Research Article QPSO-Based Adaptive DNA Computing Algorithmdownloads.hindawi.com/journals/tswj/2013/160687.pdf · of DNA computing algorithm are found using QPSO and used for the

Hindawi Publishing CorporationThe Scientific World JournalVolume 2013 Article ID 160687 8 pageshttpdxdoiorg1011552013160687

Research ArticleQPSO-Based Adaptive DNA Computing Algorithm

Mehmet Karakose1 and Ugur Cigdem2

1 Computer Engineering Department Firat University Elazig Turkey2 Computer Programming Department Gaziosmanpasa University Tokat Turkey

Correspondence should be addressed to Mehmet Karakose mkarakosefiratedutr

Received 3 May 2013 Accepted 20 June 2013

Academic Editors P Agarwal and Y Zhang

Copyright copy 2013 M Karakose and U CigdemThis is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

DNA (deoxyribonucleic acid) computing that is a new computation model based on DNA molecules for information storage hasbeen increasingly used for optimization and data analysis in recent years However DNA computing algorithmhas some limitationsin terms of convergence speed adaptability and effectiveness In this paper a new approach for improvement of DNA computing isproposedThis new approach aims to performDNA computing algorithmwith adaptive parameters towards the desired goal usingquantum-behaved particle swarm optimization (QPSO) Some contributions provided by the proposed QPSO based on adaptiveDNA computing algorithm are as follows (1) parameters of population size crossover rate maximum number of operationsenzyme and virus mutation rate and fitness function of DNA computing algorithm are simultaneously tuned for adaptive process(2) adaptive algorithm is performed using QPSO algorithm for goal-driven progress faster operation and flexibility in data and(3) numerical realization of DNA computing algorithm with proposed approach is implemented in system identification Twoexperiments with different systems were carried out to evaluate the performance of the proposed approach with comparativeresults Experimental results obtained with Matlab and FPGA demonstrate ability to provide effective optimization considerableconvergence speed and high accuracy according to DNA computing algorithm

1 Introduction

DNA computing is a new field of research which per-forms computing using the biomolecular structure of DNAmoleculesThefirst study performed in the field ofDNAcom-puting was the solution of the problem of traveling salesmencomposed of 7 cities by Adleman using real DNA molecules[1] The application was realized by creating solution envi-ronment in the biology laboratory and using biochemicalreactions The cities and distances that make up the problemof traveling salesmen were coded using DNA series and allways thatmight be solutionwere created using polymer chainreaction Upon the success of the study performed a newalgorithm for the solution of multivariable problems wasacquired In the continuation of this study Lipton solvedsatisfiability problem included in NP (nonpolynomial) prob-lem class using DNA computing in a similar way [2] Liptondemonstrated the application and that DNA computing canbe used for the solution of the problems containing logicalequations as well After this study Lin et al [3] performed the

design and optimization of PI (proportional integral) param-eters using DNA computing Ding and Ren used similarDNA computing algorithm with the abovementioned studyfor setting turbid inspecting parameters [4] Kim and Leeapplied DNA computing algorithm with a different methodfor setting the PID parameters [5] In the study performedDNA molecules were used in coding and setting the PIparameters The results of the application through computersimulation indicated that high success was acquired in thisfield Wang et al compared DNA computers and electroniccomputers and suggested that DNA computers were moreadvantageous [6] As a result of the applications givenabove DNA computing was developed rapidly and used inmany scientific studies [7ndash15] It has been used frequentlyparticularly in NP problems coloring problems of graphicsin setting the inspecting parameters arithmetic operationssignal processing problems and ciphering the data [16ndash24]DNA computing algorithm performs computing using thenatural characteristics of DNA molecules Those character-istics include parallel operations storing high amount of

2 The Scientific World Journal

data providing energy saving and having significant role incomputingThose characteristics are listed quite effectively inthe solution of complex and difficult problems In order to usethe DNA computingmore effectively scientists dealt with thedesign of total natural DNA computers composed of DNAmolecules [25]

In Section 2 of this study DNA computing algorithm isexplained in detail In Section 3 QPSO algorithm is men-tioned In Section 4 DNA computing is made applicable withQPSO algorithm parameters that increase the effectivenessof DNA computing algorithm are found using QPSO andused for the proposed algorithm And in the last section theacquired simulation results are given It is understood fromthe simulation results that the proposed method producesbetter values and is more successful

2 DNA Computing Algorithm

DNA computing is a new optimization algorithm performingcomputing using DNA molecules which store genetic infor-mation of the living things While performing the computingusing DNA DNA bases that make up the ground of the DNAmolecules are used DNA bases are Adenine (A) Guanine(G) Cytosine (C) and Thymine (T) and in this study theyhave been converted into numerical values including A-0G-1 C-2 and T-3 [9] Adenine and Thymine and Guanineand Cytosine complete each other [26ndash28] DNA moleculeshave a structure due to the double and triple hydrogen tiesamong those bases and in the computing process wherenumber of hydrogen ties is used In Particular the completionsituations of DNA bases and numbers of hydrogen ties areused in the computing performed in solution environmentsThere are 2 hydrogen ties between Adenine and Thymineand 3 hydrogen ties between Guanine and Cytosine [27]DNA molecules exist in the form of single serial and doublespiral serial Using the single DNA serials synthesis andreproduction of DNA molecules is realized Double spiralDNA serials are created according to the Watson-Crickcomplementation rule According to this rule Adenine andThymine and Guanine and Cytosine combine No emergenceis possible DNA molecules have many advantages includingperforming parallel operations storing high amount of dataand using those data for many years without any corruptionThese advantages are used sufficiently while performing thecomputing the algorithm produces better results In Figure 1the general structure of the DNA molecule was shown

