formance and the improved ABC is validated by benchmarkfunction [16 17] In recent years the OBL was introducedby Rahnamayan et al [18] and has been proven to be auseful strategy to enhance the accuracy and convergence rateof the optimization algorithm such as differential evolutionand PSO
In this paper the OGABC is proposed by incorporatingthe OBL strategy and global best search equation into theABC to enhance the performance of ABC in the inversion ofatmospheric duct In OGABC the OBL is used to acceleratethe convergence rate and the global best search equation isadopted to balance the local and global search ability
2 The Propagation Model andObjective Function
21 Parabolic Equation Method Considering that theparabolic equation method has the advantages of highstability and accuracy it has been extensively utilized toinvestigate the tropospheric electromagnetic wave propa-gation In rectangular coordinates the parabolic equationcan be represented as
1205972119906
1205971199112+ 21198941198960
120597119906
120597119909+ 1198962
0(1198992minus 1) 119906 = 0 (1)
where 119899 is the refractive index and 1198960is the free space wave
number
If the initial field is provided the split step Fouriersolution of parabolic equation method at different range canbe easily obtained by [19]
119906 (1199090+ Δ119909 119911)
= 119890(11989411989602)[119899
2minus1]Δ119909
119865minus1119890(119894Δ11990921198960)119901
2
119865 [119906 (1199090 119911)]
(2)
where 119865 and 119865minus1 are the Fourier transform and inverse
Fourier transform respectively 119901 is the transform variableΔ119909 is the distance interval and 119906(119909
0 119911) is the initial field It
should be pointed out that this researchmainly focuses on theinversion of atmospheric duct more detailed information onthe propagation problemwith parabolic equationmethod canbe found in [19]
22 Radar Sea Clutter Power In RFC technique the objec-tive function is described by the radar sea clutter powerat different propagation distances Taking the influence ofatmosphere condition into account the received radar seaclutter power based on radar equation can be expressed indB by [4]
119875119888(m) = minus2119871 + 120590
∘+ 10 lg (119903) + 119862 (3)
where 119871 is the propagation loss calculated by the parabolicequationmethod120590∘ is the radar cross section obtained by theGIT sea clutter model [20] 119903 is the propagation distance 119862 isa constant that includes wavelength transmitter power andantenna gain and m is the parameter vector of the atmo-spheric duct
In this paper the surface based duct is described by thefollowing four-parameter model [2]
119872(119911) = 1198720
+
1198881119911 119911 lt ℎ
1
1198881ℎ1+ 1198882(119911 minus ℎ
1) ℎ
1le 119911 le ℎ
2
1198881ℎ1+ 1198882ℎ2+ 0118 (119911 minus ℎ
1minus ℎ2) 119911 gt ℎ
2
(4)
where 1198720is the base refractivity and 119888
1and ℎ
1stand for
the slope and thickness of the base layer whereas 1198882and ℎ
2
represent the slope and thickness of the inversion layerrespectively
23 The Objective Function In the process of inversion thecommonly used least squares objective function is given by[4]
119891 (m) = eTe
e = Pobs119888
minus P119888(m) minus
= Pobs119888
minus P119888(m)
(5)
where Pobs119888
and P119888(m) stand for the observed and received sea
clutter power at different ranges and Pobs119888
and P119888(m) denote
the average power of Pobs119888
and P119888(m) respectively
International Journal of Antennas and Propagation 3
3 The Proposed OGABC Algorithm
31TheOBL TheOBL strategy can improve the convergencerate and accuracy of optimization algorithm by simultane-ously evaluating the initial solution and opposite solution forthe population initialization and for the generation jumpingThe probability theory indicates that the opposite solutioncan increase the opportunity of approaching the global bestsolution in the search process The definitions of oppositenumber and opposite solution are given by [18]
Definition 1 Let119909 isin [119897 119906] be a real number Its correspondingopposite number is defined by
= 119897 + 119906 minus 119909 (6)
Definition 2 Let 119883 = (1199091 1199092 119909
119863) be a solution in 119863-
dimensional space where 119909119894isin [119897119894 119906119894] and 119897
119894 119906119894are lower
and upper bounds of the 119894th dimension The correspondingopposite solution = (
1 2
119863) is defined by
119894= 119897119894+ 119906119894minus 119909119894
119894 = 1 119863 (7)
In this paper the inversion of atmospheric duct is aminimization problem With the help of the definition ofopposite solution the OBL in the inversion of atmosphericduct can be described by the following if 119891() le 119891(119883) thenrandom solution 119883 can be replaced with otherwise wecontinue with 119883 Additionally according to a jumping ratethe better population for the next iteration can be obtainedby the generation jumping using the current and theircorresponding opposite population Evidently the randomsolution and opposite solution are simultaneously evaluatedto select the better solution in the search process
32 The Proposed OGABC and Its Implementation StepsThe ABC is one of the most recent swarm intelligenceoptimization algorithms proposed by Karaboga under theinspiration of the intelligent foraging behavior of honeybeeswarm In ABC there are three types of honeybees employedbees onlooker bees and scoutsThe position of a food sourcestands for a possible solution of the optimization problemand the nectar amount of a food source is employed toevaluate the quality of the solutionThe number of employedbees is equal to the number of food sources and the halfof the population size The employed bees undertake theresponsibility of searching for food sources and share theeffective information with onlooker bees The onlooker beestry to make a further selection of the excellent food sourcesbased on the information provided by employed bees Ifthe quality of food source cannot be improved through apredetermined condition the corresponding food sourcebecomes a scoutThen the scout begins to randomly generatea new food source at the neighborhood of the hive
In order to enhance the performance of ABC in theinversion of atmospheric duct the OGABC is presentedby incorporating the OBL strategy and global best searchequation into ABC algorithmThemain steps of OGABC aresummarized as follows
Step 1 (opposition-based population initialization) Step 1contains the following
Step 11 Randomly produce119863-dimensional population1198830of
119873 solutions by
1198830119894119895
= 119897119895+ rand (0 1) (119906
119895minus 119897119895)
119894 = 1 2 119873 119895 = 1 2 119863
(8)
Step 12 Generate the opposite population 1198741198830of1198830by
1198741198830119894119895
= 119897119895+ 119906119895minus 1198830119894119895
(9)
Step 13 Choose the119873 best solutions from [1198830 1198741198830] accord-
ing to the fitness value to produce the initial population
Step 2 (in employed bees stage) Step 2 contains the following
Step 21 Update the position of food sources using the globalbest search (10) [16] and evaluate the quality of the newposition of food sources
V119894119895= 119909119894119895+ 120601119894119895(119909119894119895minus 119909119896119895) + 120595119894119895(119910119895minus 119909119894119895)
(119894 119896 = 1 119873 119895 = 1 119863)
(10)
where the subscripts 119896 119894 and 119895 are randomly selected andsatisfy 119896 = 119894 119910
119895is the 119895th element of the global best solution
and 120601119894119895and 120595
119894119895are uniform random number in [minus1 1] and
[0 15] respectively
Step 22 Apply the greedy selectionmechanism to choose thebetter food source between the old and new food source
Step 3 Calculate the probability of each food source accord-ing to
119901119894=
fit119894
sum119873
119896=1fit119896
(11)
where fit119894represents the fitness value of the food source 119894
computed in employed bees stage
Step 4 (in onlooker bees stage) Step 4 contains the following
Step 41 Update the position of food sources using (10)according to the probability computed in Step 3
Step 42 Apply the greedy selection mechanism again tochoose the better food source
Step 5 Memorize the best solution so far
Step 6 In scouts stage decidewhether a food source becomesa scout or not if it exists the food source is replaced by a newrandom solution
Step 7 (opposition-based generation jumping) Step 7 con-tains the following
4 International Journal of Antennas and Propagation
Opposition-based population initializationwith (8) and (9)
In employed bees stage update the positionof food sources using the global best search (10) and choose the better one
In onlooker bees stage update the position offood sources using the global best search (10) and choose the better one
Does scout exist
Terminal condition
No
End
Yes
Begin
Opposition-based generation jumpingaccording to the jumping rate
Replace the scout with a random solution
Yes
No
Figure 1 The flowchart of the proposed OGABC algorithm
Step 71 According to the jumping rate decide whetheropposition-based generation jumping appears or not if itappears the new opposite population 119874119883 of 119883 are producedby
119874119883119894119895= 119897
min119895
+ 119906max119895
minus 119883119894119895
119894 = 1 2 119873 119895 = 1 2 119863
(12)
where 119897min119895
and 119906max119895
are the minimum and maximum valueof the 119895th dimension in the current population
Step 72 Choose the119873 best solutions from [119883119874119883] accordingto the fitness value to generate the population for the nextiteration
Step 8 Repeat Step 2 to Step 7 until a terminating conditionis met
The flowchart of the proposed OGABC is shown inFigure 1
4 The Numerical Results and Discussions
In this section the inversion results are given to validatethe optimization performance of the proposed OGABC Inthe following we take the inversion of the four-parametersurface duct with RFC technique as an example to analyzethe performance of OGABC and the inversion results arecompared with those of the MIWO [21] and ABC
In fact the essence of the inversion of surface duct isto obtain its corresponding refractivity profile determined
International Journal of Antennas and Propagation 5
005 01 015 02 0250
5
10
15O
ccur
renc
eO
ccur
renc
eO
ccur
renc
eO
ccur
renc
e
005 01 015 02 0250
5
10
15
005 01 015 02 0250
5
10
15
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus220
5
10
15
0
5
10
15
0
5
10
15
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
14 16 18 20 22 24 26 28 300
5
10
15
MIWO ABC OGABC
14 16 18 20 22 24 26 28 300
5
10
15
14 16 18 20 22 24 26 28 300
5
10
15
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 2 The comparison of the histograms of the inversion results for different algorithms with the noise level of 0 dB
Table 1 The lower and upper search bounds of the parameters
Parameter Lower bound Upper bound Units1198881
00 025 M-unitsm1198882
minus35 minus10 M-unitsmℎ1
250 500 mℎ2
100 300 m
by (4) In other words the optimization problem can betranslated into the inversion of the parameters of the surfaceduct m = (119888
1 1198882 ℎ1 ℎ2) and the lower and upper bounds of
the surface duct parameters are shown in Table 1In numerical simulation the inversions are implemented
by the radar sea clutter power calculated by parabolic equa-tion method using the simulated and measured refractivityprofile respectively During the inversion the simulatedradar sea clutter power from 10Km to 50Km is regardedas the observed radar sea clutter power and the radarsystem operates at a frequency of 10GHz power of 914 dBmantenna gain of 528 dB antenna height of 7m beam widthof 07∘ 600m range bin and HH polarization In additionthe control parameters of OGABC are given as followsthe population size is 60 the number of food sources is30 the parameter limit is 25 the maximum number ofiterations is 120 and the jumping rate of OBL is 03 [18]the parameters settings for MIWO are given as follows theinitial population size is 30 the maximum population size is
40 the maximum number of iterations is 120 the nonlinearmodulation index is 3 theminimumandmaximumnumbersof seeds are 0 and 10 the initial and final value of standarddeviation are 100 and 00001 and the inversion results areobtained from 30 independent runs for each algorithm forthe simulated refractivity profile case For a fair comparisonbetween ABC and OGABC they are examined using thesame parameter settings and the settings of the radar systemremain unchanged in the process of inversions
For the simulated refractivity case the radar sea clutterpower computed by the parameters of the surface duct m =
(013 minus25 40 20) is utilized to the inversion of the surfaceduct Moreover the Gaussian noise with zero mean anddifferent standard deviations is added to the simulated radarsea clutter power to examine the stability of the algorithmsand the standard deviation is employed to represent thenoise level Also the histograms and convergence curves arepresented to analyze the accuracy and convergence rate indetail
Figures 2ndash5 give the comparison of the histograms ofthe inversions of the surface duct parameters for differentalgorithms at a specific noise level and the red lines denotethe actual parameter of surface duct It is obvious that thedistribution of inversion parameters obtained by theOGABCis more intensive than those of MIWO and ABC for differentnoise level In addition the inversion results of OGABCachieve the most occurrence at the vicinity of the actualparameter compared with those of theMIWO andABCThatis to say the proposed OGABC is the most stable algorithm
6 International Journal of Antennas and Propagation
006 008 01 012 014 016 0180
5
10
15
006 008 01 012 014 016 0180
5
10
15
006 008 01 012 014 016 0180
5
10
15
0
10
20
30
0
10
20
30
0
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
10 15 20 25 3005
101520
10 15 20 25 3005
101520
10 15 20 25 3005
101520
MIWO ABC OGABC
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22
Occ
urre
nce
Occ
urre
nce
Occ
urre
nce
Occ
urre
nce
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 3 The comparison of the histograms of the inversion results for different algorithms with the noise level of 1 dB
005 01 015 02 0250
5
10
15
Occ
urre
nce
005 01 015 02 0250
5
10
15
005 01 015 02 0250
5
10
15
0
5
10
15
Occ
urre
nce
0
5
10
15
0
5
10
15
38 40 42 44 46 480
5
10
15
Occ
urre
nce
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
10 15 20 25 3005
101520
Occ
urre
nce
MIWO ABC OGABC
10 15 20 25 3005
101520
10 15 20 25 3005
101520
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 4 The comparison of the histograms of the inversion results for different algorithms with the noise level of 2 dB
among the three algorithms It is likely due to the fact thatthe global best search equation not only enhances the globalsearch ability but also avoids falling into the local minimum
To study the convergence performance of the OGABCthe comparisons of the convergence curves of different
algorithms based on the inversion results given in Figures 2ndash5with the same noise level are demonstrated in Figure 6 It canbe seen that the convergence rate of OGABC is faster thanthat of the MIWO and ABC besides the OGABC has thesmallest meanminimum fitness at the end of the iteration for
International Journal of Antennas and Propagation 7
0 005 01 015 02 02505
101520
Occ
urre
nce
0 005 01 015 02 02505
101520
0 005 01 015 02 02505
101520
05
101520
Occ
urre
nce
05
101520
05
101520
36 38 40 42 44 4605
101520
Occ
urre
nce
36 38 40 42 44 4605
101520
36 38 40 42 44 4605
101520
15 20 25 300
5
10
Occ
urre
nce
MIWO ABC OGABC
15 20 25 300
5
10
15 20 25 300
5
10
minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 5 The comparison of the histograms of the inversion results for different algorithms with the noise level of 3 dB
different noise levelThe convergence curves indicate that theOGABC is the best algorithm among the three algorithmsThis can be attributed to the fact that the OBL can produce arelatively excellent initial population in the initialization stageand a generation jumping to form a better population in thesearch process
The accuracy of the inversion of atmospheric duct is ofcrucial importance to exactly predict the marine electromag-netic environment Hence the comparisons of the differencebetween the inverted and actual radar coverage diagramsimulated by parabolic equation method for different noiselevel are shown in Figure 7 and the peaks of the parameterdistributions [20] in Figures 2ndash5 are treated as the invertedparameters of the surface duct to simulate their correspond-ing coverage diagram Nevertheless it is hardly possible toevaluate the quality of the algorithms according to the resultspresented in Figure 7 Thus a new quantitative evaluationcriterion named Mean Absolute Error (MAE) is given by
MAE =
sum119873119909
119894=1sum119873119910
119895=1
1003816100381610038161003816PL119894 (119894 119895) minus PL119886(119894 119895)
1003816100381610038161003816
119873119909119873119910
(13)
where PL119894(119894 119895) and PL
119886(119894 119895) represent the inverted and actual
propagation loss calculated at discrete point (119894Δ119909 119895Δ119910) and119873119909and 119873
119910are the sample points along the horizontal and
vertical direction respectively It is easy to see that the smallerthe MAE the higher the accuracy of the algorithm
The MAE of the three algorithms for different noise levelobtained by (13) are presented in Table 2 and the minimum
Table 2 The comparison of the MAE for different algorithms withdifferent noise level
Noise level AlgorithmsMIWO ABC OGABC
0 dB 222 164 0091 dB 188 088 0752 dB 354 184 1053 dB 325 079 060
results are highlighted in bold It can be observed fromTable 2that the MAE of OGABC is the smallest one among the threealgorithms for different noise level namely the accuracy ofOGABC is superior to MIWO and ABC
Then in order to further test the performance ofOGABCthe radar sea clutter power generated bymeasured refractivityprofile [20] is also utilized for the inversion of the surfaceduct Figure 8 shows the comparison of the inverted profileobtained by the three algorithms with the measured profileand it is distinct that the inverted profile of OGABC is inexcellent agreement with the measured one
In addition Figure 9 presents the corresponding com-parison of the convergence curves of the MIWO ABC andOGABC It is observed that the three algorithms have thefaster convergence rate at the early stage of iterations andthe minimum fitness is hardly changed with iteration at themiddle and later stage of iterationsHowever theOGABChasthe smallest minimum fitness among the three algorithms
8 International Journal of Antennas and Propagation
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 0dB
(a)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 1dB
(b)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 2dB
(c)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000M
ean
min
imum
fitn
ess
Iterations
MIWOABCOGABC
Noise level 3dB
(d)
Figure 6 The comparison of the convergence curves of different algorithms with the same noise level