Parallel computing provides a quick conclusion for oper-ations It is possible to complete the problems that requirevery big operation volumes and take a long time in thesolution with DNA molecules Some similar problems havemany variables and parameters Coding these variables andparameters and modeling the system generally require dataDNA molecules make great contributions to the solution ofthese problemswith their capability to store dataWhileDNAcomputing is applying the errors are made in the sequenceof DNA serials to affect the result negatively As a result ofDNAcomputing the values acquiredwith other optimizationalgorithm are not deemed to be final results The studies

A T C

GC

C

CC

PP

PP

PP

Phosphate

Hydrogen bounds

Deoxyribose

DNA bases

5998400 ndash3

998400di

rect

ion

3998400 ndash5

998400di

rect

ion

Figure 1 The general structure of the DNA molecule

performed on DNA computing have been implemented intwo different manners The most commonly used methodamong these is the computing performed in solution envi-ronment [26ndash28] The cost of performing the computing insolution environment is more because machines are neededto produce DNA serials and DNA synthesis Furthermoresolution should be prepared and solution tubes should beused Systems named gel electrophoresis are needed foracquiring and analyzing the results [22 23] And anotheralgorithm used for DNA computing is the numerical DNAcomputing algorithm In this algorithm DNA serials arecreated in electronic computers and they are convertedinto numerical values for computing Although the cost ofcomputing performed electronically is less the mentionedadvantages of DNAmolecules cannot be applied on problemscompletely because when the DNA molecules are used insolution environment they only can use the abovementionedcharacteristics

Numerical DNA computing algorithm is similar to thegenetic algorithms however it is different from geneticalgorithms It uses A T G and C bases rather than dualnumber of the systems and the solution set of the problems iscomposed of these bases [29] Furthermore DNA computinghas twonewmutation operationsThese are enzyme and virusmutations and provide great advantage while computing isperformed Enzyme mutation is the operation of deletion inone or more DNA parts from any DNA serial And virusmutation is the operation of adding one or more DNA partsto any DNA serial [9 19 20] Enzyme and virus mutationsprovide continuous renewal of the population and preventfocusing on local optimum points thus give the algorithma global search capacity In Table 1 the conversion of DNAserials into numerical data has been given

3 QPSO Algorithm

QPSO is an algorithmhaving the capacity of global searchingThis algorithm has been developed inspired by living thingsthat move as masses including insects bees and were used forthe solution of many problems in the literature In the QPSOalgorithmwhich is a population-based algorithm individualsact together and compose the masses While the speed ofpopulation is used in PSO algorithm the next position of the

The Scientific World Journal 3

Table 1 Coding table of DNA computing algorithm

DNA codingsystem

Quartet codingsystem

Binary codingsystem Decimal coding system

Minimumvalue of119870119901 and 119870119894

Maximumvalue of119870119901 and 119870119894A G C T 0 1 2 3 00 01 10 11

119870119901 AGT TAA TTT 013 300 333 000111 110000111111

0 lowast 16 + 1 lowast 4 + 3 lowast 1 = 73 lowast 16 + 0 lowast 4 + 0 lowast 1 = 483 lowast 16 + 3 lowast 4 + 3 lowast 1 = 63

0 63

119870119894

GTT CCAAGC 133 220 012 011111 101000

000110

1 lowast 1 + 3 lowast 14 + 3 lowast 116 = 193752 lowast 1 + 2 lowast 14 lowast 0 lowast 116 = 250 lowast 1 + 1 lowast 14 + 2 lowast 116 = 0375

0 4

population is determined by taking the average of the bestresults acquired by the particles that make up the populationin QPSO locally In order to update the speeds and positionsof the particles in QPSO algorithm (1) (2) and (3) are used[30]

While preparing the algorithm the best informationacquired by each particle (pbest) and the best informationrealized by the mass (119892best = min (119901best)) variables are usedUsing these two variables the movement point of the mass 119875value is calculated using (1) The Q1 and Q2 variables in (1)represent random numbers produced in the interval 0-1 Indetermining the next position of the mass QPSO algorithmuses the variable ofmbest which is the best arithmetic averageof mass instead of gbest mbest value is calculated using (2)In (2) n represents the population size After computing thembest value for the next step of the population the 119909 value iscomputed by using (3) The 120573 variable in (3) is a value to bedetermined by the applier And the 119906 variable is another onewhich takes value at the interval of 0-1

119901 =(1198761 lowast 119901best + 1198762 lowast 119892best)

1198761 + 1198762 (1)

119898best =(sum119899

119898=1119901best119894)119899 (2)

119909 (119894 + 1) = 119901 plusmn 120573 sdot |119898best minus 119909 (119894)| sdot ln(1119906) (3)

4 QPSO-Based Adaptive DNAComputing Algorithm

In order to increase the global search capacity and effec-tiveness of DNA computing algorithm in problem solvingparameters should be applicable The size of the populationsuitable for this algorithm (Pb) maximumoperation number(N) Crossover rate (Mu) enzyme proportion (E) and virusproportion (V) affect the local and global search capacity ofthe algorithm positively Making all these parameters adapt-able simultaneously can sometimes give negative results Thereal target aimed by adapting the parameters is to increasethe diversity in the population and prevent the focusing onlocal optimum points Furthermore the adaptable algorithmscan easily be applied for the solution of many optimizationproblems In this study it is aimed to find the optimum valuesof the parameters considering the conformity values acquiredby the population in all iterations of the algorithm Using the

equation depending on conformity function together withthe QPSO algorithm the parameter values are determinedThe acquired values are taken as optimized parameter valueswhen optimum solution is achievedThese parameters whichare fixed at the beginning are made dynamic and increasingthe effectiveness of the algorithm is aimed In Figure 2the steps of QPSO based on numerical DNA computingalgorithm are shown

As shown in Figure 3 the proposed approach can bemodeled to find the parameters values of DNA computingalgorithm using a QPSO algorithm In the proposed modelvariables of 119870119901 and 119870119894 are coded with DNA serials at thelength of 3 bases119870119901 and119870119894 values are created randomly usingA G C and T bases and converted into numerical data using0 1 2 and 3 values respectively The serials are convertedinto first system of 4 then system of 10 and used in numericalenvironmentThe parameter values used formaking adaptiveDNA computing are explained as follows

Population Size (Pb) Population size is one of the mostsignificant characteristics that contribute to the solutionof the problem The size of the problem and its complexstructure are taken into consideration while determiningthe value of those parameters When the population size isincreased a broader solution area is created The time ofsolution gets longer Furthermore unnecessary expansion ofthe solution area causes big time losses and the solution timefor the problem increases rapidly For this reason optimumvalue should be selected in the applications In setting of thepopulation size (4) is used The 119891 variable given in (4) isthe conformity function used for the algorithm 119891(119899) givesthe conformity value of the nth element of the populationand 119891(1) gives the conformity value of the first element ofthe population Max(119891) gives the maximum value of thepopulation and min(119891) gives the minimum value Thesetwo values provide information about the distribution ofthe values belonging to population elements The number 15given in (4) is the base value and it is used for the purpose ofthe probability of the size of the population to be zero It isdetermined with the method of trial and error

Pop size =(max (119891) minusmin (119891))(119891 (119899) minus 119891 (1))

+ 15 (4)

Maximum Number of Operations (N) Maximum numberof operations gives the number of processes of the cycle


The creation of the firstpopulation with DNA

value

Apply virusmutation

Stop

Determine the optimalvalue

Crossover

QPSO

Pb

E

V

Mu

Is maximumgeneration

No

Yes

a b

converted intonumerical values

DNA sequences are

mutationApply enzyme

Calculate the fitness

N

(a)

System++

1

S

DNAcomputing

Pb E VMu

+minus

e u y

X

X

KpKir

N a b

QPSO algorithm

(b)

Figure 2 Block diagram of the proposed approach

PI controller SystemRef

QPSO algorithmDNA computingalgorithm

Adaptive algorithm

Figure 3 Schematic diagram of the model

in the DNA computing algorithm When the maximumnumber of operations is increased the completing time ofthe cycles is extended and the number of the performedoperations increases In Particular the number of elementsthat composes the population and the number of the variablesused in the problem are big the completion time of thealgorithm increases rapidly and this causes loss of timeand economic losses Keeping the maximum number ofoperations in the lowest level decreases the efficiency of thealgorithmThe best values should be determined considering

all those situations In this study in order to find maximumnumber of operations (5) has been usedThe number 5 givenin (5) is the base value and it is used for the purpose of theprobability of the size of the maximum number of operationsto be zero and it is determined with the method of trial anderror

Maximum generation number = (119891 (119899) + 119891 (1)) lowast Pb + 5(5)

Crossover Rate (Mu) With Crossover new elements arecreated by using the existing population elements Whiledetermining the Crossover rate the number and selectionof the elements to be crossed are taken into account Whenthe Crossover rate is kept high the number of elements tobe selected increases and the change of the set that shall becreated newly is providedHowever it is probable that the newcreated elements may be values close to the parent elementsFurthermore the number of elements to be crossed beinghigh prolongs the duration of completion of the algorithmIn this study (6) has been used for the purpose of findingthe Crossover rate When (6) is used crossover rate in all


Table 2 Comparison results of DNA and adaptive DNA computingalgorithms

Algorithm 119870119901 119870119894 Settling time Overshoot DNA-PI 30 01250 008 14Adaptive DNA-PI 17 04375 008 sim0

Table 3 Values of DNA and adaptive DNA computing parameters

Parameters DNA computing Adaptive DNAcomputing

Population size 80 60Maximum generations 20 40Crossover rate 100 32Enzyme and virus rate 30 12a and b 15 12 118203 19700

iterations is applied at a proportion of 80

Crossover rate = Pop size lowast 08 (6)