That is to say the OGABC is superior to MIWO and ABCaccording to the accuracy and convergence rate
5 Conclusion
In this paper an improved ABC algorithm called OGABCis presented by simultaneously merging the OBL strategyand global best search equation into the standard ABCalgorithm to tackle its deficiency of slow convergence rateand falling into the local best during the process of inversionof atmospheric duct Taking the inversion of the surfaceduct using RFC technique as an example the propagation
characteristics of radar sea clutter obtained from the sim-ulated and measured refractivity profile are treated as theobserved sea clutter power to examine the performance ofOGABC respectively For the simulated refractivity profilecase the Gaussian noise is added to the simulated radarsea clutter power to investigate the stability of the proposedOGABC algorithm and the histograms and the convergencecurves are used to analyze the accuracy and convergencerate Further investigation using the radar sea clutter powergenerated by themeasured refractivity profile is also involvedand the accuracy and convergence rate of the algorithmsare discussed by comparing the inverted refractivity profilewith the measured one and analyzing their convergence
International Journal of Antennas and Propagation 9
60
120
180
0 20 40 60 80 100
40
20
0
minus20
minus40
40
20
0
minus20
minus40
Hei
ght (
m)
60
120
180
Hei
ght (
m)
60
120
180
Hei
ght (
m)
Distance (km)0 20 40 60 80 100
Distance (km)0 20 40 60 80 100
Distance (km)
5
0
minus5
MIWO ABC OGABC
(a)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
40
20
0
minus20
minus40
MIWO ABC OGABC
(b)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
minus40
40
20
0
minus20
40
20
0
minus20
minus40
minus60
MIWO ABC OGABC
(c)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
50
0
minus50
MIWO ABC OGABC
(d)
Figure 7 The comparison of the difference between the inverted and actual coverage diagram with different noise level (a) 0 dB (b) 1 dB(c) 2 dB and (d) 3 dB
290 295 300 305 310 315 3200
50
100
150
200
Hei
ght (
m)
M (M-units)
MIWOABC
OGABCMeasured profile
Figure 8 The comparison of the inverted profile with the measured profile
10 International Journal of Antennas and Propagation
0 20 40 60 80 100 1204000
5000
6000
7000
8000
Min
imum
fitn
ess
Iterations
MIWOABCOGABC
Figure 9 The comparison of the convergence curves of differentalgorithms
curves In addition the inversion results are also analyzed andcompared with those of the MIWO and ABC The obtainedresults verify that the proposed OGABC algorithm outper-forms MIWO and ABC in terms of stability accuracy andconvergence rate Future work will focus on the experimentalresearch and the improvement of the inversion model
Competing Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is supported by the National Science Fund forDistinguished Young Scholars of China (no 61225002) andthe Young Scientists Fund of the National Natural ScienceFoundation of China (no 61302050)
References
[1] C Yardim P Gerstoft andW S Hodgkiss ldquoEstimation of radiorefractivity from radar clutter using BayesianMonte Carlo anal-ysisrdquo IEEE Transactions on Antennas and Propagation vol 54no 4 pp 1318ndash1327 2006
[2] C Yardim P Gerstoft andW S Hodgkiss ldquoTracking refractiv-ity from clutter using Kalman and particle filtersrdquo IEEE Trans-actions on Antennas and Propagation vol 56 no 4 pp 1058ndash1070 2008
[3] C Yardim P Gerstoft andW S Hodgkiss ldquoSensitivity analysisand performance estimation of refractivity from clutter tech-niquesrdquo Radio Science vol 44 pp 1ndash16 2009
[4] P Gerstoft L T Rogers J L Krolik andW S Hodgkiss ldquoInver-sion for refractivity parameters from radar sea clutterrdquo RadioScience vol 38 pp 1ndash22 2003
[5] A Karimian C Yardim P Gerstoft W S Hodgkiss and AE Barrios ldquoRefractivity estimation from sea clutter an invitedreviewrdquo Radio Science vol 46 pp 1ndash16 2009
[6] X-F Zhao S-X Huang and H-D Du ldquoTheoretical analysisand numerical experiments of variational adjoint approach forrefractivity estimationrdquo Radio Science vol 46 no 1 pp 1ndash122011
[7] X F Zhao and S X Huang ldquoAtmospheric duct estimationusing radar sea clutter returns by the adjoint method withregularization techniquerdquo Journal of Atmospheric and OceanicTechnology vol 31 no 6 pp 1250ndash1262 2014
[8] R Douvenot V Fabbro P Gerstoft C Bourlier and J SaillardldquoA duct mapping method using least squares support vectormachinesrdquo Radio Science vol 43 pp 1ndash12 2008
[9] B Wang Z-S Wu Z-W Zhao and H-G Wang ldquoRetrievingevaporation duct heights from radar sea clutter using particleswarm optimization (PSO) algorithmrdquo Progress In Electromag-netics Research M vol 9 pp 79ndash91 2009
[10] X-F Zhao S-X Huang J Xiang andW-L Shi ldquoRemote sens-ing of atmospheric duct parameters using simulated annealingrdquoChinese Physics B vol 20 no 9 Article ID 099201 8 pages 2011
[11] C Yang ldquoEstimation of the atmospheric duct from radarsea clutter using artificial bee colony optimization algorithmrdquoProgress in Electromagnetics Research vol 135 pp 183ndash199 2013
[12] D Karaboga and B Basturk ldquoA powerful and efficient algo-rithm for numerical function optimization artificial bee colony(ABC) algorithmrdquo Journal of Global Optimization vol 39 no 3pp 459ndash471 2007
[13] J Yang W-T Li X-W Shi L Xin and J-F Yu ldquoA hybrid ABC-DE algorithm and its application for time-modulated arrayspattern synthesisrdquo IEEE Transactions on Antennas and Pro-pagation vol 61 no 11 pp 5485ndash5495 2013
[14] X Zhang X Zhang S Y Yuen S L Ho and W N Fu ldquoAnimproved artificial bee colony algorithm for optimal design ofelectromagnetic devicesrdquo IEEE Transactions on Magnetics vol49 no 8 pp 4811ndash4816 2013
[15] S K Goudos K Siakavara A Theopoulos E E Vafiadis andJ N Sahalos ldquoApplication of Gbest-guided artificial beecolonyalgorithm to passive UHF RFID tag designrdquo InternationalJournal of Microwave and Wireless Technologies 2015
[16] G Zhu and S Kwong ldquoGbest-guided artificial bee colonyalgorithm for numerical function optimizationrdquo Applied Math-ematics and Computation vol 217 no 7 pp 3166ndash3173 2010
[17] W-F Gao S-Y Liu and L-L Huang ldquoA novel artificialbee colony algorithm based on modified search equation andorthogonal learningrdquo IEEE Transactions on Cybernetics vol 43no 3 pp 1011ndash1024 2013
[18] R S Rahnamayan H R Tizhoosh and M M A SalamaldquoOpposition-based differential evolutionrdquo IEEETransactions onEvolutionary Computation vol 12 no 1 pp 64ndash79 2008
[19] A E Barrios ldquoTerrain parabolic equation model for propaga-tion in the troposphererdquo IEEE Transactions on Antennas andPropagation vol 42 no 1 pp 90ndash98 1994
[20] A Karimian C Yardim W S Hodgkiss P Gerstoft and A EBarrios ldquoEstimation of radio refractivity using a multiple angleclutter modelrdquo Radio Science vol 47 pp 1ndash9 2012
[21] A Basak D Maity and S Das ldquoA differential invasive weedoptimization algorithm for improved global numerical opti-mizationrdquo Applied Mathematics and Computation vol 219 no12 pp 6645ndash6668 2013
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
2 International Journal of Antennas and Propagation
about RFC technique Zhao et al [6 7] derive the inversiontheoretical framework of adjoint method from parabolicequation model and the feasibility of the adjoint method isvalidated by numerical simulations As is known to all esti-mation of atmosphere duct using RFC technique is an inverseproblem Taking the nonlinear relation between the forwardpropagation model and atmospheric duct parameters intoconsideration the investigation on the optimization modelswith high performance is one of the most important researchtopics in the field of inversion of the atmospheric ductfrom radar sea clutter For instance the least square supportvector machine the particle swarm optimization (PSO)the simulated annealing algorithm and the ABC algorithmhave been applied to infer the atmospheric duct using RFCtechnique [8ndash11]
The ABC algorithm [12] is one of the most recentlyproposed swarm intelligence algorithms which simulatesthe intelligent behavior of honeybee swarm In ABC theoptimization procedures are implemented by simulating theintelligent foraging behavior of a honeybee swarm to shareinformation of bees for the purpose of finding the optimalsolution Currently the ABC algorithm has been applied tothe design of antenna and electromagnetic devices [13ndash15]In addition the ABC algorithm has been used to infer theatmospheric duct from RFC technique [11] and the compar-ative study results demonstrate that the performance of ABCis superior to that of the PSO for the inversion of atmosphericduct However the ABC also has the drawbacks of easilyfalling into local optima and slower convergence rate Toovercome this issue the improvedABChas been proposed byupdating the search equation to enhance its optimization per-formance and the improved ABC is validated by benchmarkfunction [16 17] In recent years the OBL was introducedby Rahnamayan et al [18] and has been proven to be auseful strategy to enhance the accuracy and convergence rateof the optimization algorithm such as differential evolutionand PSO
In this paper the OGABC is proposed by incorporatingthe OBL strategy and global best search equation into theABC to enhance the performance of ABC in the inversion ofatmospheric duct In OGABC the OBL is used to acceleratethe convergence rate and the global best search equation isadopted to balance the local and global search ability
2 The Propagation Model andObjective Function
21 Parabolic Equation Method Considering that theparabolic equation method has the advantages of highstability and accuracy it has been extensively utilized toinvestigate the tropospheric electromagnetic wave propa-gation In rectangular coordinates the parabolic equationcan be represented as
1205972119906
1205971199112+ 21198941198960
120597119906
120597119909+ 1198962
0(1198992minus 1) 119906 = 0 (1)
where 119899 is the refractive index and 1198960is the free space wave
number
If the initial field is provided the split step Fouriersolution of parabolic equation method at different range canbe easily obtained by [19]
119906 (1199090+ Δ119909 119911)
= 119890(11989411989602)[119899
2minus1]Δ119909
119865minus1119890(119894Δ11990921198960)119901
2
119865 [119906 (1199090 119911)]
(2)
where 119865 and 119865minus1 are the Fourier transform and inverse
Fourier transform respectively 119901 is the transform variableΔ119909 is the distance interval and 119906(119909
0 119911) is the initial field It
should be pointed out that this researchmainly focuses on theinversion of atmospheric duct more detailed information onthe propagation problemwith parabolic equationmethod canbe found in [19]
22 Radar Sea Clutter Power In RFC technique the objec-tive function is described by the radar sea clutter powerat different propagation distances Taking the influence ofatmosphere condition into account the received radar seaclutter power based on radar equation can be expressed indB by [4]
119875119888(m) = minus2119871 + 120590
∘+ 10 lg (119903) + 119862 (3)
where 119871 is the propagation loss calculated by the parabolicequationmethod120590∘ is the radar cross section obtained by theGIT sea clutter model [20] 119903 is the propagation distance 119862 isa constant that includes wavelength transmitter power andantenna gain and m is the parameter vector of the atmo-spheric duct
In this paper the surface based duct is described by thefollowing four-parameter model [2]
119872(119911) = 1198720
+
1198881119911 119911 lt ℎ
1
1198881ℎ1+ 1198882(119911 minus ℎ
1) ℎ
1le 119911 le ℎ
2
1198881ℎ1+ 1198882ℎ2+ 0118 (119911 minus ℎ
1minus ℎ2) 119911 gt ℎ
2
(4)
where 1198720is the base refractivity and 119888
1and ℎ
1stand for
the slope and thickness of the base layer whereas 1198882and ℎ
2
represent the slope and thickness of the inversion layerrespectively
23 The Objective Function In the process of inversion thecommonly used least squares objective function is given by[4]
119891 (m) = eTe
e = Pobs119888
minus P119888(m) minus
= Pobs119888
minus P119888(m)
(5)
where Pobs119888
and P119888(m) stand for the observed and received sea
clutter power at different ranges and Pobs119888
and P119888(m) denote
the average power of Pobs119888
and P119888(m) respectively
International Journal of Antennas and Propagation 3
3 The Proposed OGABC Algorithm
31TheOBL TheOBL strategy can improve the convergencerate and accuracy of optimization algorithm by simultane-ously evaluating the initial solution and opposite solution forthe population initialization and for the generation jumpingThe probability theory indicates that the opposite solutioncan increase the opportunity of approaching the global bestsolution in the search process The definitions of oppositenumber and opposite solution are given by [18]
Definition 1 Let119909 isin [119897 119906] be a real number Its correspondingopposite number is defined by
= 119897 + 119906 minus 119909 (6)
Definition 2 Let 119883 = (1199091 1199092 119909
119863) be a solution in 119863-
dimensional space where 119909119894isin [119897119894 119906119894] and 119897
119894 119906119894are lower
and upper bounds of the 119894th dimension The correspondingopposite solution = (
1 2
119863) is defined by
119894= 119897119894+ 119906119894minus 119909119894
119894 = 1 119863 (7)
In this paper the inversion of atmospheric duct is aminimization problem With the help of the definition ofopposite solution the OBL in the inversion of atmosphericduct can be described by the following if 119891() le 119891(119883) thenrandom solution 119883 can be replaced with otherwise wecontinue with 119883 Additionally according to a jumping ratethe better population for the next iteration can be obtainedby the generation jumping using the current and theircorresponding opposite population Evidently the randomsolution and opposite solution are simultaneously evaluatedto select the better solution in the search process
32 The Proposed OGABC and Its Implementation StepsThe ABC is one of the most recent swarm intelligenceoptimization algorithms proposed by Karaboga under theinspiration of the intelligent foraging behavior of honeybeeswarm In ABC there are three types of honeybees employedbees onlooker bees and scoutsThe position of a food sourcestands for a possible solution of the optimization problemand the nectar amount of a food source is employed toevaluate the quality of the solutionThe number of employedbees is equal to the number of food sources and the halfof the population size The employed bees undertake theresponsibility of searching for food sources and share theeffective information with onlooker bees The onlooker beestry to make a further selection of the excellent food sourcesbased on the information provided by employed bees Ifthe quality of food source cannot be improved through apredetermined condition the corresponding food sourcebecomes a scoutThen the scout begins to randomly generatea new food source at the neighborhood of the hive
In order to enhance the performance of ABC in theinversion of atmospheric duct the OGABC is presentedby incorporating the OBL strategy and global best searchequation into ABC algorithmThemain steps of OGABC aresummarized as follows
Step 1 (opposition-based population initialization) Step 1contains the following
Step 11 Randomly produce119863-dimensional population1198830of
119873 solutions by
1198830119894119895
= 119897119895+ rand (0 1) (119906
119895minus 119897119895)
119894 = 1 2 119873 119895 = 1 2 119863
(8)
Step 12 Generate the opposite population 1198741198830of1198830by
1198741198830119894119895
= 119897119895+ 119906119895minus 1198830119894119895
(9)
Step 13 Choose the119873 best solutions from [1198830 1198741198830] accord-
ing to the fitness value to produce the initial population
Step 2 (in employed bees stage) Step 2 contains the following
Step 21 Update the position of food sources using the globalbest search (10) [16] and evaluate the quality of the newposition of food sources
V119894119895= 119909119894119895+ 120601119894119895(119909119894119895minus 119909119896119895) + 120595119894119895(119910119895minus 119909119894119895)
(119894 119896 = 1 119873 119895 = 1 119863)
(10)
where the subscripts 119896 119894 and 119895 are randomly selected andsatisfy 119896 = 119894 119910
119895is the 119895th element of the global best solution
and 120601119894119895and 120595
119894119895are uniform random number in [minus1 1] and
[0 15] respectively
Step 22 Apply the greedy selectionmechanism to choose thebetter food source between the old and new food source
Step 3 Calculate the probability of each food source accord-ing to
119901119894=
fit119894
sum119873
119896=1fit119896
(11)
where fit119894represents the fitness value of the food source 119894
computed in employed bees stage
Step 4 (in onlooker bees stage) Step 4 contains the following
Step 41 Update the position of food sources using (10)according to the probability computed in Step 3
Step 42 Apply the greedy selection mechanism again tochoose the better food source
Step 5 Memorize the best solution so far
Step 6 In scouts stage decidewhether a food source becomesa scout or not if it exists the food source is replaced by a newrandom solution
Step 7 (opposition-based generation jumping) Step 7 con-tains the following
4 International Journal of Antennas and Propagation
Opposition-based population initializationwith (8) and (9)
In employed bees stage update the positionof food sources using the global best search (10) and choose the better one
In onlooker bees stage update the position offood sources using the global best search (10) and choose the better one
Does scout exist
Terminal condition
No
End
Yes
Begin
Opposition-based generation jumpingaccording to the jumping rate
Replace the scout with a random solution
Yes
No
Figure 1 The flowchart of the proposed OGABC algorithm
Step 71 According to the jumping rate decide whetheropposition-based generation jumping appears or not if itappears the new opposite population 119874119883 of 119883 are producedby
119874119883119894119895= 119897
min119895
+ 119906max119895
minus 119883119894119895
119894 = 1 2 119873 119895 = 1 2 119863
(12)
where 119897min119895
and 119906max119895
are the minimum and maximum valueof the 119895th dimension in the current population
Step 72 Choose the119873 best solutions from [119883119874119883] accordingto the fitness value to generate the population for the nextiteration
Step 8 Repeat Step 2 to Step 7 until a terminating conditionis met
The flowchart of the proposed OGABC is shown inFigure 1
4 The Numerical Results and Discussions
In this section the inversion results are given to validatethe optimization performance of the proposed OGABC Inthe following we take the inversion of the four-parametersurface duct with RFC technique as an example to analyzethe performance of OGABC and the inversion results arecompared with those of the MIWO [21] and ABC
In fact the essence of the inversion of surface duct isto obtain its corresponding refractivity profile determined
International Journal of Antennas and Propagation 5
005 01 015 02 0250
5
10
15O
ccur
renc
eO
ccur
renc
eO
ccur
renc
eO
ccur
renc
e
005 01 015 02 0250
5
10
15
005 01 015 02 0250
5
10
15
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus220
5
10
15
0
5
10
15
0
5
10
15
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
14 16 18 20 22 24 26 28 300
5
10
15
MIWO ABC OGABC
14 16 18 20 22 24 26 28 300
5
10
15
14 16 18 20 22 24 26 28 300
5
10
15
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 2 The comparison of the histograms of the inversion results for different algorithms with the noise level of 0 dB
Table 1 The lower and upper search bounds of the parameters
Parameter Lower bound Upper bound Units1198881
00 025 M-unitsm1198882
minus35 minus10 M-unitsmℎ1
250 500 mℎ2
100 300 m
by (4) In other words the optimization problem can betranslated into the inversion of the parameters of the surfaceduct m = (119888
1 1198882 ℎ1 ℎ2) and the lower and upper bounds of
the surface duct parameters are shown in Table 1In numerical simulation the inversions are implemented
by the radar sea clutter power calculated by parabolic equa-tion method using the simulated and measured refractivityprofile respectively During the inversion the simulatedradar sea clutter power from 10Km to 50Km is regardedas the observed radar sea clutter power and the radarsystem operates at a frequency of 10GHz power of 914 dBmantenna gain of 528 dB antenna height of 7m beam widthof 07∘ 600m range bin and HH polarization In additionthe control parameters of OGABC are given as followsthe population size is 60 the number of food sources is30 the parameter limit is 25 the maximum number ofiterations is 120 and the jumping rate of OBL is 03 [18]the parameters settings for MIWO are given as follows theinitial population size is 30 the maximum population size is
40 the maximum number of iterations is 120 the nonlinearmodulation index is 3 theminimumandmaximumnumbersof seeds are 0 and 10 the initial and final value of standarddeviation are 100 and 00001 and the inversion results areobtained from 30 independent runs for each algorithm forthe simulated refractivity profile case For a fair comparisonbetween ABC and OGABC they are examined using thesame parameter settings and the settings of the radar systemremain unchanged in the process of inversions
For the simulated refractivity case the radar sea clutterpower computed by the parameters of the surface duct m =
(013 minus25 40 20) is utilized to the inversion of the surfaceduct Moreover the Gaussian noise with zero mean anddifferent standard deviations is added to the simulated radarsea clutter power to examine the stability of the algorithmsand the standard deviation is employed to represent thenoise level Also the histograms and convergence curves arepresented to analyze the accuracy and convergence rate indetail
Figures 2ndash5 give the comparison of the histograms ofthe inversions of the surface duct parameters for differentalgorithms at a specific noise level and the red lines denotethe actual parameter of surface duct It is obvious that thedistribution of inversion parameters obtained by theOGABCis more intensive than those of MIWO and ABC for differentnoise level In addition the inversion results of OGABCachieve the most occurrence at the vicinity of the actualparameter compared with those of theMIWO andABCThatis to say the proposed OGABC is the most stable algorithm
6 International Journal of Antennas and Propagation
006 008 01 012 014 016 0180
5
10
15
006 008 01 012 014 016 0180
5
10
15
006 008 01 012 014 016 0180
5
10
15
0
10
20
30
0
10
20
30
0
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
10 15 20 25 3005
101520
10 15 20 25 3005
101520
10 15 20 25 3005
101520
MIWO ABC OGABC
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22
Occ
urre
nce
Occ
urre
nce
Occ
urre
nce
Occ
urre
nce
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 3 The comparison of the histograms of the inversion results for different algorithms with the noise level of 1 dB
005 01 015 02 0250
5
10
15
Occ
urre
nce
005 01 015 02 0250
5
10
15
005 01 015 02 0250
5
10
15
0
5
10
15
Occ
urre
nce
0
5
10
15
0
5
10
15
38 40 42 44 46 480
5
10
15
Occ
urre
nce
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
10 15 20 25 3005
101520
Occ
urre
nce
MIWO ABC OGABC
10 15 20 25 3005
101520
10 15 20 25 3005
101520
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 4 The comparison of the histograms of the inversion results for different algorithms with the noise level of 2 dB
among the three algorithms It is likely due to the fact thatthe global best search equation not only enhances the globalsearch ability but also avoids falling into the local minimum
To study the convergence performance of the OGABCthe comparisons of the convergence curves of different
algorithms based on the inversion results given in Figures 2ndash5with the same noise level are demonstrated in Figure 6 It canbe seen that the convergence rate of OGABC is faster thanthat of the MIWO and ABC besides the OGABC has thesmallest meanminimum fitness at the end of the iteration for
International Journal of Antennas and Propagation 7
0 005 01 015 02 02505
101520
Occ
urre
nce
0 005 01 015 02 02505
101520
0 005 01 015 02 02505
101520
05
101520
Occ
urre
nce
05
101520
05
101520
36 38 40 42 44 4605
101520
Occ
urre
nce
36 38 40 42 44 4605
101520
36 38 40 42 44 4605
101520
15 20 25 300
5
10
Occ
urre
nce
MIWO ABC OGABC
15 20 25 300
5
10
15 20 25 300
5
10
minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 5 The comparison of the histograms of the inversion results for different algorithms with the noise level of 3 dB
different noise levelThe convergence curves indicate that theOGABC is the best algorithm among the three algorithmsThis can be attributed to the fact that the OBL can produce arelatively excellent initial population in the initialization stageand a generation jumping to form a better population in thesearch process
The accuracy of the inversion of atmospheric duct is ofcrucial importance to exactly predict the marine electromag-netic environment Hence the comparisons of the differencebetween the inverted and actual radar coverage diagramsimulated by parabolic equation method for different noiselevel are shown in Figure 7 and the peaks of the parameterdistributions [20] in Figures 2ndash5 are treated as the invertedparameters of the surface duct to simulate their correspond-ing coverage diagram Nevertheless it is hardly possible toevaluate the quality of the algorithms according to the resultspresented in Figure 7 Thus a new quantitative evaluationcriterion named Mean Absolute Error (MAE) is given by
MAE =
sum119873119909
119894=1sum119873119910
119895=1
1003816100381610038161003816PL119894 (119894 119895) minus PL119886(119894 119895)
1003816100381610038161003816
119873119909119873119910
(13)
where PL119894(119894 119895) and PL
119886(119894 119895) represent the inverted and actual
propagation loss calculated at discrete point (119894Δ119909 119895Δ119910) and119873119909and 119873
119910are the sample points along the horizontal and
vertical direction respectively It is easy to see that the smallerthe MAE the higher the accuracy of the algorithm
The MAE of the three algorithms for different noise levelobtained by (13) are presented in Table 2 and the minimum
Table 2 The comparison of the MAE for different algorithms withdifferent noise level
Noise level AlgorithmsMIWO ABC OGABC
0 dB 222 164 0091 dB 188 088 0752 dB 354 184 1053 dB 325 079 060
results are highlighted in bold It can be observed fromTable 2that the MAE of OGABC is the smallest one among the threealgorithms for different noise level namely the accuracy ofOGABC is superior to MIWO and ABC
Then in order to further test the performance ofOGABCthe radar sea clutter power generated bymeasured refractivityprofile [20] is also utilized for the inversion of the surfaceduct Figure 8 shows the comparison of the inverted profileobtained by the three algorithms with the measured profileand it is distinct that the inverted profile of OGABC is inexcellent agreement with the measured one
In addition Figure 9 presents the corresponding com-parison of the convergence curves of the MIWO ABC andOGABC It is observed that the three algorithms have thefaster convergence rate at the early stage of iterations andthe minimum fitness is hardly changed with iteration at themiddle and later stage of iterationsHowever theOGABChasthe smallest minimum fitness among the three algorithms
8 International Journal of Antennas and Propagation
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 0dB
(a)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 1dB
(b)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 2dB
(c)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000M
ean
min
imum
fitn
ess
Iterations
MIWOABCOGABC
Noise level 3dB
(d)
Figure 6 The comparison of the convergence curves of different algorithms with the same noise level
That is to say the OGABC is superior to MIWO and ABCaccording to the accuracy and convergence rate
5 Conclusion
In this paper an improved ABC algorithm called OGABCis presented by simultaneously merging the OBL strategyand global best search equation into the standard ABCalgorithm to tackle its deficiency of slow convergence rateand falling into the local best during the process of inversionof atmospheric duct Taking the inversion of the surfaceduct using RFC technique as an example the propagation
characteristics of radar sea clutter obtained from the sim-ulated and measured refractivity profile are treated as theobserved sea clutter power to examine the performance ofOGABC respectively For the simulated refractivity profilecase the Gaussian noise is added to the simulated radarsea clutter power to investigate the stability of the proposedOGABC algorithm and the histograms and the convergencecurves are used to analyze the accuracy and convergencerate Further investigation using the radar sea clutter powergenerated by themeasured refractivity profile is also involvedand the accuracy and convergence rate of the algorithmsare discussed by comparing the inverted refractivity profilewith the measured one and analyzing their convergence
International Journal of Antennas and Propagation 9
60
120
180
0 20 40 60 80 100
40
20
0
minus20
minus40
40
20
0
minus20
minus40
Hei
ght (
m)
60
120
180
Hei
ght (
m)
60
120
180
Hei
ght (
m)
Distance (km)0 20 40 60 80 100
Distance (km)0 20 40 60 80 100
Distance (km)
5
0
minus5
MIWO ABC OGABC
(a)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
40
20
0
minus20
minus40
MIWO ABC OGABC
(b)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
minus40
40
20
0
minus20
40
20
0
minus20
minus40
minus60
MIWO ABC OGABC
(c)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
50
0
minus50
MIWO ABC OGABC
(d)
Figure 7 The comparison of the difference between the inverted and actual coverage diagram with different noise level (a) 0 dB (b) 1 dB(c) 2 dB and (d) 3 dB
290 295 300 305 310 315 3200
50
100
150
200
Hei
ght (
m)
M (M-units)
MIWOABC
OGABCMeasured profile
Figure 8 The comparison of the inverted profile with the measured profile
10 International Journal of Antennas and Propagation
0 20 40 60 80 100 1204000
5000
6000
7000
8000
Min
imum
fitn
ess
Iterations
MIWOABCOGABC
Figure 9 The comparison of the convergence curves of differentalgorithms
curves In addition the inversion results are also analyzed andcompared with those of the MIWO and ABC The obtainedresults verify that the proposed OGABC algorithm outper-forms MIWO and ABC in terms of stability accuracy andconvergence rate Future work will focus on the experimentalresearch and the improvement of the inversion model
Competing Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is supported by the National Science Fund forDistinguished Young Scholars of China (no 61225002) andthe Young Scientists Fund of the National Natural ScienceFoundation of China (no 61302050)
References
[1] C Yardim P Gerstoft andW S Hodgkiss ldquoEstimation of radiorefractivity from radar clutter using BayesianMonte Carlo anal-ysisrdquo IEEE Transactions on Antennas and Propagation vol 54no 4 pp 1318ndash1327 2006
[2] C Yardim P Gerstoft andW S Hodgkiss ldquoTracking refractiv-ity from clutter using Kalman and particle filtersrdquo IEEE Trans-actions on Antennas and Propagation vol 56 no 4 pp 1058ndash1070 2008
[3] C Yardim P Gerstoft andW S Hodgkiss ldquoSensitivity analysisand performance estimation of refractivity from clutter tech-niquesrdquo Radio Science vol 44 pp 1ndash16 2009
[4] P Gerstoft L T Rogers J L Krolik andW S Hodgkiss ldquoInver-sion for refractivity parameters from radar sea clutterrdquo RadioScience vol 38 pp 1ndash22 2003
[5] A Karimian C Yardim P Gerstoft W S Hodgkiss and AE Barrios ldquoRefractivity estimation from sea clutter an invitedreviewrdquo Radio Science vol 46 pp 1ndash16 2009
[6] X-F Zhao S-X Huang and H-D Du ldquoTheoretical analysisand numerical experiments of variational adjoint approach forrefractivity estimationrdquo Radio Science vol 46 no 1 pp 1ndash122011
[7] X F Zhao and S X Huang ldquoAtmospheric duct estimationusing radar sea clutter returns by the adjoint method withregularization techniquerdquo Journal of Atmospheric and OceanicTechnology vol 31 no 6 pp 1250ndash1262 2014
[8] R Douvenot V Fabbro P Gerstoft C Bourlier and J SaillardldquoA duct mapping method using least squares support vectormachinesrdquo Radio Science vol 43 pp 1ndash12 2008
[9] B Wang Z-S Wu Z-W Zhao and H-G Wang ldquoRetrievingevaporation duct heights from radar sea clutter using particleswarm optimization (PSO) algorithmrdquo Progress In Electromag-netics Research M vol 9 pp 79ndash91 2009
[10] X-F Zhao S-X Huang J Xiang andW-L Shi ldquoRemote sens-ing of atmospheric duct parameters using simulated annealingrdquoChinese Physics B vol 20 no 9 Article ID 099201 8 pages 2011
[11] C Yang ldquoEstimation of the atmospheric duct from radarsea clutter using artificial bee colony optimization algorithmrdquoProgress in Electromagnetics Research vol 135 pp 183ndash199 2013
[12] D Karaboga and B Basturk ldquoA powerful and efficient algo-rithm for numerical function optimization artificial bee colony(ABC) algorithmrdquo Journal of Global Optimization vol 39 no 3pp 459ndash471 2007
[13] J Yang W-T Li X-W Shi L Xin and J-F Yu ldquoA hybrid ABC-DE algorithm and its application for time-modulated arrayspattern synthesisrdquo IEEE Transactions on Antennas and Pro-pagation vol 61 no 11 pp 5485ndash5495 2013
[14] X Zhang X Zhang S Y Yuen S L Ho and W N Fu ldquoAnimproved artificial bee colony algorithm for optimal design ofelectromagnetic devicesrdquo IEEE Transactions on Magnetics vol49 no 8 pp 4811ndash4816 2013
[15] S K Goudos K Siakavara A Theopoulos E E Vafiadis andJ N Sahalos ldquoApplication of Gbest-guided artificial beecolonyalgorithm to passive UHF RFID tag designrdquo InternationalJournal of Microwave and Wireless Technologies 2015
[16] G Zhu and S Kwong ldquoGbest-guided artificial bee colonyalgorithm for numerical function optimizationrdquo Applied Math-ematics and Computation vol 217 no 7 pp 3166ndash3173 2010
[17] W-F Gao S-Y Liu and L-L Huang ldquoA novel artificialbee colony algorithm based on modified search equation andorthogonal learningrdquo IEEE Transactions on Cybernetics vol 43no 3 pp 1011ndash1024 2013
[18] R S Rahnamayan H R Tizhoosh and M M A SalamaldquoOpposition-based differential evolutionrdquo IEEETransactions onEvolutionary Computation vol 12 no 1 pp 64ndash79 2008
[19] A E Barrios ldquoTerrain parabolic equation model for propaga-tion in the troposphererdquo IEEE Transactions on Antennas andPropagation vol 42 no 1 pp 90ndash98 1994
[20] A Karimian C Yardim W S Hodgkiss P Gerstoft and A EBarrios ldquoEstimation of radio refractivity using a multiple angleclutter modelrdquo Radio Science vol 47 pp 1ndash9 2012
[21] A Basak D Maity and S Das ldquoA differential invasive weedoptimization algorithm for improved global numerical opti-mizationrdquo Applied Mathematics and Computation vol 219 no12 pp 6645ndash6668 2013
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of Antennas and Propagation 3
3 The Proposed OGABC Algorithm
31TheOBL TheOBL strategy can improve the convergencerate and accuracy of optimization algorithm by simultane-ously evaluating the initial solution and opposite solution forthe population initialization and for the generation jumpingThe probability theory indicates that the opposite solutioncan increase the opportunity of approaching the global bestsolution in the search process The definitions of oppositenumber and opposite solution are given by [18]
Definition 1 Let119909 isin [119897 119906] be a real number Its correspondingopposite number is defined by
= 119897 + 119906 minus 119909 (6)
Definition 2 Let 119883 = (1199091 1199092 119909
119863) be a solution in 119863-
dimensional space where 119909119894isin [119897119894 119906119894] and 119897
119894 119906119894are lower
and upper bounds of the 119894th dimension The correspondingopposite solution = (
1 2
119863) is defined by
119894= 119897119894+ 119906119894minus 119909119894
119894 = 1 119863 (7)
In this paper the inversion of atmospheric duct is aminimization problem With the help of the definition ofopposite solution the OBL in the inversion of atmosphericduct can be described by the following if 119891() le 119891(119883) thenrandom solution 119883 can be replaced with otherwise wecontinue with 119883 Additionally according to a jumping ratethe better population for the next iteration can be obtainedby the generation jumping using the current and theircorresponding opposite population Evidently the randomsolution and opposite solution are simultaneously evaluatedto select the better solution in the search process
32 The Proposed OGABC and Its Implementation StepsThe ABC is one of the most recent swarm intelligenceoptimization algorithms proposed by Karaboga under theinspiration of the intelligent foraging behavior of honeybeeswarm In ABC there are three types of honeybees employedbees onlooker bees and scoutsThe position of a food sourcestands for a possible solution of the optimization problemand the nectar amount of a food source is employed toevaluate the quality of the solutionThe number of employedbees is equal to the number of food sources and the halfof the population size The employed bees undertake theresponsibility of searching for food sources and share theeffective information with onlooker bees The onlooker beestry to make a further selection of the excellent food sourcesbased on the information provided by employed bees Ifthe quality of food source cannot be improved through apredetermined condition the corresponding food sourcebecomes a scoutThen the scout begins to randomly generatea new food source at the neighborhood of the hive
In order to enhance the performance of ABC in theinversion of atmospheric duct the OGABC is presentedby incorporating the OBL strategy and global best searchequation into ABC algorithmThemain steps of OGABC aresummarized as follows
Step 1 (opposition-based population initialization) Step 1contains the following
Step 11 Randomly produce119863-dimensional population1198830of
119873 solutions by
1198830119894119895
= 119897119895+ rand (0 1) (119906
119895minus 119897119895)
119894 = 1 2 119873 119895 = 1 2 119863