Enzyme (E) and Virus (V) Proportion The most significanttwo parameters used in this study are enzyme and virusmutations While enzyme mutation provides the deletion ofa certain proportion of elements from the population virusmutation provides the addition of a certain proportion ofelements to the population These two parameters are onlyused with DNA computing unlike others The implementingadaptive DNA computing enzyme and virus proportionsshould be selected well because when the enzyme mutationis not applied in the correct time it may lead to the loss ofelements with good conformity values One should be carefulthat the algorithm does not create unnecessary solution areawhile implementing virus mutation Equation (7) has beenused in the study performed in order to determine enzymeand virus proportions With the change of the populationsize the variation of population at a proportion of 30 isprovided in ever iteration In the selection of the equationsfound through the trial and error method were taken intoconsideration and the values acquired by running the systemmany times were examined in detail and the equations havebeen given their final forms

e v = Pop size lowast 03 (7)

The algorithmic steps of DNA computing algorithm areexplained in detail as follows

Step 1 The first population is created with DNA serialsrandomly

Step 2 The serials created were firstly converted into systemof four and then into decimal system as it is explained inTable 3 and converted into numerical data The operations

of coding and converting are given in Table 3 in detail Thepopulation elements converted into numerical data are sentto the simulation environment and the error values createdin the system are determined

Step 3 The error values from the simulation are placed in theconformity function and conformity values are obtainedThepopulation elements that minimize sum conformity valuesare used as the new 119870119901 and 119870119894 values Equation (8) hasbeen selected for the determination of the conformity valueThis conformity function may change in accordance with thepreference of the applier as it is explained above

Step 4 The adaptive parameter values to be used for theadaptive DNA computing algorithm and cycle are foundusing QPSO and sent to the system Equation (9) is usedfor the QPSO conformity function In the selection of theconformity function the results found through themethod oftrial and error were taken into consideration and the valuesacquired by running the system many times were examinedin detail and the equations have been given their final forms

Step 5 The elements of the population created firstly aresubjected toCrossover process andnewpopulation is createdUsing Steps 2 and 3 new 119870119901 and119870119894 are determined

Step 6 The population applied enzyme and virus mutationand new population is created Applying Steps 2 3 and 4 new119870119901 and119870119894 values are determined

Step 7 The best conformity values found in the above stepsare compared 119870119901 and 119870119894 values in the step where theconformity value is minimum are sent to the system

Step 8 Those steps are applied till the maximum number ofoperations is achieved At the end of maximum operationsthe most suitable 119870119901 and119870119894 values are determined

119891 (119870119901 119870119894) = sum (abs (119890)) (8)

119891 qpso = 119886 lowast sum (1198902) + 119887 lowastmax (yout) (9)

119886 = average (119891 qpso) lowast 1198881 (10)


5 Experimental Results

A transfer function given with equation (12) that is modelingposition control of a DCmotor was used for the performancemeasurement of the proposed method DNA computingalgorithm was applied for setting the PI parameters andMatlab m-file was written For finding the simulation resultsthe model given in Figure 3 was created in the Simulinkenvironment and the results were obtained The parametervalues used for DNA computing are given in Table 3

119866 (119904) =22

896119890 minus 61199043 + 727119890 minus 31199042 + 0945119904 (12)


0 05 1 15 2 25 30

02

04

06

08

1

12

14

Time (s)

Fitn

ess v

alue

RefferanceDNA-PI

Adaptive DNA-PI

Figure 4 Adaptive algorithm results

0 5 10 15 20 25 30 35 40minus58297

minus58296

minus58295

minus58294

minus58293

minus58292

minus58291

minus5829

Iterations

Fitn

ess v

alue

Figure 5The fitness values of adaptive DNA computing algorithm

In the application performed QPSO-based DNA com-puting algorithm was used for the optimization of the PIparameters Although various conformity functions wereused in the optimization of the PI parameters in this studythe sum of absolute value of error was selected as theconformity function With the use of this function whichis sensitive even to the errors with minimal values it wastargeted that upper excess increase and setting times wouldgive better results In the MatlabSimulink study performedthe size of the population used in the application was takenas 80 maximum number of operations as 20 and referencevalue as 1 For the detection of 119870119901 and 119870119894 values (8) hasbeen used as the conformity function While performingcoding with DNA computing algorithm 119870119901 and 119870119894 valueswere coded with DNA basis using data of 6 bytes Firstly 80individuals in the population were used and each individualwas representedwith data of 12 bytesThe first 6 bytes of thosedata of 12 bytes were used for 119870119901 and the other 6 bytes wereused for 119870119894 In every iteration enzyme and virus mutation asmuch as 30 of the population was applied and the changeof the individual was provided as much as population sizelowast 03 and the population was renewed In the applicationperformed the population elements created in the algorithmswere sent to the system and determination of the PI elementshas been provided The results produced by the system as a

0 5 10 15 20 25 300

2

4

6

8

10

12

14

Time (ms)

Resp

onse

syste

m

DNAGA

QPSOYBS

times103

(a)

0 5 10 15 20 25 300

1

2

3

4

5

6

7

8

9

10times104

Time (ms)

Resp

onse

syste

m

DNAGA

QPSOYBS

(b)

Figure 6 Comparison results

result of running the programmany times are given in Table 3and Figure 4

As it is given in Table 2 using the adaptive DNA com-puting algorithm 119870119901 is found to be 17 119870119894 04375 placementtime 008 seconds and maximum excess approximately 0In Figure 4 the comparison of the results foundwith adaptiveDNA computing to DNA computing is given The maxi-mum excess value found with DNA computing algorithmmade adaptive is approximately 0 while the maximumexcess value found with DNA computing is computed asapproximately 14Those results indicate that adaptive DNAcomputing gets better results


In Table 3 DNA computing parameters made adaptivewith the QPSO algorithm and DNA computing parametersare given With the activation of the algorithm with QPSOparameter values have changed With the QPSO algorithmpopulation size was revised as 60 maximum number ofoperations as 40Crossover rate as 32 and enzyme and virusproportion as 12 New values were applied by being sent toDNA computing algorithm The values of 119886 and 119887 used forthe conformity function were 15 and 12 when the algorithmstarted and with the QPSO algorithm they were acquired as118203 and 19700 and used for the systemThe performanceof the proposed approach and comparison results are shownin Figures 5 and 6 respectively As shown in these figures theproposed approach finds faster than other algorithms

6 Conclusions

DNA computing that the key advantage is parallelism hasthe natural capabilities of DNA and large information storageabilities But this computation model has some problemsfor numerical implementation in real world applications Inparticular the parameters of the DNA computing algorithmare determined by trial and error in the literature Thisdetermination method is usually very difficult inefficientand inflexible A newQPSO-based adaptive DNA computingalgorithm is proposed and implemented for optimizationproblems in this study The seven parameters of DNAcomputing algorithm has been simultaneously and onlinetuned with QPSO for adaptive operation in the proposedapproach Experimental results obtained with computationof parameters of PI controller for system identificationapplication consistently show that proposed approach haseffective optimization high accuracy and high convergencespeed according to traditional DNA computing algorithmIn addition general search capability of the method is notdependent on local optimum points and applicability ofthe proposed approach is easier simpler and faster thantraditional DNA computing algorithm

References

[1] L M Adleman ldquoMolecular computation of solutions to com-binatorial problemsrdquo Science vol 266 no 5187 pp 1021ndash10241994

[2] R J Lipton ldquoUsing DNA to solve NP-complete problemsrdquoScience vol 268 no 5210 pp 542ndash545 1995

[3] C-L Lin H-Y Jan and T-S Hwang ldquoStructure variable PIDcontrol design based on DNA coding methodrdquo in Proceedingsof the IEEE International Symposium on Industrial Electronics(IEEE-ISlE rsquo04) pp 423ndash428 May 2004

[4] Y Ding and L Ren ldquoDNA genetic algorithm for design ofthe generalized membership-type Takagi-Sugeno fuzzy controlsystemrdquo in Proceedings of the IEEE International Conference onSystems Man and Cybernetics vol 5 pp 3862ndash3867 October2000

[5] J J Kim and J J Lee ldquoPID controller design using double helixstructured DNA algorithms with a recovery functionrdquoArtificialLife and Robotics vol 12 no 1-2 pp 241ndash244 2008

[6] Y Wang G Cui and Z Wang ldquoResearch on DNA computerwith coprocessor organizational modelrdquo in Proceedings of the8th International Conference on High-Performance Computingin Asia-Pacific Region (HPC Asia rsquo05) vol 6 pp 544ndash549December 2005

[7] X Zhang Y Niu and Y Wang ldquoDNA computing in microre-actors a solution to the minimum vertex cover problemrdquo inProceedings of the 6th International Conference on Bio-InspiredComputingTheories andApplications (BIC-TA rsquo11) pp 236ndash240September 2011

[8] R Sridhar and S Balasubramaniam ldquoGIS integratedDNA com-puting for solving travelling salesman problemrdquo in Proceedingsof the IEEE Symposium on Computers amp Informatics (ISCI rsquo11)pp 402ndash406 mys March 2011

[9] U Cigdem and M Karakose ldquoUsing of DNA computing fortuning of parameters of PI and fuzzy controllersrdquo in Proceedingsof the Automatic Control National Symposium pp 665ndash670Izmir Turkey 2011

[10] F Li Z Li and J Xu ldquoDNA computing model based on photo-electric detection system with magnetic beadsrdquo in Proceedingsof the 6th International Conference on Bio-Inspired ComputingTheories and Applications (BIC-TA rsquo11) pp 170ndash175 September2011

[11] Y Huang and L He ldquoDNA computing research progress andapplicationrdquo in Proceedings of the 6th International Conferenceon Computer Science and Education (ICCSE rsquo11) pp 232ndash235August 2011

[12] J Xu X Qiang Y Yang et al ldquoAn unenumerative DNA com-puting model for vertex coloring problemrdquo IEEE Transactionson Nanobioscience vol 10 no 2 pp 94ndash98 2011

[13] S Mitra R Das and Y Hayashi ldquoGenetic networks and softcomputingrdquo IEEEACMTransactions on Computational Biologyand Bioinformatics vol 8 no 1 pp 94ndash107 2011