(8)
Step 12 Generate the opposite population 1198741198830of1198830by
1198741198830119894119895
= 119897119895+ 119906119895minus 1198830119894119895
(9)
Step 13 Choose the119873 best solutions from [1198830 1198741198830] accord-
ing to the fitness value to produce the initial population
Step 2 (in employed bees stage) Step 2 contains the following
Step 21 Update the position of food sources using the globalbest search (10) [16] and evaluate the quality of the newposition of food sources
V119894119895= 119909119894119895+ 120601119894119895(119909119894119895minus 119909119896119895) + 120595119894119895(119910119895minus 119909119894119895)
(119894 119896 = 1 119873 119895 = 1 119863)
(10)
where the subscripts 119896 119894 and 119895 are randomly selected andsatisfy 119896 = 119894 119910
119895is the 119895th element of the global best solution
and 120601119894119895and 120595
119894119895are uniform random number in [minus1 1] and
[0 15] respectively
Step 22 Apply the greedy selectionmechanism to choose thebetter food source between the old and new food source
Step 3 Calculate the probability of each food source accord-ing to
119901119894=
fit119894
sum119873
119896=1fit119896
(11)
where fit119894represents the fitness value of the food source 119894
computed in employed bees stage
Step 4 (in onlooker bees stage) Step 4 contains the following
Step 41 Update the position of food sources using (10)according to the probability computed in Step 3
Step 42 Apply the greedy selection mechanism again tochoose the better food source
Step 5 Memorize the best solution so far
Step 6 In scouts stage decidewhether a food source becomesa scout or not if it exists the food source is replaced by a newrandom solution
Step 7 (opposition-based generation jumping) Step 7 con-tains the following
4 International Journal of Antennas and Propagation
Opposition-based population initializationwith (8) and (9)
In employed bees stage update the positionof food sources using the global best search (10) and choose the better one
In onlooker bees stage update the position offood sources using the global best search (10) and choose the better one
Does scout exist
Terminal condition
No
End
Yes
Begin
Opposition-based generation jumpingaccording to the jumping rate
Replace the scout with a random solution
Yes
No
Figure 1 The flowchart of the proposed OGABC algorithm
Step 71 According to the jumping rate decide whetheropposition-based generation jumping appears or not if itappears the new opposite population 119874119883 of 119883 are producedby
119874119883119894119895= 119897
min119895
+ 119906max119895
minus 119883119894119895
119894 = 1 2 119873 119895 = 1 2 119863
(12)
where 119897min119895
and 119906max119895
are the minimum and maximum valueof the 119895th dimension in the current population
Step 72 Choose the119873 best solutions from [119883119874119883] accordingto the fitness value to generate the population for the nextiteration
Step 8 Repeat Step 2 to Step 7 until a terminating conditionis met
The flowchart of the proposed OGABC is shown inFigure 1
4 The Numerical Results and Discussions
In this section the inversion results are given to validatethe optimization performance of the proposed OGABC Inthe following we take the inversion of the four-parametersurface duct with RFC technique as an example to analyzethe performance of OGABC and the inversion results arecompared with those of the MIWO [21] and ABC
In fact the essence of the inversion of surface duct isto obtain its corresponding refractivity profile determined
International Journal of Antennas and Propagation 5
005 01 015 02 0250
5
10
15O
ccur
renc
eO
ccur
renc
eO
ccur
renc
eO
ccur
renc
e
005 01 015 02 0250
5
10
15
005 01 015 02 0250
5
10
15
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus220
5
10
15
0
5
10
15
0
5
10
15
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
14 16 18 20 22 24 26 28 300
5
10
15
MIWO ABC OGABC
14 16 18 20 22 24 26 28 300
5
10
15
14 16 18 20 22 24 26 28 300
5
10
15
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 2 The comparison of the histograms of the inversion results for different algorithms with the noise level of 0 dB
Table 1 The lower and upper search bounds of the parameters
Parameter Lower bound Upper bound Units1198881
00 025 M-unitsm1198882
minus35 minus10 M-unitsmℎ1
250 500 mℎ2
100 300 m
by (4) In other words the optimization problem can betranslated into the inversion of the parameters of the surfaceduct m = (119888
1 1198882 ℎ1 ℎ2) and the lower and upper bounds of
the surface duct parameters are shown in Table 1In numerical simulation the inversions are implemented
by the radar sea clutter power calculated by parabolic equa-tion method using the simulated and measured refractivityprofile respectively During the inversion the simulatedradar sea clutter power from 10Km to 50Km is regardedas the observed radar sea clutter power and the radarsystem operates at a frequency of 10GHz power of 914 dBmantenna gain of 528 dB antenna height of 7m beam widthof 07∘ 600m range bin and HH polarization In additionthe control parameters of OGABC are given as followsthe population size is 60 the number of food sources is30 the parameter limit is 25 the maximum number ofiterations is 120 and the jumping rate of OBL is 03 [18]the parameters settings for MIWO are given as follows theinitial population size is 30 the maximum population size is
40 the maximum number of iterations is 120 the nonlinearmodulation index is 3 theminimumandmaximumnumbersof seeds are 0 and 10 the initial and final value of standarddeviation are 100 and 00001 and the inversion results areobtained from 30 independent runs for each algorithm forthe simulated refractivity profile case For a fair comparisonbetween ABC and OGABC they are examined using thesame parameter settings and the settings of the radar systemremain unchanged in the process of inversions
For the simulated refractivity case the radar sea clutterpower computed by the parameters of the surface duct m =
(013 minus25 40 20) is utilized to the inversion of the surfaceduct Moreover the Gaussian noise with zero mean anddifferent standard deviations is added to the simulated radarsea clutter power to examine the stability of the algorithmsand the standard deviation is employed to represent thenoise level Also the histograms and convergence curves arepresented to analyze the accuracy and convergence rate indetail
Figures 2ndash5 give the comparison of the histograms ofthe inversions of the surface duct parameters for differentalgorithms at a specific noise level and the red lines denotethe actual parameter of surface duct It is obvious that thedistribution of inversion parameters obtained by theOGABCis more intensive than those of MIWO and ABC for differentnoise level In addition the inversion results of OGABCachieve the most occurrence at the vicinity of the actualparameter compared with those of theMIWO andABCThatis to say the proposed OGABC is the most stable algorithm
6 International Journal of Antennas and Propagation
006 008 01 012 014 016 0180
5
10
15
006 008 01 012 014 016 0180
5
10
15
006 008 01 012 014 016 0180
5
10
15
0
10
20
30
0
10
20
30
0
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
10 15 20 25 3005
101520
10 15 20 25 3005
101520
10 15 20 25 3005
101520
MIWO ABC OGABC
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22
Occ
urre
nce
Occ
urre
nce
Occ
urre
nce
Occ
urre
nce
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 3 The comparison of the histograms of the inversion results for different algorithms with the noise level of 1 dB
005 01 015 02 0250
5
10
15
Occ
urre
nce
005 01 015 02 0250
5
10
15
005 01 015 02 0250
5
10
15
0
5
10
15
Occ
urre
nce
0
5
10
15
0
5
10
15
38 40 42 44 46 480
5
10
15
Occ
urre
nce
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
10 15 20 25 3005
101520
Occ
urre
nce
MIWO ABC OGABC
10 15 20 25 3005
101520
10 15 20 25 3005
101520
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 4 The comparison of the histograms of the inversion results for different algorithms with the noise level of 2 dB
among the three algorithms It is likely due to the fact thatthe global best search equation not only enhances the globalsearch ability but also avoids falling into the local minimum
To study the convergence performance of the OGABCthe comparisons of the convergence curves of different
algorithms based on the inversion results given in Figures 2ndash5with the same noise level are demonstrated in Figure 6 It canbe seen that the convergence rate of OGABC is faster thanthat of the MIWO and ABC besides the OGABC has thesmallest meanminimum fitness at the end of the iteration for
International Journal of Antennas and Propagation 7
0 005 01 015 02 02505
101520
Occ
urre
nce
0 005 01 015 02 02505
101520
0 005 01 015 02 02505
101520
05
101520
Occ
urre
nce
05
101520
05
101520
36 38 40 42 44 4605
101520
Occ
urre
nce
36 38 40 42 44 4605
101520
36 38 40 42 44 4605
101520
15 20 25 300
5
10
Occ
urre
nce
MIWO ABC OGABC
15 20 25 300
5
10
15 20 25 300
5
10
minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 5 The comparison of the histograms of the inversion results for different algorithms with the noise level of 3 dB
different noise levelThe convergence curves indicate that theOGABC is the best algorithm among the three algorithmsThis can be attributed to the fact that the OBL can produce arelatively excellent initial population in the initialization stageand a generation jumping to form a better population in thesearch process
The accuracy of the inversion of atmospheric duct is ofcrucial importance to exactly predict the marine electromag-netic environment Hence the comparisons of the differencebetween the inverted and actual radar coverage diagramsimulated by parabolic equation method for different noiselevel are shown in Figure 7 and the peaks of the parameterdistributions [20] in Figures 2ndash5 are treated as the invertedparameters of the surface duct to simulate their correspond-ing coverage diagram Nevertheless it is hardly possible toevaluate the quality of the algorithms according to the resultspresented in Figure 7 Thus a new quantitative evaluationcriterion named Mean Absolute Error (MAE) is given by
MAE =
sum119873119909
119894=1sum119873119910
119895=1
1003816100381610038161003816PL119894 (119894 119895) minus PL119886(119894 119895)
1003816100381610038161003816
119873119909119873119910
(13)
where PL119894(119894 119895) and PL
119886(119894 119895) represent the inverted and actual
propagation loss calculated at discrete point (119894Δ119909 119895Δ119910) and119873119909and 119873
119910are the sample points along the horizontal and
vertical direction respectively It is easy to see that the smallerthe MAE the higher the accuracy of the algorithm
The MAE of the three algorithms for different noise levelobtained by (13) are presented in Table 2 and the minimum
Table 2 The comparison of the MAE for different algorithms withdifferent noise level
Noise level AlgorithmsMIWO ABC OGABC
0 dB 222 164 0091 dB 188 088 0752 dB 354 184 1053 dB 325 079 060
results are highlighted in bold It can be observed fromTable 2that the MAE of OGABC is the smallest one among the threealgorithms for different noise level namely the accuracy ofOGABC is superior to MIWO and ABC
Then in order to further test the performance ofOGABCthe radar sea clutter power generated bymeasured refractivityprofile [20] is also utilized for the inversion of the surfaceduct Figure 8 shows the comparison of the inverted profileobtained by the three algorithms with the measured profileand it is distinct that the inverted profile of OGABC is inexcellent agreement with the measured one
In addition Figure 9 presents the corresponding com-parison of the convergence curves of the MIWO ABC andOGABC It is observed that the three algorithms have thefaster convergence rate at the early stage of iterations andthe minimum fitness is hardly changed with iteration at themiddle and later stage of iterationsHowever theOGABChasthe smallest minimum fitness among the three algorithms
8 International Journal of Antennas and Propagation
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 0dB
(a)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 1dB
(b)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 2dB
(c)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000M
ean
min
imum
fitn
ess
Iterations
MIWOABCOGABC
Noise level 3dB
(d)
Figure 6 The comparison of the convergence curves of different algorithms with the same noise level
That is to say the OGABC is superior to MIWO and ABCaccording to the accuracy and convergence rate
5 Conclusion
In this paper an improved ABC algorithm called OGABCis presented by simultaneously merging the OBL strategyand global best search equation into the standard ABCalgorithm to tackle its deficiency of slow convergence rateand falling into the local best during the process of inversionof atmospheric duct Taking the inversion of the surfaceduct using RFC technique as an example the propagation
characteristics of radar sea clutter obtained from the sim-ulated and measured refractivity profile are treated as theobserved sea clutter power to examine the performance ofOGABC respectively For the simulated refractivity profilecase the Gaussian noise is added to the simulated radarsea clutter power to investigate the stability of the proposedOGABC algorithm and the histograms and the convergencecurves are used to analyze the accuracy and convergencerate Further investigation using the radar sea clutter powergenerated by themeasured refractivity profile is also involvedand the accuracy and convergence rate of the algorithmsare discussed by comparing the inverted refractivity profilewith the measured one and analyzing their convergence
International Journal of Antennas and Propagation 9
60
120
180
0 20 40 60 80 100
40
20
0
minus20
minus40
40
20
0
minus20
minus40
Hei
ght (
m)
60
120
180
Hei
ght (
m)
60
120
180
Hei
ght (
m)
Distance (km)0 20 40 60 80 100
Distance (km)0 20 40 60 80 100
Distance (km)
5
0
minus5
MIWO ABC OGABC
(a)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
40
20
0
minus20
minus40
MIWO ABC OGABC
(b)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
minus40
40
20
0
minus20
40
20
0
minus20
minus40
minus60
MIWO ABC OGABC
(c)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
50
0
minus50
MIWO ABC OGABC
(d)
Figure 7 The comparison of the difference between the inverted and actual coverage diagram with different noise level (a) 0 dB (b) 1 dB(c) 2 dB and (d) 3 dB
290 295 300 305 310 315 3200
50
100
150
200
Hei
ght (
m)
M (M-units)
MIWOABC
OGABCMeasured profile
Figure 8 The comparison of the inverted profile with the measured profile
10 International Journal of Antennas and Propagation
0 20 40 60 80 100 1204000
5000
6000
7000
8000
Min
imum
fitn
ess
Iterations
MIWOABCOGABC
Figure 9 The comparison of the convergence curves of differentalgorithms
curves In addition the inversion results are also analyzed andcompared with those of the MIWO and ABC The obtainedresults verify that the proposed OGABC algorithm outper-forms MIWO and ABC in terms of stability accuracy andconvergence rate Future work will focus on the experimentalresearch and the improvement of the inversion model
Competing Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is supported by the National Science Fund forDistinguished Young Scholars of China (no 61225002) andthe Young Scientists Fund of the National Natural ScienceFoundation of China (no 61302050)
References
[1] C Yardim P Gerstoft andW S Hodgkiss ldquoEstimation of radiorefractivity from radar clutter using BayesianMonte Carlo anal-ysisrdquo IEEE Transactions on Antennas and Propagation vol 54no 4 pp 1318ndash1327 2006
[2] C Yardim P Gerstoft andW S Hodgkiss ldquoTracking refractiv-ity from clutter using Kalman and particle filtersrdquo IEEE Trans-actions on Antennas and Propagation vol 56 no 4 pp 1058ndash1070 2008
[3] C Yardim P Gerstoft andW S Hodgkiss ldquoSensitivity analysisand performance estimation of refractivity from clutter tech-niquesrdquo Radio Science vol 44 pp 1ndash16 2009
[4] P Gerstoft L T Rogers J L Krolik andW S Hodgkiss ldquoInver-sion for refractivity parameters from radar sea clutterrdquo RadioScience vol 38 pp 1ndash22 2003
[5] A Karimian C Yardim P Gerstoft W S Hodgkiss and AE Barrios ldquoRefractivity estimation from sea clutter an invitedreviewrdquo Radio Science vol 46 pp 1ndash16 2009
[6] X-F Zhao S-X Huang and H-D Du ldquoTheoretical analysisand numerical experiments of variational adjoint approach forrefractivity estimationrdquo Radio Science vol 46 no 1 pp 1ndash122011
[7] X F Zhao and S X Huang ldquoAtmospheric duct estimationusing radar sea clutter returns by the adjoint method withregularization techniquerdquo Journal of Atmospheric and OceanicTechnology vol 31 no 6 pp 1250ndash1262 2014
[8] R Douvenot V Fabbro P Gerstoft C Bourlier and J SaillardldquoA duct mapping method using least squares support vectormachinesrdquo Radio Science vol 43 pp 1ndash12 2008
[9] B Wang Z-S Wu Z-W Zhao and H-G Wang ldquoRetrievingevaporation duct heights from radar sea clutter using particleswarm optimization (PSO) algorithmrdquo Progress In Electromag-netics Research M vol 9 pp 79ndash91 2009
[10] X-F Zhao S-X Huang J Xiang andW-L Shi ldquoRemote sens-ing of atmospheric duct parameters using simulated annealingrdquoChinese Physics B vol 20 no 9 Article ID 099201 8 pages 2011
[11] C Yang ldquoEstimation of the atmospheric duct from radarsea clutter using artificial bee colony optimization algorithmrdquoProgress in Electromagnetics Research vol 135 pp 183ndash199 2013
[12] D Karaboga and B Basturk ldquoA powerful and efficient algo-rithm for numerical function optimization artificial bee colony(ABC) algorithmrdquo Journal of Global Optimization vol 39 no 3pp 459ndash471 2007
[13] J Yang W-T Li X-W Shi L Xin and J-F Yu ldquoA hybrid ABC-DE algorithm and its application for time-modulated arrayspattern synthesisrdquo IEEE Transactions on Antennas and Pro-pagation vol 61 no 11 pp 5485ndash5495 2013
[14] X Zhang X Zhang S Y Yuen S L Ho and W N Fu ldquoAnimproved artificial bee colony algorithm for optimal design ofelectromagnetic devicesrdquo IEEE Transactions on Magnetics vol49 no 8 pp 4811ndash4816 2013
[15] S K Goudos K Siakavara A Theopoulos E E Vafiadis andJ N Sahalos ldquoApplication of Gbest-guided artificial beecolonyalgorithm to passive UHF RFID tag designrdquo InternationalJournal of Microwave and Wireless Technologies 2015
[16] G Zhu and S Kwong ldquoGbest-guided artificial bee colonyalgorithm for numerical function optimizationrdquo Applied Math-ematics and Computation vol 217 no 7 pp 3166ndash3173 2010
[17] W-F Gao S-Y Liu and L-L Huang ldquoA novel artificialbee colony algorithm based on modified search equation andorthogonal learningrdquo IEEE Transactions on Cybernetics vol 43no 3 pp 1011ndash1024 2013
[18] R S Rahnamayan H R Tizhoosh and M M A SalamaldquoOpposition-based differential evolutionrdquo IEEETransactions onEvolutionary Computation vol 12 no 1 pp 64ndash79 2008
[19] A E Barrios ldquoTerrain parabolic equation model for propaga-tion in the troposphererdquo IEEE Transactions on Antennas andPropagation vol 42 no 1 pp 90ndash98 1994