[14] H Jiao Y Zhong L Zhang and P Li ldquoUnsupervised remotesensing image classificationusing an artificialDNAcomputingrdquoin Proceedings of the 2011 7th International Conference onNatural Computation (ICNC rsquo11) pp 1341ndash1345 July 2011

[15] Z Yin C Hua and B Song ldquoPlasmid DNA computing modelof 0-1 programming problemrdquo in Proceedings of the IEEE 5thInternational Conference on Bio-Inspired Computing Theoriesand Applications (BIC-TA rsquo10) pp 148ndash151 September 2010

[16] J Xu X Qiang K Zhang et al ldquoA parallel type of DNAcomputing model for graph vertex coloring problemrdquo in Pro-ceedings of the IEEE 5th International Conference on Bio-InspiredComputingTheories and Applications (BIC-TA rsquo10) pp 231ndash235September 2010

[17] Y Huang Z Yin and Y Tian ldquoDesign of PID controller basedon DNA computingrdquo in Proceedings of the International Con-ference on Artificial Intelligence and Computational Intelligence(AICI rsquo10) pp 195ndash198 October 2010

[18] Z Yin B Song C Zhen and C Hua ldquoMolecular beacon-based DNA computing model for maximum independent setproblemrdquo in Proceedings of the International Conference onIntelligent Computation Technology and Automation (ICICTArsquo10) pp 732ndash735 May 2010

[19] C-L Lin H-Y Jan and T-S Hwang ldquoStructure variable PIDcontrol design based onDNA codingmethodrdquo in Proceedings ofthe IEEE International Symposium on Industrial Electronics pp423ndash428 May 2004


[20] C-L Lin H-Y Jan and T-H Huang ldquoSelf-organizing PIDcontrol design based on DNA computing methodrdquo in Proceed-ings of the IEEE International Conference on Control Applica-tions pp 568ndash573 September 2004

[21] C V Henkel Experimental DNA computing [PhD thesis]Leiden University 2005

[22] Z F Qiu Advance the DNA computing [PhD thesis] TexasAampM University College Station Tex USA 2003

[23] M O Rahman A Hussain E Scavino M A Hannan and HBasri ldquoObject identification using DNA computing algorithmrdquoin Proceedings of the IEEE Congress on Evolutionary Computa-tion pp 1ndash7 2012

[24] J M Chaves-Gonzalez and M A Vega-Rodrıguez ldquoDNAsequence design for reliable DNA computing by using amultiobjective approachrdquo in Proceedings of the 13th IEEEInternational Symposium on Computational Intelligence andInformatics pp 73ndash79 2012

[25] H Zhang and X Liu ldquoA general object oriented description forDNA computing techniquerdquo in Proceedings of the InternationalConference on Information Technology and Computer Science(ITCS rsquo09) pp 3ndash6 July 2009

[26] H Jiao Y Zhong and L Zhang ldquoArtificial DNA computing-based spectral encoding andmatching algorithm for hyperspec-tral remote sensing datardquo IEEE Transactions on Geoscience andRemote Sensing vol 50 no 10 pp 4085ndash4105 2012

[27] W Liu S Sun and Y Guo ldquoA DNA computing model ofperceptronrdquo in Proceedings of the Pacific-Asia Conference onCircuits Communications and System (PACCS rsquo09) pp 710ndash712May 2009

[28] Q Zhang X Xue and X Wei ldquoA novel image encryptionalgorithm based onDNA subsequence operationrdquoThe ScientificWorld Journal vol 2012 Article ID 286741 10 pages 2012


[30] M Xi J Sun and W Xu ldquoParameter optimization of PID con-troller based on Quantum-Behaved Particle Swarm Optimiza-tion Algorithmrdquo Complex Systems and Applications-Modellingvol 14 supplement 2 pp 603ndash607 2007

Submit your manuscripts athttpwwwhindawicom

Computer Games Technology

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Distributed Sensor Networks


Advances in

FuzzySystems

Hindawi Publishing Corporationhttpwwwhindawicom

Volume 2014


ReconfigurableComputing

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014


Applied Computational Intelligence and Soft Computing

thinspAdvancesthinspinthinsp

Artificial Intelligence

HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014

Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Electrical and Computer Engineering

Journal of

Journal of

Computer Networks and Communications


Hindawi Publishing Corporation

httpwwwhindawicom Volume 2014

Advances in

Multimedia


Biomedical Imaging


ArtificialNeural Systems

Advances in


RoboticsJournal of



Computational Intelligence and Neuroscience

Industrial EngineeringJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Human-ComputerInteraction

Advances in

Computer EngineeringAdvances in



data providing energy saving and having significant role incomputingThose characteristics are listed quite effectively inthe solution of complex and difficult problems In order to usethe DNA computingmore effectively scientists dealt with thedesign of total natural DNA computers composed of DNAmolecules [25]

In Section 2 of this study DNA computing algorithm isexplained in detail In Section 3 QPSO algorithm is men-tioned In Section 4 DNA computing is made applicable withQPSO algorithm parameters that increase the effectivenessof DNA computing algorithm are found using QPSO andused for the proposed algorithm And in the last section theacquired simulation results are given It is understood fromthe simulation results that the proposed method producesbetter values and is more successful

2 DNA Computing Algorithm

DNA computing is a new optimization algorithm performingcomputing using DNA molecules which store genetic infor-mation of the living things While performing the computingusing DNA DNA bases that make up the ground of the DNAmolecules are used DNA bases are Adenine (A) Guanine(G) Cytosine (C) and Thymine (T) and in this study theyhave been converted into numerical values including A-0G-1 C-2 and T-3 [9] Adenine and Thymine and Guanineand Cytosine complete each other [26ndash28] DNA moleculeshave a structure due to the double and triple hydrogen tiesamong those bases and in the computing process wherenumber of hydrogen ties is used In Particular the completionsituations of DNA bases and numbers of hydrogen ties areused in the computing performed in solution environmentsThere are 2 hydrogen ties between Adenine and Thymineand 3 hydrogen ties between Guanine and Cytosine [27]DNA molecules exist in the form of single serial and doublespiral serial Using the single DNA serials synthesis andreproduction of DNA molecules is realized Double spiralDNA serials are created according to the Watson-Crickcomplementation rule According to this rule Adenine andThymine and Guanine and Cytosine combine No emergenceis possible DNA molecules have many advantages includingperforming parallel operations storing high amount of dataand using those data for many years without any corruptionThese advantages are used sufficiently while performing thecomputing the algorithm produces better results In Figure 1the general structure of the DNA molecule was shown

Parallel computing provides a quick conclusion for oper-ations It is possible to complete the problems that requirevery big operation volumes and take a long time in thesolution with DNA molecules Some similar problems havemany variables and parameters Coding these variables andparameters and modeling the system generally require dataDNA molecules make great contributions to the solution ofthese problemswith their capability to store dataWhileDNAcomputing is applying the errors are made in the sequenceof DNA serials to affect the result negatively As a result ofDNAcomputing the values acquiredwith other optimizationalgorithm are not deemed to be final results The studies

A T C

GC

C

CC

PP

PP

PP

Phosphate

Hydrogen bounds

Deoxyribose

DNA bases

5998400 ndash3

998400di

rect

ion

3998400 ndash5

998400di

rect

ion

Figure 1 The general structure of the DNA molecule

performed on DNA computing have been implemented intwo different manners The most commonly used methodamong these is the computing performed in solution envi-ronment [26ndash28] The cost of performing the computing insolution environment is more because machines are neededto produce DNA serials and DNA synthesis Furthermoresolution should be prepared and solution tubes should beused Systems named gel electrophoresis are needed foracquiring and analyzing the results [22 23] And anotheralgorithm used for DNA computing is the numerical DNAcomputing algorithm In this algorithm DNA serials arecreated in electronic computers and they are convertedinto numerical values for computing Although the cost ofcomputing performed electronically is less the mentionedadvantages of DNAmolecules cannot be applied on problemscompletely because when the DNA molecules are used insolution environment they only can use the abovementionedcharacteristics

Numerical DNA computing algorithm is similar to thegenetic algorithms however it is different from geneticalgorithms It uses A T G and C bases rather than dualnumber of the systems and the solution set of the problems iscomposed of these bases [29] Furthermore DNA computinghas twonewmutation operationsThese are enzyme and virusmutations and provide great advantage while computing isperformed Enzyme mutation is the operation of deletion inone or more DNA parts from any DNA serial And virusmutation is the operation of adding one or more DNA partsto any DNA serial [9 19 20] Enzyme and virus mutationsprovide continuous renewal of the population and preventfocusing on local optimum points thus give the algorithma global search capacity In Table 1 the conversion of DNAserials into numerical data has been given

3 QPSO Algorithm

QPSO is an algorithmhaving the capacity of global searchingThis algorithm has been developed inspired by living thingsthat move as masses including insects bees and were used forthe solution of many problems in the literature In the QPSOalgorithmwhich is a population-based algorithm individualsact together and compose the masses While the speed ofpopulation is used in PSO algorithm the next position of the



DNA codingsystem





119870119901 AGT TAA TTT 013 300 333 000111 110000111111


0 63

119870119894

GTT CCAAGC 133 220 012 011111 101000

000110


0 4




1198761 + 1198762 (1)

119898best =(sum119899

119898=1119901best119894)119899 (2)








+ 15 (4)




value

Apply virusmutation

Stop


Crossover

QPSO

Pb

E

V

Mu


No

Yes

a b


DNA sequences are



N

(a)

System++

1

S

DNAcomputing

Pb E VMu

+minus

e u y

X

X

KpKir

N a b

QPSO algorithm

(b)




Adaptive algorithm


























119891 (119870119901 119870119894) = sum (abs (119890)) (8)






119866 (119904) =22

896119890 minus 61199043 + 727119890 minus 31199042 + 0945119904 (12)


0 05 1 15 2 25 30

02

04

06

08

1

12

14

Time (s)

Fitn

ess v

alue

RefferanceDNA-PI

Adaptive DNA-PI


0 5 10 15 20 25 30 35 40minus58297

minus58296

minus58295

minus58294

minus58293

minus58292

minus58291

minus5829

Iterations

Fitn

ess v

alue



0 5 10 15 20 25 300

2

4

6

8

10

12

14

Time (ms)