[20] A Karimian C Yardim W S Hodgkiss P Gerstoft and A EBarrios ldquoEstimation of radio refractivity using a multiple angleclutter modelrdquo Radio Science vol 47 pp 1ndash9 2012
[21] A Basak D Maity and S Das ldquoA differential invasive weedoptimization algorithm for improved global numerical opti-mizationrdquo Applied Mathematics and Computation vol 219 no12 pp 6645ndash6668 2013
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
4 International Journal of Antennas and Propagation
Opposition-based population initializationwith (8) and (9)
In employed bees stage update the positionof food sources using the global best search (10) and choose the better one
In onlooker bees stage update the position offood sources using the global best search (10) and choose the better one
Does scout exist
Terminal condition
No
End
Yes
Begin
Opposition-based generation jumpingaccording to the jumping rate
Replace the scout with a random solution
Yes
No
Figure 1 The flowchart of the proposed OGABC algorithm
Step 71 According to the jumping rate decide whetheropposition-based generation jumping appears or not if itappears the new opposite population 119874119883 of 119883 are producedby
119874119883119894119895= 119897
min119895
+ 119906max119895
minus 119883119894119895
119894 = 1 2 119873 119895 = 1 2 119863
(12)
where 119897min119895
and 119906max119895
are the minimum and maximum valueof the 119895th dimension in the current population
Step 72 Choose the119873 best solutions from [119883119874119883] accordingto the fitness value to generate the population for the nextiteration
Step 8 Repeat Step 2 to Step 7 until a terminating conditionis met
The flowchart of the proposed OGABC is shown inFigure 1
4 The Numerical Results and Discussions
In this section the inversion results are given to validatethe optimization performance of the proposed OGABC Inthe following we take the inversion of the four-parametersurface duct with RFC technique as an example to analyzethe performance of OGABC and the inversion results arecompared with those of the MIWO [21] and ABC
In fact the essence of the inversion of surface duct isto obtain its corresponding refractivity profile determined
International Journal of Antennas and Propagation 5
005 01 015 02 0250
5
10
15O
ccur
renc
eO
ccur
renc
eO
ccur
renc
eO
ccur
renc
e
005 01 015 02 0250
5
10
15
005 01 015 02 0250
5
10
15
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus220
5
10
15
0
5
10
15
0
5
10
15
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
14 16 18 20 22 24 26 28 300
5
10
15
MIWO ABC OGABC
14 16 18 20 22 24 26 28 300
5
10
15
14 16 18 20 22 24 26 28 300
5
10
15
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 2 The comparison of the histograms of the inversion results for different algorithms with the noise level of 0 dB
Table 1 The lower and upper search bounds of the parameters
Parameter Lower bound Upper bound Units1198881
00 025 M-unitsm1198882
minus35 minus10 M-unitsmℎ1
250 500 mℎ2
100 300 m
by (4) In other words the optimization problem can betranslated into the inversion of the parameters of the surfaceduct m = (119888
1 1198882 ℎ1 ℎ2) and the lower and upper bounds of
the surface duct parameters are shown in Table 1In numerical simulation the inversions are implemented
by the radar sea clutter power calculated by parabolic equa-tion method using the simulated and measured refractivityprofile respectively During the inversion the simulatedradar sea clutter power from 10Km to 50Km is regardedas the observed radar sea clutter power and the radarsystem operates at a frequency of 10GHz power of 914 dBmantenna gain of 528 dB antenna height of 7m beam widthof 07∘ 600m range bin and HH polarization In additionthe control parameters of OGABC are given as followsthe population size is 60 the number of food sources is30 the parameter limit is 25 the maximum number ofiterations is 120 and the jumping rate of OBL is 03 [18]the parameters settings for MIWO are given as follows theinitial population size is 30 the maximum population size is
40 the maximum number of iterations is 120 the nonlinearmodulation index is 3 theminimumandmaximumnumbersof seeds are 0 and 10 the initial and final value of standarddeviation are 100 and 00001 and the inversion results areobtained from 30 independent runs for each algorithm forthe simulated refractivity profile case For a fair comparisonbetween ABC and OGABC they are examined using thesame parameter settings and the settings of the radar systemremain unchanged in the process of inversions
For the simulated refractivity case the radar sea clutterpower computed by the parameters of the surface duct m =
(013 minus25 40 20) is utilized to the inversion of the surfaceduct Moreover the Gaussian noise with zero mean anddifferent standard deviations is added to the simulated radarsea clutter power to examine the stability of the algorithmsand the standard deviation is employed to represent thenoise level Also the histograms and convergence curves arepresented to analyze the accuracy and convergence rate indetail
Figures 2ndash5 give the comparison of the histograms ofthe inversions of the surface duct parameters for differentalgorithms at a specific noise level and the red lines denotethe actual parameter of surface duct It is obvious that thedistribution of inversion parameters obtained by theOGABCis more intensive than those of MIWO and ABC for differentnoise level In addition the inversion results of OGABCachieve the most occurrence at the vicinity of the actualparameter compared with those of theMIWO andABCThatis to say the proposed OGABC is the most stable algorithm
6 International Journal of Antennas and Propagation
006 008 01 012 014 016 0180
5
10
15
006 008 01 012 014 016 0180
5
10
15
006 008 01 012 014 016 0180
5
10
15
0
10
20
30
0
10
20
30
0
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
10 15 20 25 3005
101520
10 15 20 25 3005
101520
10 15 20 25 3005
101520
MIWO ABC OGABC
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22
Occ
urre
nce
Occ
urre
nce
Occ
urre
nce
Occ
urre
nce
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 3 The comparison of the histograms of the inversion results for different algorithms with the noise level of 1 dB
005 01 015 02 0250
5
10
15
Occ
urre
nce
005 01 015 02 0250
5
10
15
005 01 015 02 0250
5
10
15
0
5
10
15
Occ
urre
nce
0
5
10
15
0
5
10
15
38 40 42 44 46 480
5
10
15
Occ
urre
nce
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
10 15 20 25 3005
101520
Occ
urre
nce
MIWO ABC OGABC
10 15 20 25 3005
101520
10 15 20 25 3005
101520
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 4 The comparison of the histograms of the inversion results for different algorithms with the noise level of 2 dB
among the three algorithms It is likely due to the fact thatthe global best search equation not only enhances the globalsearch ability but also avoids falling into the local minimum
To study the convergence performance of the OGABCthe comparisons of the convergence curves of different
algorithms based on the inversion results given in Figures 2ndash5with the same noise level are demonstrated in Figure 6 It canbe seen that the convergence rate of OGABC is faster thanthat of the MIWO and ABC besides the OGABC has thesmallest meanminimum fitness at the end of the iteration for
International Journal of Antennas and Propagation 7
0 005 01 015 02 02505
101520
Occ
urre
nce
0 005 01 015 02 02505
101520
0 005 01 015 02 02505
101520
05
101520
Occ
urre
nce
05
101520
05
101520
36 38 40 42 44 4605
101520
Occ
urre
nce
36 38 40 42 44 4605
101520
36 38 40 42 44 4605
101520
15 20 25 300
5
10
Occ
urre
nce
MIWO ABC OGABC
15 20 25 300
5
10
15 20 25 300
5
10
minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 5 The comparison of the histograms of the inversion results for different algorithms with the noise level of 3 dB
different noise levelThe convergence curves indicate that theOGABC is the best algorithm among the three algorithmsThis can be attributed to the fact that the OBL can produce arelatively excellent initial population in the initialization stageand a generation jumping to form a better population in thesearch process
The accuracy of the inversion of atmospheric duct is ofcrucial importance to exactly predict the marine electromag-netic environment Hence the comparisons of the differencebetween the inverted and actual radar coverage diagramsimulated by parabolic equation method for different noiselevel are shown in Figure 7 and the peaks of the parameterdistributions [20] in Figures 2ndash5 are treated as the invertedparameters of the surface duct to simulate their correspond-ing coverage diagram Nevertheless it is hardly possible toevaluate the quality of the algorithms according to the resultspresented in Figure 7 Thus a new quantitative evaluationcriterion named Mean Absolute Error (MAE) is given by
MAE =
sum119873119909
119894=1sum119873119910
119895=1
1003816100381610038161003816PL119894 (119894 119895) minus PL119886(119894 119895)
1003816100381610038161003816
119873119909119873119910
(13)
where PL119894(119894 119895) and PL
119886(119894 119895) represent the inverted and actual
propagation loss calculated at discrete point (119894Δ119909 119895Δ119910) and119873119909and 119873
119910are the sample points along the horizontal and
vertical direction respectively It is easy to see that the smallerthe MAE the higher the accuracy of the algorithm
The MAE of the three algorithms for different noise levelobtained by (13) are presented in Table 2 and the minimum
Table 2 The comparison of the MAE for different algorithms withdifferent noise level
Noise level AlgorithmsMIWO ABC OGABC
0 dB 222 164 0091 dB 188 088 0752 dB 354 184 1053 dB 325 079 060
results are highlighted in bold It can be observed fromTable 2that the MAE of OGABC is the smallest one among the threealgorithms for different noise level namely the accuracy ofOGABC is superior to MIWO and ABC
Then in order to further test the performance ofOGABCthe radar sea clutter power generated bymeasured refractivityprofile [20] is also utilized for the inversion of the surfaceduct Figure 8 shows the comparison of the inverted profileobtained by the three algorithms with the measured profileand it is distinct that the inverted profile of OGABC is inexcellent agreement with the measured one
In addition Figure 9 presents the corresponding com-parison of the convergence curves of the MIWO ABC andOGABC It is observed that the three algorithms have thefaster convergence rate at the early stage of iterations andthe minimum fitness is hardly changed with iteration at themiddle and later stage of iterationsHowever theOGABChasthe smallest minimum fitness among the three algorithms
8 International Journal of Antennas and Propagation
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 0dB
(a)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 1dB
(b)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 2dB
(c)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000M
ean
min
imum
fitn
ess
Iterations
MIWOABCOGABC
Noise level 3dB
(d)
Figure 6 The comparison of the convergence curves of different algorithms with the same noise level
That is to say the OGABC is superior to MIWO and ABCaccording to the accuracy and convergence rate
5 Conclusion
In this paper an improved ABC algorithm called OGABCis presented by simultaneously merging the OBL strategyand global best search equation into the standard ABCalgorithm to tackle its deficiency of slow convergence rateand falling into the local best during the process of inversionof atmospheric duct Taking the inversion of the surfaceduct using RFC technique as an example the propagation
characteristics of radar sea clutter obtained from the sim-ulated and measured refractivity profile are treated as theobserved sea clutter power to examine the performance ofOGABC respectively For the simulated refractivity profilecase the Gaussian noise is added to the simulated radarsea clutter power to investigate the stability of the proposedOGABC algorithm and the histograms and the convergencecurves are used to analyze the accuracy and convergencerate Further investigation using the radar sea clutter powergenerated by themeasured refractivity profile is also involvedand the accuracy and convergence rate of the algorithmsare discussed by comparing the inverted refractivity profilewith the measured one and analyzing their convergence
International Journal of Antennas and Propagation 9
60
120
180
0 20 40 60 80 100
40
20
0
minus20
minus40
40
20
0
minus20
minus40
Hei
ght (
m)
60
120
180
Hei
ght (
m)
60
120
180
Hei
ght (
m)
Distance (km)0 20 40 60 80 100
Distance (km)0 20 40 60 80 100
Distance (km)
5
0
minus5
MIWO ABC OGABC
(a)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
40
20
0
minus20
minus40
MIWO ABC OGABC
(b)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
minus40
40
20
0
minus20
40
20
0
minus20
minus40
minus60
MIWO ABC OGABC
(c)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
50
0
minus50
MIWO ABC OGABC
(d)
Figure 7 The comparison of the difference between the inverted and actual coverage diagram with different noise level (a) 0 dB (b) 1 dB(c) 2 dB and (d) 3 dB
290 295 300 305 310 315 3200
50
100
150
200
Hei
ght (
m)
M (M-units)
MIWOABC
OGABCMeasured profile
Figure 8 The comparison of the inverted profile with the measured profile
10 International Journal of Antennas and Propagation
0 20 40 60 80 100 1204000
5000
6000
7000
8000
Min
imum
fitn
ess
Iterations
MIWOABCOGABC
Figure 9 The comparison of the convergence curves of differentalgorithms
curves In addition the inversion results are also analyzed andcompared with those of the MIWO and ABC The obtainedresults verify that the proposed OGABC algorithm outper-forms MIWO and ABC in terms of stability accuracy andconvergence rate Future work will focus on the experimentalresearch and the improvement of the inversion model
Competing Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is supported by the National Science Fund forDistinguished Young Scholars of China (no 61225002) andthe Young Scientists Fund of the National Natural ScienceFoundation of China (no 61302050)
References
[1] C Yardim P Gerstoft andW S Hodgkiss ldquoEstimation of radiorefractivity from radar clutter using BayesianMonte Carlo anal-ysisrdquo IEEE Transactions on Antennas and Propagation vol 54no 4 pp 1318ndash1327 2006
[2] C Yardim P Gerstoft andW S Hodgkiss ldquoTracking refractiv-ity from clutter using Kalman and particle filtersrdquo IEEE Trans-actions on Antennas and Propagation vol 56 no 4 pp 1058ndash1070 2008
[3] C Yardim P Gerstoft andW S Hodgkiss ldquoSensitivity analysisand performance estimation of refractivity from clutter tech-niquesrdquo Radio Science vol 44 pp 1ndash16 2009
[4] P Gerstoft L T Rogers J L Krolik andW S Hodgkiss ldquoInver-sion for refractivity parameters from radar sea clutterrdquo RadioScience vol 38 pp 1ndash22 2003
[5] A Karimian C Yardim P Gerstoft W S Hodgkiss and AE Barrios ldquoRefractivity estimation from sea clutter an invitedreviewrdquo Radio Science vol 46 pp 1ndash16 2009
[6] X-F Zhao S-X Huang and H-D Du ldquoTheoretical analysisand numerical experiments of variational adjoint approach forrefractivity estimationrdquo Radio Science vol 46 no 1 pp 1ndash122011
[7] X F Zhao and S X Huang ldquoAtmospheric duct estimationusing radar sea clutter returns by the adjoint method withregularization techniquerdquo Journal of Atmospheric and OceanicTechnology vol 31 no 6 pp 1250ndash1262 2014
[8] R Douvenot V Fabbro P Gerstoft C Bourlier and J SaillardldquoA duct mapping method using least squares support vectormachinesrdquo Radio Science vol 43 pp 1ndash12 2008
[9] B Wang Z-S Wu Z-W Zhao and H-G Wang ldquoRetrievingevaporation duct heights from radar sea clutter using particleswarm optimization (PSO) algorithmrdquo Progress In Electromag-netics Research M vol 9 pp 79ndash91 2009
[10] X-F Zhao S-X Huang J Xiang andW-L Shi ldquoRemote sens-ing of atmospheric duct parameters using simulated annealingrdquoChinese Physics B vol 20 no 9 Article ID 099201 8 pages 2011
[11] C Yang ldquoEstimation of the atmospheric duct from radarsea clutter using artificial bee colony optimization algorithmrdquoProgress in Electromagnetics Research vol 135 pp 183ndash199 2013
[12] D Karaboga and B Basturk ldquoA powerful and efficient algo-rithm for numerical function optimization artificial bee colony(ABC) algorithmrdquo Journal of Global Optimization vol 39 no 3pp 459ndash471 2007
[13] J Yang W-T Li X-W Shi L Xin and J-F Yu ldquoA hybrid ABC-DE algorithm and its application for time-modulated arrayspattern synthesisrdquo IEEE Transactions on Antennas and Pro-pagation vol 61 no 11 pp 5485ndash5495 2013
[14] X Zhang X Zhang S Y Yuen S L Ho and W N Fu ldquoAnimproved artificial bee colony algorithm for optimal design ofelectromagnetic devicesrdquo IEEE Transactions on Magnetics vol49 no 8 pp 4811ndash4816 2013
[15] S K Goudos K Siakavara A Theopoulos E E Vafiadis andJ N Sahalos ldquoApplication of Gbest-guided artificial beecolonyalgorithm to passive UHF RFID tag designrdquo InternationalJournal of Microwave and Wireless Technologies 2015
[16] G Zhu and S Kwong ldquoGbest-guided artificial bee colonyalgorithm for numerical function optimizationrdquo Applied Math-ematics and Computation vol 217 no 7 pp 3166ndash3173 2010
[17] W-F Gao S-Y Liu and L-L Huang ldquoA novel artificialbee colony algorithm based on modified search equation andorthogonal learningrdquo IEEE Transactions on Cybernetics vol 43no 3 pp 1011ndash1024 2013
[18] R S Rahnamayan H R Tizhoosh and M M A SalamaldquoOpposition-based differential evolutionrdquo IEEETransactions onEvolutionary Computation vol 12 no 1 pp 64ndash79 2008
[19] A E Barrios ldquoTerrain parabolic equation model for propaga-tion in the troposphererdquo IEEE Transactions on Antennas andPropagation vol 42 no 1 pp 90ndash98 1994
[20] A Karimian C Yardim W S Hodgkiss P Gerstoft and A EBarrios ldquoEstimation of radio refractivity using a multiple angleclutter modelrdquo Radio Science vol 47 pp 1ndash9 2012
[21] A Basak D Maity and S Das ldquoA differential invasive weedoptimization algorithm for improved global numerical opti-mizationrdquo Applied Mathematics and Computation vol 219 no12 pp 6645ndash6668 2013
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of Antennas and Propagation 5
005 01 015 02 0250
5
10
15O
ccur
renc
eO
ccur
renc
eO
ccur
renc
eO
ccur
renc
e
005 01 015 02 0250
5
10
15
005 01 015 02 0250
5
10
15
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus220
5
10
15
0
5
10
15
0
5
10
15
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
14 16 18 20 22 24 26 28 300
5
10
15
MIWO ABC OGABC
14 16 18 20 22 24 26 28 300
5
10
15
14 16 18 20 22 24 26 28 300
5
10
15
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 2 The comparison of the histograms of the inversion results for different algorithms with the noise level of 0 dB
Table 1 The lower and upper search bounds of the parameters
Parameter Lower bound Upper bound Units1198881
00 025 M-unitsm1198882
minus35 minus10 M-unitsmℎ1
250 500 mℎ2
100 300 m
by (4) In other words the optimization problem can betranslated into the inversion of the parameters of the surfaceduct m = (119888
1 1198882 ℎ1 ℎ2) and the lower and upper bounds of
the surface duct parameters are shown in Table 1In numerical simulation the inversions are implemented
by the radar sea clutter power calculated by parabolic equa-tion method using the simulated and measured refractivityprofile respectively During the inversion the simulatedradar sea clutter power from 10Km to 50Km is regardedas the observed radar sea clutter power and the radarsystem operates at a frequency of 10GHz power of 914 dBmantenna gain of 528 dB antenna height of 7m beam widthof 07∘ 600m range bin and HH polarization In additionthe control parameters of OGABC are given as followsthe population size is 60 the number of food sources is30 the parameter limit is 25 the maximum number ofiterations is 120 and the jumping rate of OBL is 03 [18]the parameters settings for MIWO are given as follows theinitial population size is 30 the maximum population size is
40 the maximum number of iterations is 120 the nonlinearmodulation index is 3 theminimumandmaximumnumbersof seeds are 0 and 10 the initial and final value of standarddeviation are 100 and 00001 and the inversion results areobtained from 30 independent runs for each algorithm forthe simulated refractivity profile case For a fair comparisonbetween ABC and OGABC they are examined using thesame parameter settings and the settings of the radar systemremain unchanged in the process of inversions
For the simulated refractivity case the radar sea clutterpower computed by the parameters of the surface duct m =
(013 minus25 40 20) is utilized to the inversion of the surfaceduct Moreover the Gaussian noise with zero mean anddifferent standard deviations is added to the simulated radarsea clutter power to examine the stability of the algorithmsand the standard deviation is employed to represent thenoise level Also the histograms and convergence curves arepresented to analyze the accuracy and convergence rate indetail
Figures 2ndash5 give the comparison of the histograms ofthe inversions of the surface duct parameters for differentalgorithms at a specific noise level and the red lines denotethe actual parameter of surface duct It is obvious that thedistribution of inversion parameters obtained by theOGABCis more intensive than those of MIWO and ABC for differentnoise level In addition the inversion results of OGABCachieve the most occurrence at the vicinity of the actualparameter compared with those of theMIWO andABCThatis to say the proposed OGABC is the most stable algorithm
6 International Journal of Antennas and Propagation
006 008 01 012 014 016 0180
5
10
15
006 008 01 012 014 016 0180
5
10
15
006 008 01 012 014 016 0180
5
10
15
0
10
20
30
0
10
20
30
0
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
10 15 20 25 3005
101520
10 15 20 25 3005
101520
10 15 20 25 3005
101520
MIWO ABC OGABC
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22
Occ
urre
nce
Occ
urre
nce
Occ
urre
nce
Occ
urre
nce
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 3 The comparison of the histograms of the inversion results for different algorithms with the noise level of 1 dB
005 01 015 02 0250
5
10
15
Occ
urre
nce
005 01 015 02 0250
5
10
15
005 01 015 02 0250
5
10
15
0
5
10
15
Occ
urre
nce
0
5
10
15
0
5
10
15
38 40 42 44 46 480
5
10
15
Occ
urre
nce
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
10 15 20 25 3005
101520
Occ
urre
nce
MIWO ABC OGABC
10 15 20 25 3005
101520
10 15 20 25 3005
101520
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 4 The comparison of the histograms of the inversion results for different algorithms with the noise level of 2 dB
among the three algorithms It is likely due to the fact thatthe global best search equation not only enhances the globalsearch ability but also avoids falling into the local minimum
To study the convergence performance of the OGABCthe comparisons of the convergence curves of different
algorithms based on the inversion results given in Figures 2ndash5with the same noise level are demonstrated in Figure 6 It canbe seen that the convergence rate of OGABC is faster thanthat of the MIWO and ABC besides the OGABC has thesmallest meanminimum fitness at the end of the iteration for
International Journal of Antennas and Propagation 7
0 005 01 015 02 02505
101520
Occ
urre
nce
0 005 01 015 02 02505
101520
0 005 01 015 02 02505
101520
05
101520
Occ
urre
nce
05
101520
05
101520
36 38 40 42 44 4605
101520
Occ
urre
nce
36 38 40 42 44 4605
101520
36 38 40 42 44 4605
101520
15 20 25 300
5
10
Occ
urre
nce
MIWO ABC OGABC
15 20 25 300
5
10
15 20 25 300
5
10
minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 5 The comparison of the histograms of the inversion results for different algorithms with the noise level of 3 dB
different noise levelThe convergence curves indicate that theOGABC is the best algorithm among the three algorithmsThis can be attributed to the fact that the OBL can produce arelatively excellent initial population in the initialization stageand a generation jumping to form a better population in thesearch process
The accuracy of the inversion of atmospheric duct is ofcrucial importance to exactly predict the marine electromag-netic environment Hence the comparisons of the differencebetween the inverted and actual radar coverage diagramsimulated by parabolic equation method for different noiselevel are shown in Figure 7 and the peaks of the parameterdistributions [20] in Figures 2ndash5 are treated as the invertedparameters of the surface duct to simulate their correspond-ing coverage diagram Nevertheless it is hardly possible toevaluate the quality of the algorithms according to the resultspresented in Figure 7 Thus a new quantitative evaluationcriterion named Mean Absolute Error (MAE) is given by
MAE =
sum119873119909
119894=1sum119873119910
119895=1
1003816100381610038161003816PL119894 (119894 119895) minus PL119886(119894 119895)
1003816100381610038161003816
119873119909119873119910
(13)
where PL119894(119894 119895) and PL
119886(119894 119895) represent the inverted and actual
propagation loss calculated at discrete point (119894Δ119909 119895Δ119910) and119873119909and 119873
119910are the sample points along the horizontal and
vertical direction respectively It is easy to see that the smallerthe MAE the higher the accuracy of the algorithm
The MAE of the three algorithms for different noise levelobtained by (13) are presented in Table 2 and the minimum
Table 2 The comparison of the MAE for different algorithms withdifferent noise level
Noise level AlgorithmsMIWO ABC OGABC
0 dB 222 164 0091 dB 188 088 0752 dB 354 184 1053 dB 325 079 060
results are highlighted in bold It can be observed fromTable 2that the MAE of OGABC is the smallest one among the threealgorithms for different noise level namely the accuracy ofOGABC is superior to MIWO and ABC
Then in order to further test the performance ofOGABCthe radar sea clutter power generated bymeasured refractivityprofile [20] is also utilized for the inversion of the surfaceduct Figure 8 shows the comparison of the inverted profileobtained by the three algorithms with the measured profileand it is distinct that the inverted profile of OGABC is inexcellent agreement with the measured one
In addition Figure 9 presents the corresponding com-parison of the convergence curves of the MIWO ABC andOGABC It is observed that the three algorithms have thefaster convergence rate at the early stage of iterations andthe minimum fitness is hardly changed with iteration at themiddle and later stage of iterationsHowever theOGABChasthe smallest minimum fitness among the three algorithms
8 International Journal of Antennas and Propagation
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 0dB
(a)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 1dB
(b)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 2dB
(c)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000M
ean
min
imum
fitn
ess
Iterations
MIWOABCOGABC
Noise level 3dB
(d)
Figure 6 The comparison of the convergence curves of different algorithms with the same noise level
That is to say the OGABC is superior to MIWO and ABCaccording to the accuracy and convergence rate
5 Conclusion
In this paper an improved ABC algorithm called OGABCis presented by simultaneously merging the OBL strategyand global best search equation into the standard ABCalgorithm to tackle its deficiency of slow convergence rateand falling into the local best during the process of inversionof atmospheric duct Taking the inversion of the surfaceduct using RFC technique as an example the propagation
characteristics of radar sea clutter obtained from the sim-ulated and measured refractivity profile are treated as theobserved sea clutter power to examine the performance ofOGABC respectively For the simulated refractivity profilecase the Gaussian noise is added to the simulated radarsea clutter power to investigate the stability of the proposedOGABC algorithm and the histograms and the convergencecurves are used to analyze the accuracy and convergencerate Further investigation using the radar sea clutter powergenerated by themeasured refractivity profile is also involvedand the accuracy and convergence rate of the algorithmsare discussed by comparing the inverted refractivity profilewith the measured one and analyzing their convergence
International Journal of Antennas and Propagation 9
60
120
180
0 20 40 60 80 100
40
20
0
minus20
minus40
40
20
0
minus20
minus40
Hei
ght (
m)
60
120
180
Hei
ght (
m)
60
120
180
Hei
ght (
m)
Distance (km)0 20 40 60 80 100
Distance (km)0 20 40 60 80 100
Distance (km)
5
0
minus5
MIWO ABC OGABC
(a)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
40
20
0
minus20
minus40
MIWO ABC OGABC
(b)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
minus40
40
20
0
minus20
40
20
0
minus20
minus40
minus60
MIWO ABC OGABC
(c)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
50
0
minus50
MIWO ABC OGABC
(d)
Figure 7 The comparison of the difference between the inverted and actual coverage diagram with different noise level (a) 0 dB (b) 1 dB(c) 2 dB and (d) 3 dB
290 295 300 305 310 315 3200
50
100
150
200
Hei
ght (
m)
M (M-units)
MIWOABC
OGABCMeasured profile
Figure 8 The comparison of the inverted profile with the measured profile
10 International Journal of Antennas and Propagation
0 20 40 60 80 100 1204000
5000
6000
7000
8000
Min
imum
fitn
ess
Iterations
MIWOABCOGABC
Figure 9 The comparison of the convergence curves of differentalgorithms
curves In addition the inversion results are also analyzed andcompared with those of the MIWO and ABC The obtainedresults verify that the proposed OGABC algorithm outper-forms MIWO and ABC in terms of stability accuracy andconvergence rate Future work will focus on the experimentalresearch and the improvement of the inversion model
Competing Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is supported by the National Science Fund forDistinguished Young Scholars of China (no 61225002) andthe Young Scientists Fund of the National Natural ScienceFoundation of China (no 61302050)
References
[1] C Yardim P Gerstoft andW S Hodgkiss ldquoEstimation of radiorefractivity from radar clutter using BayesianMonte Carlo anal-ysisrdquo IEEE Transactions on Antennas and Propagation vol 54no 4 pp 1318ndash1327 2006
[2] C Yardim P Gerstoft andW S Hodgkiss ldquoTracking refractiv-ity from clutter using Kalman and particle filtersrdquo IEEE Trans-actions on Antennas and Propagation vol 56 no 4 pp 1058ndash1070 2008
[3] C Yardim P Gerstoft andW S Hodgkiss ldquoSensitivity analysisand performance estimation of refractivity from clutter tech-niquesrdquo Radio Science vol 44 pp 1ndash16 2009
[4] P Gerstoft L T Rogers J L Krolik andW S Hodgkiss ldquoInver-sion for refractivity parameters from radar sea clutterrdquo RadioScience vol 38 pp 1ndash22 2003
[5] A Karimian C Yardim P Gerstoft W S Hodgkiss and AE Barrios ldquoRefractivity estimation from sea clutter an invitedreviewrdquo Radio Science vol 46 pp 1ndash16 2009
[6] X-F Zhao S-X Huang and H-D Du ldquoTheoretical analysisand numerical experiments of variational adjoint approach forrefractivity estimationrdquo Radio Science vol 46 no 1 pp 1ndash122011
[7] X F Zhao and S X Huang ldquoAtmospheric duct estimationusing radar sea clutter returns by the adjoint method withregularization techniquerdquo Journal of Atmospheric and OceanicTechnology vol 31 no 6 pp 1250ndash1262 2014
[8] R Douvenot V Fabbro P Gerstoft C Bourlier and J SaillardldquoA duct mapping method using least squares support vectormachinesrdquo Radio Science vol 43 pp 1ndash12 2008
[9] B Wang Z-S Wu Z-W Zhao and H-G Wang ldquoRetrievingevaporation duct heights from radar sea clutter using particleswarm optimization (PSO) algorithmrdquo Progress In Electromag-netics Research M vol 9 pp 79ndash91 2009
[10] X-F Zhao S-X Huang J Xiang andW-L Shi ldquoRemote sens-ing of atmospheric duct parameters using simulated annealingrdquoChinese Physics B vol 20 no 9 Article ID 099201 8 pages 2011
[11] C Yang ldquoEstimation of the atmospheric duct from radarsea clutter using artificial bee colony optimization algorithmrdquoProgress in Electromagnetics Research vol 135 pp 183ndash199 2013
[12] D Karaboga and B Basturk ldquoA powerful and efficient algo-rithm for numerical function optimization artificial bee colony(ABC) algorithmrdquo Journal of Global Optimization vol 39 no 3pp 459ndash471 2007
[13] J Yang W-T Li X-W Shi L Xin and J-F Yu ldquoA hybrid ABC-DE algorithm and its application for time-modulated arrayspattern synthesisrdquo IEEE Transactions on Antennas and Pro-pagation vol 61 no 11 pp 5485ndash5495 2013
[14] X Zhang X Zhang S Y Yuen S L Ho and W N Fu ldquoAnimproved artificial bee colony algorithm for optimal design ofelectromagnetic devicesrdquo IEEE Transactions on Magnetics vol49 no 8 pp 4811ndash4816 2013
[15] S K Goudos K Siakavara A Theopoulos E E Vafiadis andJ N Sahalos ldquoApplication of Gbest-guided artificial beecolonyalgorithm to passive UHF RFID tag designrdquo InternationalJournal of Microwave and Wireless Technologies 2015
[16] G Zhu and S Kwong ldquoGbest-guided artificial bee colonyalgorithm for numerical function optimizationrdquo Applied Math-ematics and Computation vol 217 no 7 pp 3166ndash3173 2010
[17] W-F Gao S-Y Liu and L-L Huang ldquoA novel artificialbee colony algorithm based on modified search equation andorthogonal learningrdquo IEEE Transactions on Cybernetics vol 43no 3 pp 1011ndash1024 2013
[18] R S Rahnamayan H R Tizhoosh and M M A SalamaldquoOpposition-based differential evolutionrdquo IEEETransactions onEvolutionary Computation vol 12 no 1 pp 64ndash79 2008
[19] A E Barrios ldquoTerrain parabolic equation model for propaga-tion in the troposphererdquo IEEE Transactions on Antennas andPropagation vol 42 no 1 pp 90ndash98 1994
[20] A Karimian C Yardim W S Hodgkiss P Gerstoft and A EBarrios ldquoEstimation of radio refractivity using a multiple angleclutter modelrdquo Radio Science vol 47 pp 1ndash9 2012
[21] A Basak D Maity and S Das ldquoA differential invasive weedoptimization algorithm for improved global numerical opti-mizationrdquo Applied Mathematics and Computation vol 219 no12 pp 6645ndash6668 2013
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
6 International Journal of Antennas and Propagation
006 008 01 012 014 016 0180
5
10
15
006 008 01 012 014 016 0180
5
10
15
006 008 01 012 014 016 0180
5
10
15
0
10
20
30
0
10
20
30
0
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
38 39 40 41 42 43 44 45 460
10
20
30
10 15 20 25 3005
101520
10 15 20 25 3005
101520
10 15 20 25 3005
101520
MIWO ABC OGABC
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22
Occ
urre
nce
Occ
urre
nce
Occ
urre
nce
Occ
urre
nce
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 3 The comparison of the histograms of the inversion results for different algorithms with the noise level of 1 dB
005 01 015 02 0250
5
10
15
Occ
urre
nce
005 01 015 02 0250
5
10
15
005 01 015 02 0250
5
10
15
0
5
10
15
Occ
urre
nce
0
5
10
15
0
5
10
15
38 40 42 44 46 480
5
10
15
Occ
urre
nce
38 40 42 44 46 480
5
10
15
38 40 42 44 46 480
5
10
15
10 15 20 25 3005
101520
Occ
urre
nce
MIWO ABC OGABC
10 15 20 25 3005
101520
10 15 20 25 3005
101520
minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus36 minus34 minus32 minus3 minus28 minus26 minus24 minus22
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 4 The comparison of the histograms of the inversion results for different algorithms with the noise level of 2 dB
among the three algorithms It is likely due to the fact thatthe global best search equation not only enhances the globalsearch ability but also avoids falling into the local minimum
To study the convergence performance of the OGABCthe comparisons of the convergence curves of different
algorithms based on the inversion results given in Figures 2ndash5with the same noise level are demonstrated in Figure 6 It canbe seen that the convergence rate of OGABC is faster thanthat of the MIWO and ABC besides the OGABC has thesmallest meanminimum fitness at the end of the iteration for
International Journal of Antennas and Propagation 7
0 005 01 015 02 02505
101520
Occ
urre
nce
0 005 01 015 02 02505
101520
0 005 01 015 02 02505
101520
05
101520
Occ
urre
nce
05
101520
05
101520
36 38 40 42 44 4605
101520
Occ
urre
nce
36 38 40 42 44 4605
101520
36 38 40 42 44 4605
101520
15 20 25 300
5
10
Occ
urre
nce
MIWO ABC OGABC
15 20 25 300
5
10
15 20 25 300
5
10
minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 5 The comparison of the histograms of the inversion results for different algorithms with the noise level of 3 dB
different noise levelThe convergence curves indicate that theOGABC is the best algorithm among the three algorithmsThis can be attributed to the fact that the OBL can produce arelatively excellent initial population in the initialization stageand a generation jumping to form a better population in thesearch process
The accuracy of the inversion of atmospheric duct is ofcrucial importance to exactly predict the marine electromag-netic environment Hence the comparisons of the differencebetween the inverted and actual radar coverage diagramsimulated by parabolic equation method for different noiselevel are shown in Figure 7 and the peaks of the parameterdistributions [20] in Figures 2ndash5 are treated as the invertedparameters of the surface duct to simulate their correspond-ing coverage diagram Nevertheless it is hardly possible toevaluate the quality of the algorithms according to the resultspresented in Figure 7 Thus a new quantitative evaluationcriterion named Mean Absolute Error (MAE) is given by
MAE =
sum119873119909
119894=1sum119873119910
119895=1
1003816100381610038161003816PL119894 (119894 119895) minus PL119886(119894 119895)
1003816100381610038161003816
119873119909119873119910
(13)
where PL119894(119894 119895) and PL
119886(119894 119895) represent the inverted and actual
propagation loss calculated at discrete point (119894Δ119909 119895Δ119910) and119873119909and 119873
119910are the sample points along the horizontal and
vertical direction respectively It is easy to see that the smallerthe MAE the higher the accuracy of the algorithm
The MAE of the three algorithms for different noise levelobtained by (13) are presented in Table 2 and the minimum