Resp

onse

syste

m

DNAGA

QPSOYBS

times103

(a)

0 5 10 15 20 25 300

1

2

3

4

5

6

7

8

9

10times104

Time (ms)

Resp

onse

syste

m

DNAGA

QPSOYBS

(b)






6 Conclusions


References







































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





DNA codingsystem





119870119901 AGT TAA TTT 013 300 333 000111 110000111111


0 63

119870119894

GTT CCAAGC 133 220 012 011111 101000

000110


0 4




1198761 + 1198762 (1)

119898best =(sum119899

119898=1119901best119894)119899 (2)








+ 15 (4)




value

Apply virusmutation

Stop


Crossover

QPSO

Pb

E

V

Mu


No

Yes

a b


DNA sequences are



N

(a)

System++

1

S

DNAcomputing

Pb E VMu

+minus

e u y

X

X

KpKir

N a b

QPSO algorithm

(b)




Adaptive algorithm


























119891 (119870119901 119870119894) = sum (abs (119890)) (8)






119866 (119904) =22

896119890 minus 61199043 + 727119890 minus 31199042 + 0945119904 (12)


0 05 1 15 2 25 30

02

04

06

08

1

12

14

Time (s)

Fitn

ess v

alue

RefferanceDNA-PI

Adaptive DNA-PI


0 5 10 15 20 25 30 35 40minus58297

minus58296

minus58295

minus58294

minus58293

minus58292

minus58291

minus5829

Iterations

Fitn

ess v

alue



0 5 10 15 20 25 300

2

4

6

8

10

12

14

Time (ms)

Resp

onse

syste

m

DNAGA

QPSOYBS

times103

(a)

0 5 10 15 20 25 300

1

2

3

4

5

6

7

8

9

10times104

Time (ms)

Resp

onse

syste

m

DNAGA

QPSOYBS

(b)






6 Conclusions


References







































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





value

Apply virusmutation

Stop


Crossover

QPSO

Pb

E

V

Mu


No

Yes

a b


DNA sequences are



N

(a)

System++

1

S

DNAcomputing

Pb E VMu

+minus

e u y

X

X

KpKir

N a b

QPSO algorithm

(b)




Adaptive algorithm


























119891 (119870119901 119870119894) = sum (abs (119890)) (8)






119866 (119904) =22

896119890 minus 61199043 + 727119890 minus 31199042 + 0945119904 (12)


0 05 1 15 2 25 30

02

04

06

08

1

12

14

Time (s)

Fitn

ess v

alue

RefferanceDNA-PI

Adaptive DNA-PI


0 5 10 15 20 25 30 35 40minus58297

minus58296

minus58295

minus58294

minus58293

minus58292

minus58291

minus5829

Iterations

Fitn

ess v

alue



0 5 10 15 20 25 300

2

4

6

8

10

12

14

Time (ms)

Resp

onse

syste

m

DNAGA

QPSOYBS

times103

(a)

0 5 10 15 20 25 300

1

2

3

4

5

6

7

8

9

10times104

Time (ms)

Resp

onse

syste

m

DNAGA

QPSOYBS

(b)






6 Conclusions


References







































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in























119891 (119870119901 119870119894) = sum (abs (119890)) (8)






119866 (119904) =22

896119890 minus 61199043 + 727119890 minus 31199042 + 0945119904 (12)


0 05 1 15 2 25 30

02

04

06

08

1

12

14

Time (s)

Fitn

ess v

alue

RefferanceDNA-PI

Adaptive DNA-PI


0 5 10 15 20 25 30 35 40minus58297

minus58296

minus58295

minus58294

minus58293

minus58292

minus58291

minus5829

Iterations

Fitn

ess v

alue



0 5 10 15 20 25 300

2

4

6

8

10

12

14

Time (ms)

Resp

onse

syste

m

DNAGA

QPSOYBS

times103

(a)

0 5 10 15 20 25 300

1

2

3

4

5

6

7

8

9

10times104

Time (ms)

Resp

onse

syste

m

DNAGA

QPSOYBS

(b)






6 Conclusions


References







































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




0 05 1 15 2 25 30

02

04

06

08

1

12

14

Time (s)

Fitn

ess v

alue

RefferanceDNA-PI

Adaptive DNA-PI


0 5 10 15 20 25 30 35 40minus58297

minus58296

minus58295

minus58294

minus58293

minus58292

minus58291

minus5829

Iterations

Fitn

ess v

alue



0 5 10 15 20 25 300

2

4

6

8

10

12

14

Time (ms)

Resp

onse

syste

m

DNAGA

QPSOYBS

times103

(a)

0 5 10 15 20 25 300

1

2

3

4

5

6

7

8

9

10times104

Time (ms)

Resp

onse

syste

m

DNAGA

QPSOYBS

(b)






6 Conclusions


References







































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





6 Conclusions


References







































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in






















Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in










Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in



Documents

Research Article QPSO-Based Adaptive DNA Computing Algorithmdownloads.hindawi.com/journals/tswj/2013/160687.pdf · of DNA computing algorithm are found using QPSO and used for the