Table 2 The comparison of the MAE for different algorithms withdifferent noise level
Noise level AlgorithmsMIWO ABC OGABC
0 dB 222 164 0091 dB 188 088 0752 dB 354 184 1053 dB 325 079 060
results are highlighted in bold It can be observed fromTable 2that the MAE of OGABC is the smallest one among the threealgorithms for different noise level namely the accuracy ofOGABC is superior to MIWO and ABC
Then in order to further test the performance ofOGABCthe radar sea clutter power generated bymeasured refractivityprofile [20] is also utilized for the inversion of the surfaceduct Figure 8 shows the comparison of the inverted profileobtained by the three algorithms with the measured profileand it is distinct that the inverted profile of OGABC is inexcellent agreement with the measured one
In addition Figure 9 presents the corresponding com-parison of the convergence curves of the MIWO ABC andOGABC It is observed that the three algorithms have thefaster convergence rate at the early stage of iterations andthe minimum fitness is hardly changed with iteration at themiddle and later stage of iterationsHowever theOGABChasthe smallest minimum fitness among the three algorithms
8 International Journal of Antennas and Propagation
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 0dB
(a)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 1dB
(b)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 2dB
(c)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000M
ean
min
imum
fitn
ess
Iterations
MIWOABCOGABC
Noise level 3dB
(d)
Figure 6 The comparison of the convergence curves of different algorithms with the same noise level
That is to say the OGABC is superior to MIWO and ABCaccording to the accuracy and convergence rate
5 Conclusion
In this paper an improved ABC algorithm called OGABCis presented by simultaneously merging the OBL strategyand global best search equation into the standard ABCalgorithm to tackle its deficiency of slow convergence rateand falling into the local best during the process of inversionof atmospheric duct Taking the inversion of the surfaceduct using RFC technique as an example the propagation
characteristics of radar sea clutter obtained from the sim-ulated and measured refractivity profile are treated as theobserved sea clutter power to examine the performance ofOGABC respectively For the simulated refractivity profilecase the Gaussian noise is added to the simulated radarsea clutter power to investigate the stability of the proposedOGABC algorithm and the histograms and the convergencecurves are used to analyze the accuracy and convergencerate Further investigation using the radar sea clutter powergenerated by themeasured refractivity profile is also involvedand the accuracy and convergence rate of the algorithmsare discussed by comparing the inverted refractivity profilewith the measured one and analyzing their convergence
International Journal of Antennas and Propagation 9
60
120
180
0 20 40 60 80 100
40
20
0
minus20
minus40
40
20
0
minus20
minus40
Hei
ght (
m)
60
120
180
Hei
ght (
m)
60
120
180
Hei
ght (
m)
Distance (km)0 20 40 60 80 100
Distance (km)0 20 40 60 80 100
Distance (km)
5
0
minus5
MIWO ABC OGABC
(a)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
40
20
0
minus20
minus40
MIWO ABC OGABC
(b)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
minus40
40
20
0
minus20
40
20
0
minus20
minus40
minus60
MIWO ABC OGABC
(c)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
50
0
minus50
MIWO ABC OGABC
(d)
Figure 7 The comparison of the difference between the inverted and actual coverage diagram with different noise level (a) 0 dB (b) 1 dB(c) 2 dB and (d) 3 dB
290 295 300 305 310 315 3200
50
100
150
200
Hei
ght (
m)
M (M-units)
MIWOABC
OGABCMeasured profile
Figure 8 The comparison of the inverted profile with the measured profile
10 International Journal of Antennas and Propagation
0 20 40 60 80 100 1204000
5000
6000
7000
8000
Min
imum
fitn
ess
Iterations
MIWOABCOGABC
Figure 9 The comparison of the convergence curves of differentalgorithms
curves In addition the inversion results are also analyzed andcompared with those of the MIWO and ABC The obtainedresults verify that the proposed OGABC algorithm outper-forms MIWO and ABC in terms of stability accuracy andconvergence rate Future work will focus on the experimentalresearch and the improvement of the inversion model
Competing Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is supported by the National Science Fund forDistinguished Young Scholars of China (no 61225002) andthe Young Scientists Fund of the National Natural ScienceFoundation of China (no 61302050)
References
[1] C Yardim P Gerstoft andW S Hodgkiss ldquoEstimation of radiorefractivity from radar clutter using BayesianMonte Carlo anal-ysisrdquo IEEE Transactions on Antennas and Propagation vol 54no 4 pp 1318ndash1327 2006
[2] C Yardim P Gerstoft andW S Hodgkiss ldquoTracking refractiv-ity from clutter using Kalman and particle filtersrdquo IEEE Trans-actions on Antennas and Propagation vol 56 no 4 pp 1058ndash1070 2008
[3] C Yardim P Gerstoft andW S Hodgkiss ldquoSensitivity analysisand performance estimation of refractivity from clutter tech-niquesrdquo Radio Science vol 44 pp 1ndash16 2009
[4] P Gerstoft L T Rogers J L Krolik andW S Hodgkiss ldquoInver-sion for refractivity parameters from radar sea clutterrdquo RadioScience vol 38 pp 1ndash22 2003
[5] A Karimian C Yardim P Gerstoft W S Hodgkiss and AE Barrios ldquoRefractivity estimation from sea clutter an invitedreviewrdquo Radio Science vol 46 pp 1ndash16 2009
[6] X-F Zhao S-X Huang and H-D Du ldquoTheoretical analysisand numerical experiments of variational adjoint approach forrefractivity estimationrdquo Radio Science vol 46 no 1 pp 1ndash122011
[7] X F Zhao and S X Huang ldquoAtmospheric duct estimationusing radar sea clutter returns by the adjoint method withregularization techniquerdquo Journal of Atmospheric and OceanicTechnology vol 31 no 6 pp 1250ndash1262 2014
[8] R Douvenot V Fabbro P Gerstoft C Bourlier and J SaillardldquoA duct mapping method using least squares support vectormachinesrdquo Radio Science vol 43 pp 1ndash12 2008
[9] B Wang Z-S Wu Z-W Zhao and H-G Wang ldquoRetrievingevaporation duct heights from radar sea clutter using particleswarm optimization (PSO) algorithmrdquo Progress In Electromag-netics Research M vol 9 pp 79ndash91 2009
[10] X-F Zhao S-X Huang J Xiang andW-L Shi ldquoRemote sens-ing of atmospheric duct parameters using simulated annealingrdquoChinese Physics B vol 20 no 9 Article ID 099201 8 pages 2011
[11] C Yang ldquoEstimation of the atmospheric duct from radarsea clutter using artificial bee colony optimization algorithmrdquoProgress in Electromagnetics Research vol 135 pp 183ndash199 2013
[12] D Karaboga and B Basturk ldquoA powerful and efficient algo-rithm for numerical function optimization artificial bee colony(ABC) algorithmrdquo Journal of Global Optimization vol 39 no 3pp 459ndash471 2007
[13] J Yang W-T Li X-W Shi L Xin and J-F Yu ldquoA hybrid ABC-DE algorithm and its application for time-modulated arrayspattern synthesisrdquo IEEE Transactions on Antennas and Pro-pagation vol 61 no 11 pp 5485ndash5495 2013
[14] X Zhang X Zhang S Y Yuen S L Ho and W N Fu ldquoAnimproved artificial bee colony algorithm for optimal design ofelectromagnetic devicesrdquo IEEE Transactions on Magnetics vol49 no 8 pp 4811ndash4816 2013
[15] S K Goudos K Siakavara A Theopoulos E E Vafiadis andJ N Sahalos ldquoApplication of Gbest-guided artificial beecolonyalgorithm to passive UHF RFID tag designrdquo InternationalJournal of Microwave and Wireless Technologies 2015
[16] G Zhu and S Kwong ldquoGbest-guided artificial bee colonyalgorithm for numerical function optimizationrdquo Applied Math-ematics and Computation vol 217 no 7 pp 3166ndash3173 2010
[17] W-F Gao S-Y Liu and L-L Huang ldquoA novel artificialbee colony algorithm based on modified search equation andorthogonal learningrdquo IEEE Transactions on Cybernetics vol 43no 3 pp 1011ndash1024 2013
[18] R S Rahnamayan H R Tizhoosh and M M A SalamaldquoOpposition-based differential evolutionrdquo IEEETransactions onEvolutionary Computation vol 12 no 1 pp 64ndash79 2008
[19] A E Barrios ldquoTerrain parabolic equation model for propaga-tion in the troposphererdquo IEEE Transactions on Antennas andPropagation vol 42 no 1 pp 90ndash98 1994
[20] A Karimian C Yardim W S Hodgkiss P Gerstoft and A EBarrios ldquoEstimation of radio refractivity using a multiple angleclutter modelrdquo Radio Science vol 47 pp 1ndash9 2012
[21] A Basak D Maity and S Das ldquoA differential invasive weedoptimization algorithm for improved global numerical opti-mizationrdquo Applied Mathematics and Computation vol 219 no12 pp 6645ndash6668 2013
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of Antennas and Propagation 7
0 005 01 015 02 02505
101520
Occ
urre
nce
0 005 01 015 02 02505
101520
0 005 01 015 02 02505
101520
05
101520
Occ
urre
nce
05
101520
05
101520
36 38 40 42 44 4605
101520
Occ
urre
nce
36 38 40 42 44 4605
101520
36 38 40 42 44 4605
101520
15 20 25 300
5
10
Occ
urre
nce
MIWO ABC OGABC
15 20 25 300
5
10
15 20 25 300
5
10
minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2 minus34 minus32 minus3 minus28 minus26 minus24 minus22 minus2
c 1(M
-uni
tsm
)c 2
(M-u
nits
m)
h1
(m)
h2
(m)
Figure 5 The comparison of the histograms of the inversion results for different algorithms with the noise level of 3 dB
different noise levelThe convergence curves indicate that theOGABC is the best algorithm among the three algorithmsThis can be attributed to the fact that the OBL can produce arelatively excellent initial population in the initialization stageand a generation jumping to form a better population in thesearch process
The accuracy of the inversion of atmospheric duct is ofcrucial importance to exactly predict the marine electromag-netic environment Hence the comparisons of the differencebetween the inverted and actual radar coverage diagramsimulated by parabolic equation method for different noiselevel are shown in Figure 7 and the peaks of the parameterdistributions [20] in Figures 2ndash5 are treated as the invertedparameters of the surface duct to simulate their correspond-ing coverage diagram Nevertheless it is hardly possible toevaluate the quality of the algorithms according to the resultspresented in Figure 7 Thus a new quantitative evaluationcriterion named Mean Absolute Error (MAE) is given by
MAE =
sum119873119909
119894=1sum119873119910
119895=1
1003816100381610038161003816PL119894 (119894 119895) minus PL119886(119894 119895)
1003816100381610038161003816
119873119909119873119910
(13)
where PL119894(119894 119895) and PL
119886(119894 119895) represent the inverted and actual
propagation loss calculated at discrete point (119894Δ119909 119895Δ119910) and119873119909and 119873
119910are the sample points along the horizontal and
vertical direction respectively It is easy to see that the smallerthe MAE the higher the accuracy of the algorithm
The MAE of the three algorithms for different noise levelobtained by (13) are presented in Table 2 and the minimum
Table 2 The comparison of the MAE for different algorithms withdifferent noise level
Noise level AlgorithmsMIWO ABC OGABC
0 dB 222 164 0091 dB 188 088 0752 dB 354 184 1053 dB 325 079 060
results are highlighted in bold It can be observed fromTable 2that the MAE of OGABC is the smallest one among the threealgorithms for different noise level namely the accuracy ofOGABC is superior to MIWO and ABC
Then in order to further test the performance ofOGABCthe radar sea clutter power generated bymeasured refractivityprofile [20] is also utilized for the inversion of the surfaceduct Figure 8 shows the comparison of the inverted profileobtained by the three algorithms with the measured profileand it is distinct that the inverted profile of OGABC is inexcellent agreement with the measured one
In addition Figure 9 presents the corresponding com-parison of the convergence curves of the MIWO ABC andOGABC It is observed that the three algorithms have thefaster convergence rate at the early stage of iterations andthe minimum fitness is hardly changed with iteration at themiddle and later stage of iterationsHowever theOGABChasthe smallest minimum fitness among the three algorithms
8 International Journal of Antennas and Propagation
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 0dB
(a)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 1dB
(b)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 2dB
(c)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000M
ean
min
imum
fitn
ess
Iterations
MIWOABCOGABC
Noise level 3dB
(d)
Figure 6 The comparison of the convergence curves of different algorithms with the same noise level
That is to say the OGABC is superior to MIWO and ABCaccording to the accuracy and convergence rate
5 Conclusion
In this paper an improved ABC algorithm called OGABCis presented by simultaneously merging the OBL strategyand global best search equation into the standard ABCalgorithm to tackle its deficiency of slow convergence rateand falling into the local best during the process of inversionof atmospheric duct Taking the inversion of the surfaceduct using RFC technique as an example the propagation
characteristics of radar sea clutter obtained from the sim-ulated and measured refractivity profile are treated as theobserved sea clutter power to examine the performance ofOGABC respectively For the simulated refractivity profilecase the Gaussian noise is added to the simulated radarsea clutter power to investigate the stability of the proposedOGABC algorithm and the histograms and the convergencecurves are used to analyze the accuracy and convergencerate Further investigation using the radar sea clutter powergenerated by themeasured refractivity profile is also involvedand the accuracy and convergence rate of the algorithmsare discussed by comparing the inverted refractivity profilewith the measured one and analyzing their convergence
International Journal of Antennas and Propagation 9
60
120
180
0 20 40 60 80 100
40
20
0
minus20
minus40
40
20
0
minus20
minus40
Hei
ght (
m)
60
120
180
Hei
ght (
m)
60
120
180
Hei
ght (
m)
Distance (km)0 20 40 60 80 100
Distance (km)0 20 40 60 80 100
Distance (km)
5
0
minus5
MIWO ABC OGABC
(a)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
40
20
0
minus20
minus40
MIWO ABC OGABC
(b)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
minus40
40
20
0
minus20
40
20
0
minus20
minus40
minus60
MIWO ABC OGABC
(c)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
50
0
minus50
MIWO ABC OGABC
(d)
Figure 7 The comparison of the difference between the inverted and actual coverage diagram with different noise level (a) 0 dB (b) 1 dB(c) 2 dB and (d) 3 dB
290 295 300 305 310 315 3200
50
100
150
200
Hei
ght (
m)
M (M-units)
MIWOABC
OGABCMeasured profile
Figure 8 The comparison of the inverted profile with the measured profile
10 International Journal of Antennas and Propagation
0 20 40 60 80 100 1204000
5000
6000
7000
8000
Min
imum
fitn
ess
Iterations
MIWOABCOGABC
Figure 9 The comparison of the convergence curves of differentalgorithms
curves In addition the inversion results are also analyzed andcompared with those of the MIWO and ABC The obtainedresults verify that the proposed OGABC algorithm outper-forms MIWO and ABC in terms of stability accuracy andconvergence rate Future work will focus on the experimentalresearch and the improvement of the inversion model
Competing Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is supported by the National Science Fund forDistinguished Young Scholars of China (no 61225002) andthe Young Scientists Fund of the National Natural ScienceFoundation of China (no 61302050)
References
[1] C Yardim P Gerstoft andW S Hodgkiss ldquoEstimation of radiorefractivity from radar clutter using BayesianMonte Carlo anal-ysisrdquo IEEE Transactions on Antennas and Propagation vol 54no 4 pp 1318ndash1327 2006
[2] C Yardim P Gerstoft andW S Hodgkiss ldquoTracking refractiv-ity from clutter using Kalman and particle filtersrdquo IEEE Trans-actions on Antennas and Propagation vol 56 no 4 pp 1058ndash1070 2008
[3] C Yardim P Gerstoft andW S Hodgkiss ldquoSensitivity analysisand performance estimation of refractivity from clutter tech-niquesrdquo Radio Science vol 44 pp 1ndash16 2009
[4] P Gerstoft L T Rogers J L Krolik andW S Hodgkiss ldquoInver-sion for refractivity parameters from radar sea clutterrdquo RadioScience vol 38 pp 1ndash22 2003
[5] A Karimian C Yardim P Gerstoft W S Hodgkiss and AE Barrios ldquoRefractivity estimation from sea clutter an invitedreviewrdquo Radio Science vol 46 pp 1ndash16 2009
[6] X-F Zhao S-X Huang and H-D Du ldquoTheoretical analysisand numerical experiments of variational adjoint approach forrefractivity estimationrdquo Radio Science vol 46 no 1 pp 1ndash122011
[7] X F Zhao and S X Huang ldquoAtmospheric duct estimationusing radar sea clutter returns by the adjoint method withregularization techniquerdquo Journal of Atmospheric and OceanicTechnology vol 31 no 6 pp 1250ndash1262 2014
[8] R Douvenot V Fabbro P Gerstoft C Bourlier and J SaillardldquoA duct mapping method using least squares support vectormachinesrdquo Radio Science vol 43 pp 1ndash12 2008
[9] B Wang Z-S Wu Z-W Zhao and H-G Wang ldquoRetrievingevaporation duct heights from radar sea clutter using particleswarm optimization (PSO) algorithmrdquo Progress In Electromag-netics Research M vol 9 pp 79ndash91 2009
[10] X-F Zhao S-X Huang J Xiang andW-L Shi ldquoRemote sens-ing of atmospheric duct parameters using simulated annealingrdquoChinese Physics B vol 20 no 9 Article ID 099201 8 pages 2011
[11] C Yang ldquoEstimation of the atmospheric duct from radarsea clutter using artificial bee colony optimization algorithmrdquoProgress in Electromagnetics Research vol 135 pp 183ndash199 2013
[12] D Karaboga and B Basturk ldquoA powerful and efficient algo-rithm for numerical function optimization artificial bee colony(ABC) algorithmrdquo Journal of Global Optimization vol 39 no 3pp 459ndash471 2007
[13] J Yang W-T Li X-W Shi L Xin and J-F Yu ldquoA hybrid ABC-DE algorithm and its application for time-modulated arrayspattern synthesisrdquo IEEE Transactions on Antennas and Pro-pagation vol 61 no 11 pp 5485ndash5495 2013
[14] X Zhang X Zhang S Y Yuen S L Ho and W N Fu ldquoAnimproved artificial bee colony algorithm for optimal design ofelectromagnetic devicesrdquo IEEE Transactions on Magnetics vol49 no 8 pp 4811ndash4816 2013
[15] S K Goudos K Siakavara A Theopoulos E E Vafiadis andJ N Sahalos ldquoApplication of Gbest-guided artificial beecolonyalgorithm to passive UHF RFID tag designrdquo InternationalJournal of Microwave and Wireless Technologies 2015
[16] G Zhu and S Kwong ldquoGbest-guided artificial bee colonyalgorithm for numerical function optimizationrdquo Applied Math-ematics and Computation vol 217 no 7 pp 3166ndash3173 2010
[17] W-F Gao S-Y Liu and L-L Huang ldquoA novel artificialbee colony algorithm based on modified search equation andorthogonal learningrdquo IEEE Transactions on Cybernetics vol 43no 3 pp 1011ndash1024 2013
[18] R S Rahnamayan H R Tizhoosh and M M A SalamaldquoOpposition-based differential evolutionrdquo IEEETransactions onEvolutionary Computation vol 12 no 1 pp 64ndash79 2008
[19] A E Barrios ldquoTerrain parabolic equation model for propaga-tion in the troposphererdquo IEEE Transactions on Antennas andPropagation vol 42 no 1 pp 90ndash98 1994
[20] A Karimian C Yardim W S Hodgkiss P Gerstoft and A EBarrios ldquoEstimation of radio refractivity using a multiple angleclutter modelrdquo Radio Science vol 47 pp 1ndash9 2012
[21] A Basak D Maity and S Das ldquoA differential invasive weedoptimization algorithm for improved global numerical opti-mizationrdquo Applied Mathematics and Computation vol 219 no12 pp 6645ndash6668 2013
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
8 International Journal of Antennas and Propagation
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 0dB
(a)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 1dB
(b)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000
Mea
n m
inim
um fi
tnes
s
IterationsMIWOABCOGABC
Noise level 2dB
(c)
0 20 40 60 80 100 1200
1000
2000
3000
4000
5000
6000M
ean
min
imum
fitn
ess
Iterations
MIWOABCOGABC
Noise level 3dB
(d)
Figure 6 The comparison of the convergence curves of different algorithms with the same noise level
That is to say the OGABC is superior to MIWO and ABCaccording to the accuracy and convergence rate
5 Conclusion
In this paper an improved ABC algorithm called OGABCis presented by simultaneously merging the OBL strategyand global best search equation into the standard ABCalgorithm to tackle its deficiency of slow convergence rateand falling into the local best during the process of inversionof atmospheric duct Taking the inversion of the surfaceduct using RFC technique as an example the propagation
characteristics of radar sea clutter obtained from the sim-ulated and measured refractivity profile are treated as theobserved sea clutter power to examine the performance ofOGABC respectively For the simulated refractivity profilecase the Gaussian noise is added to the simulated radarsea clutter power to investigate the stability of the proposedOGABC algorithm and the histograms and the convergencecurves are used to analyze the accuracy and convergencerate Further investigation using the radar sea clutter powergenerated by themeasured refractivity profile is also involvedand the accuracy and convergence rate of the algorithmsare discussed by comparing the inverted refractivity profilewith the measured one and analyzing their convergence
International Journal of Antennas and Propagation 9
60
120
180
0 20 40 60 80 100
40
20
0
minus20
minus40
40
20
0
minus20
minus40
Hei
ght (
m)
60
120
180
Hei
ght (
m)
60
120
180
Hei
ght (
m)
Distance (km)0 20 40 60 80 100
Distance (km)0 20 40 60 80 100
Distance (km)
5
0
minus5
MIWO ABC OGABC
(a)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
40
20
0
minus20
minus40
MIWO ABC OGABC
(b)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
minus40
40
20
0
minus20
40
20
0
minus20
minus40
minus60
MIWO ABC OGABC
(c)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
50
0
minus50
MIWO ABC OGABC
(d)
Figure 7 The comparison of the difference between the inverted and actual coverage diagram with different noise level (a) 0 dB (b) 1 dB(c) 2 dB and (d) 3 dB
290 295 300 305 310 315 3200
50
100
150
200
Hei
ght (
m)
M (M-units)
MIWOABC
OGABCMeasured profile
Figure 8 The comparison of the inverted profile with the measured profile
10 International Journal of Antennas and Propagation
0 20 40 60 80 100 1204000
5000
6000
7000
8000
Min
imum
fitn
ess
Iterations
MIWOABCOGABC
Figure 9 The comparison of the convergence curves of differentalgorithms
curves In addition the inversion results are also analyzed andcompared with those of the MIWO and ABC The obtainedresults verify that the proposed OGABC algorithm outper-forms MIWO and ABC in terms of stability accuracy andconvergence rate Future work will focus on the experimentalresearch and the improvement of the inversion model
Competing Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is supported by the National Science Fund forDistinguished Young Scholars of China (no 61225002) andthe Young Scientists Fund of the National Natural ScienceFoundation of China (no 61302050)
References
[1] C Yardim P Gerstoft andW S Hodgkiss ldquoEstimation of radiorefractivity from radar clutter using BayesianMonte Carlo anal-ysisrdquo IEEE Transactions on Antennas and Propagation vol 54no 4 pp 1318ndash1327 2006
[2] C Yardim P Gerstoft andW S Hodgkiss ldquoTracking refractiv-ity from clutter using Kalman and particle filtersrdquo IEEE Trans-actions on Antennas and Propagation vol 56 no 4 pp 1058ndash1070 2008
[3] C Yardim P Gerstoft andW S Hodgkiss ldquoSensitivity analysisand performance estimation of refractivity from clutter tech-niquesrdquo Radio Science vol 44 pp 1ndash16 2009
[4] P Gerstoft L T Rogers J L Krolik andW S Hodgkiss ldquoInver-sion for refractivity parameters from radar sea clutterrdquo RadioScience vol 38 pp 1ndash22 2003
[5] A Karimian C Yardim P Gerstoft W S Hodgkiss and AE Barrios ldquoRefractivity estimation from sea clutter an invitedreviewrdquo Radio Science vol 46 pp 1ndash16 2009
[6] X-F Zhao S-X Huang and H-D Du ldquoTheoretical analysisand numerical experiments of variational adjoint approach forrefractivity estimationrdquo Radio Science vol 46 no 1 pp 1ndash122011
[7] X F Zhao and S X Huang ldquoAtmospheric duct estimationusing radar sea clutter returns by the adjoint method withregularization techniquerdquo Journal of Atmospheric and OceanicTechnology vol 31 no 6 pp 1250ndash1262 2014
[8] R Douvenot V Fabbro P Gerstoft C Bourlier and J SaillardldquoA duct mapping method using least squares support vectormachinesrdquo Radio Science vol 43 pp 1ndash12 2008
[9] B Wang Z-S Wu Z-W Zhao and H-G Wang ldquoRetrievingevaporation duct heights from radar sea clutter using particleswarm optimization (PSO) algorithmrdquo Progress In Electromag-netics Research M vol 9 pp 79ndash91 2009
[10] X-F Zhao S-X Huang J Xiang andW-L Shi ldquoRemote sens-ing of atmospheric duct parameters using simulated annealingrdquoChinese Physics B vol 20 no 9 Article ID 099201 8 pages 2011
[11] C Yang ldquoEstimation of the atmospheric duct from radarsea clutter using artificial bee colony optimization algorithmrdquoProgress in Electromagnetics Research vol 135 pp 183ndash199 2013
[12] D Karaboga and B Basturk ldquoA powerful and efficient algo-rithm for numerical function optimization artificial bee colony(ABC) algorithmrdquo Journal of Global Optimization vol 39 no 3pp 459ndash471 2007
[13] J Yang W-T Li X-W Shi L Xin and J-F Yu ldquoA hybrid ABC-DE algorithm and its application for time-modulated arrayspattern synthesisrdquo IEEE Transactions on Antennas and Pro-pagation vol 61 no 11 pp 5485ndash5495 2013
[14] X Zhang X Zhang S Y Yuen S L Ho and W N Fu ldquoAnimproved artificial bee colony algorithm for optimal design ofelectromagnetic devicesrdquo IEEE Transactions on Magnetics vol49 no 8 pp 4811ndash4816 2013
[15] S K Goudos K Siakavara A Theopoulos E E Vafiadis andJ N Sahalos ldquoApplication of Gbest-guided artificial beecolonyalgorithm to passive UHF RFID tag designrdquo InternationalJournal of Microwave and Wireless Technologies 2015
[16] G Zhu and S Kwong ldquoGbest-guided artificial bee colonyalgorithm for numerical function optimizationrdquo Applied Math-ematics and Computation vol 217 no 7 pp 3166ndash3173 2010
[17] W-F Gao S-Y Liu and L-L Huang ldquoA novel artificialbee colony algorithm based on modified search equation andorthogonal learningrdquo IEEE Transactions on Cybernetics vol 43no 3 pp 1011ndash1024 2013
[18] R S Rahnamayan H R Tizhoosh and M M A SalamaldquoOpposition-based differential evolutionrdquo IEEETransactions onEvolutionary Computation vol 12 no 1 pp 64ndash79 2008
[19] A E Barrios ldquoTerrain parabolic equation model for propaga-tion in the troposphererdquo IEEE Transactions on Antennas andPropagation vol 42 no 1 pp 90ndash98 1994
[20] A Karimian C Yardim W S Hodgkiss P Gerstoft and A EBarrios ldquoEstimation of radio refractivity using a multiple angleclutter modelrdquo Radio Science vol 47 pp 1ndash9 2012
[21] A Basak D Maity and S Das ldquoA differential invasive weedoptimization algorithm for improved global numerical opti-mizationrdquo Applied Mathematics and Computation vol 219 no12 pp 6645ndash6668 2013
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of Antennas and Propagation 9
60
120
180
0 20 40 60 80 100
40
20
0
minus20
minus40
40
20
0
minus20
minus40
Hei
ght (
m)
60
120
180
Hei
ght (
m)
60
120
180
Hei
ght (
m)
Distance (km)0 20 40 60 80 100
Distance (km)0 20 40 60 80 100
Distance (km)
5
0
minus5
MIWO ABC OGABC
(a)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
40
20
0
minus20
minus40
MIWO ABC OGABC
(b)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
minus40
40
20
0
minus20
40
20
0
minus20
minus40
minus60
MIWO ABC OGABC
(c)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
60
120
180
0 20 40 60 80 100
Hei
ght (
m)
Distance (km)
40
20
0
minus20
40
20
0
minus20
50
0
minus50
MIWO ABC OGABC
(d)
Figure 7 The comparison of the difference between the inverted and actual coverage diagram with different noise level (a) 0 dB (b) 1 dB(c) 2 dB and (d) 3 dB
290 295 300 305 310 315 3200
50
100
150
200
Hei
ght (
m)
M (M-units)
MIWOABC
OGABCMeasured profile
Figure 8 The comparison of the inverted profile with the measured profile
10 International Journal of Antennas and Propagation
0 20 40 60 80 100 1204000
5000
6000
7000
8000
Min
imum
fitn
ess
Iterations
MIWOABCOGABC
Figure 9 The comparison of the convergence curves of differentalgorithms
curves In addition the inversion results are also analyzed andcompared with those of the MIWO and ABC The obtainedresults verify that the proposed OGABC algorithm outper-forms MIWO and ABC in terms of stability accuracy andconvergence rate Future work will focus on the experimentalresearch and the improvement of the inversion model
Competing Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is supported by the National Science Fund forDistinguished Young Scholars of China (no 61225002) andthe Young Scientists Fund of the National Natural ScienceFoundation of China (no 61302050)
References
[1] C Yardim P Gerstoft andW S Hodgkiss ldquoEstimation of radiorefractivity from radar clutter using BayesianMonte Carlo anal-ysisrdquo IEEE Transactions on Antennas and Propagation vol 54no 4 pp 1318ndash1327 2006
[2] C Yardim P Gerstoft andW S Hodgkiss ldquoTracking refractiv-ity from clutter using Kalman and particle filtersrdquo IEEE Trans-actions on Antennas and Propagation vol 56 no 4 pp 1058ndash1070 2008
[3] C Yardim P Gerstoft andW S Hodgkiss ldquoSensitivity analysisand performance estimation of refractivity from clutter tech-niquesrdquo Radio Science vol 44 pp 1ndash16 2009
[4] P Gerstoft L T Rogers J L Krolik andW S Hodgkiss ldquoInver-sion for refractivity parameters from radar sea clutterrdquo RadioScience vol 38 pp 1ndash22 2003
[5] A Karimian C Yardim P Gerstoft W S Hodgkiss and AE Barrios ldquoRefractivity estimation from sea clutter an invitedreviewrdquo Radio Science vol 46 pp 1ndash16 2009
[6] X-F Zhao S-X Huang and H-D Du ldquoTheoretical analysisand numerical experiments of variational adjoint approach forrefractivity estimationrdquo Radio Science vol 46 no 1 pp 1ndash122011
[7] X F Zhao and S X Huang ldquoAtmospheric duct estimationusing radar sea clutter returns by the adjoint method withregularization techniquerdquo Journal of Atmospheric and OceanicTechnology vol 31 no 6 pp 1250ndash1262 2014
[8] R Douvenot V Fabbro P Gerstoft C Bourlier and J SaillardldquoA duct mapping method using least squares support vectormachinesrdquo Radio Science vol 43 pp 1ndash12 2008
[9] B Wang Z-S Wu Z-W Zhao and H-G Wang ldquoRetrievingevaporation duct heights from radar sea clutter using particleswarm optimization (PSO) algorithmrdquo Progress In Electromag-netics Research M vol 9 pp 79ndash91 2009
[10] X-F Zhao S-X Huang J Xiang andW-L Shi ldquoRemote sens-ing of atmospheric duct parameters using simulated annealingrdquoChinese Physics B vol 20 no 9 Article ID 099201 8 pages 2011
[11] C Yang ldquoEstimation of the atmospheric duct from radarsea clutter using artificial bee colony optimization algorithmrdquoProgress in Electromagnetics Research vol 135 pp 183ndash199 2013
[12] D Karaboga and B Basturk ldquoA powerful and efficient algo-rithm for numerical function optimization artificial bee colony(ABC) algorithmrdquo Journal of Global Optimization vol 39 no 3pp 459ndash471 2007
[13] J Yang W-T Li X-W Shi L Xin and J-F Yu ldquoA hybrid ABC-DE algorithm and its application for time-modulated arrayspattern synthesisrdquo IEEE Transactions on Antennas and Pro-pagation vol 61 no 11 pp 5485ndash5495 2013
[14] X Zhang X Zhang S Y Yuen S L Ho and W N Fu ldquoAnimproved artificial bee colony algorithm for optimal design ofelectromagnetic devicesrdquo IEEE Transactions on Magnetics vol49 no 8 pp 4811ndash4816 2013
[15] S K Goudos K Siakavara A Theopoulos E E Vafiadis andJ N Sahalos ldquoApplication of Gbest-guided artificial beecolonyalgorithm to passive UHF RFID tag designrdquo InternationalJournal of Microwave and Wireless Technologies 2015
[16] G Zhu and S Kwong ldquoGbest-guided artificial bee colonyalgorithm for numerical function optimizationrdquo Applied Math-ematics and Computation vol 217 no 7 pp 3166ndash3173 2010
[17] W-F Gao S-Y Liu and L-L Huang ldquoA novel artificialbee colony algorithm based on modified search equation andorthogonal learningrdquo IEEE Transactions on Cybernetics vol 43no 3 pp 1011ndash1024 2013
[18] R S Rahnamayan H R Tizhoosh and M M A SalamaldquoOpposition-based differential evolutionrdquo IEEETransactions onEvolutionary Computation vol 12 no 1 pp 64ndash79 2008
[19] A E Barrios ldquoTerrain parabolic equation model for propaga-tion in the troposphererdquo IEEE Transactions on Antennas andPropagation vol 42 no 1 pp 90ndash98 1994
[20] A Karimian C Yardim W S Hodgkiss P Gerstoft and A EBarrios ldquoEstimation of radio refractivity using a multiple angleclutter modelrdquo Radio Science vol 47 pp 1ndash9 2012
[21] A Basak D Maity and S Das ldquoA differential invasive weedoptimization algorithm for improved global numerical opti-mizationrdquo Applied Mathematics and Computation vol 219 no12 pp 6645ndash6668 2013
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
10 International Journal of Antennas and Propagation
0 20 40 60 80 100 1204000
5000
6000
7000
8000
Min
imum
fitn
ess
Iterations
MIWOABCOGABC
Figure 9 The comparison of the convergence curves of differentalgorithms
curves In addition the inversion results are also analyzed andcompared with those of the MIWO and ABC The obtainedresults verify that the proposed OGABC algorithm outper-forms MIWO and ABC in terms of stability accuracy andconvergence rate Future work will focus on the experimentalresearch and the improvement of the inversion model
Competing Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is supported by the National Science Fund forDistinguished Young Scholars of China (no 61225002) andthe Young Scientists Fund of the National Natural ScienceFoundation of China (no 61302050)
References
[1] C Yardim P Gerstoft andW S Hodgkiss ldquoEstimation of radiorefractivity from radar clutter using BayesianMonte Carlo anal-ysisrdquo IEEE Transactions on Antennas and Propagation vol 54no 4 pp 1318ndash1327 2006
[2] C Yardim P Gerstoft andW S Hodgkiss ldquoTracking refractiv-ity from clutter using Kalman and particle filtersrdquo IEEE Trans-actions on Antennas and Propagation vol 56 no 4 pp 1058ndash1070 2008
[3] C Yardim P Gerstoft andW S Hodgkiss ldquoSensitivity analysisand performance estimation of refractivity from clutter tech-niquesrdquo Radio Science vol 44 pp 1ndash16 2009
[4] P Gerstoft L T Rogers J L Krolik andW S Hodgkiss ldquoInver-sion for refractivity parameters from radar sea clutterrdquo RadioScience vol 38 pp 1ndash22 2003
[5] A Karimian C Yardim P Gerstoft W S Hodgkiss and AE Barrios ldquoRefractivity estimation from sea clutter an invitedreviewrdquo Radio Science vol 46 pp 1ndash16 2009
[6] X-F Zhao S-X Huang and H-D Du ldquoTheoretical analysisand numerical experiments of variational adjoint approach forrefractivity estimationrdquo Radio Science vol 46 no 1 pp 1ndash122011
[7] X F Zhao and S X Huang ldquoAtmospheric duct estimationusing radar sea clutter returns by the adjoint method withregularization techniquerdquo Journal of Atmospheric and OceanicTechnology vol 31 no 6 pp 1250ndash1262 2014
[8] R Douvenot V Fabbro P Gerstoft C Bourlier and J SaillardldquoA duct mapping method using least squares support vectormachinesrdquo Radio Science vol 43 pp 1ndash12 2008
[9] B Wang Z-S Wu Z-W Zhao and H-G Wang ldquoRetrievingevaporation duct heights from radar sea clutter using particleswarm optimization (PSO) algorithmrdquo Progress In Electromag-netics Research M vol 9 pp 79ndash91 2009
[10] X-F Zhao S-X Huang J Xiang andW-L Shi ldquoRemote sens-ing of atmospheric duct parameters using simulated annealingrdquoChinese Physics B vol 20 no 9 Article ID 099201 8 pages 2011
[11] C Yang ldquoEstimation of the atmospheric duct from radarsea clutter using artificial bee colony optimization algorithmrdquoProgress in Electromagnetics Research vol 135 pp 183ndash199 2013
[12] D Karaboga and B Basturk ldquoA powerful and efficient algo-rithm for numerical function optimization artificial bee colony(ABC) algorithmrdquo Journal of Global Optimization vol 39 no 3pp 459ndash471 2007
[13] J Yang W-T Li X-W Shi L Xin and J-F Yu ldquoA hybrid ABC-DE algorithm and its application for time-modulated arrayspattern synthesisrdquo IEEE Transactions on Antennas and Pro-pagation vol 61 no 11 pp 5485ndash5495 2013
[14] X Zhang X Zhang S Y Yuen S L Ho and W N Fu ldquoAnimproved artificial bee colony algorithm for optimal design ofelectromagnetic devicesrdquo IEEE Transactions on Magnetics vol49 no 8 pp 4811ndash4816 2013
[15] S K Goudos K Siakavara A Theopoulos E E Vafiadis andJ N Sahalos ldquoApplication of Gbest-guided artificial beecolonyalgorithm to passive UHF RFID tag designrdquo InternationalJournal of Microwave and Wireless Technologies 2015
[16] G Zhu and S Kwong ldquoGbest-guided artificial bee colonyalgorithm for numerical function optimizationrdquo Applied Math-ematics and Computation vol 217 no 7 pp 3166ndash3173 2010
[17] W-F Gao S-Y Liu and L-L Huang ldquoA novel artificialbee colony algorithm based on modified search equation andorthogonal learningrdquo IEEE Transactions on Cybernetics vol 43no 3 pp 1011ndash1024 2013
[18] R S Rahnamayan H R Tizhoosh and M M A SalamaldquoOpposition-based differential evolutionrdquo IEEETransactions onEvolutionary Computation vol 12 no 1 pp 64ndash79 2008
[19] A E Barrios ldquoTerrain parabolic equation model for propaga-tion in the troposphererdquo IEEE Transactions on Antennas andPropagation vol 42 no 1 pp 90ndash98 1994
[20] A Karimian C Yardim W S Hodgkiss P Gerstoft and A EBarrios ldquoEstimation of radio refractivity using a multiple angleclutter modelrdquo Radio Science vol 47 pp 1ndash9 2012
[21] A Basak D Maity and S Das ldquoA differential invasive weedoptimization algorithm for improved global numerical opti-mizationrdquo Applied Mathematics and Computation vol 219 no12 pp 6645ndash6668 2013
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of