Job Scheduling Mechanisms in Fog Computing Using Soft Computing Techniques.

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

Job Scheduling Mechanisms in Fog Computing Using Soft Computing Techniques.

1Mehraj Maqbool, 2Harwant Singh Arri

1M Tech, School of Computer Science and Engineering, Lovely Professional University, Punjab 2Assistant professor, Dept. of computer science and engineering, Lovely Professional university, Punjab ***

Abstract-The internet of things may be a flexible and emerging technology, it's large and complicated network of devices during which fog plays a crucial role, and fog nodes handle the information flow of such an oversized and sophisticated network, because the cloud and fog provides the on demand virtualized resources like computing resources and storage resources to its users. To allocate the resources to the IoT tasks, dynamic and efficient load balancing algorithms is wont to improve the general performance and reduce energy consumption, i.e. the target is to a way to allocate the suitable resources to the tasks and dispatching the computing tasks from the available resource pool. the choice may affect the execution time, energy consumption, load distribution and can affect the value incurred by the user .this paper specialize in the dynamic workflow scheduling mechanisms in fog computing and can give the comparison of various workflow algorithms on the idea of various parameters and can classify the various workflow and task scheduling algorithms and at the top of the paper we'll determine the realm of research and supply some directions of future work.

Key words: fog computing, cloud computing, PSO algorithm, Monkey algorithm, genetic algorithm.

1. INTRODUCTION

ManyTechnologiesareontrendthesedays,InthesamewayIoThasgainedamuchmarketandhasbecomeanimportantand coreofmanyapplications,therearemanyapplicationwhereIoThasgainedthepopularitylike,Smartgrids,smarthealthcare ,5G connectivity ,transportation and supply chain smart cities ,wearables ,Traffic monitoring, Agriculture ,Hospitality ,and Maintenance management . According to one of the research work done by Abbasi and sha [1], 50 billion devices would be connectedtotheinternetby2020andifthisnumbercanbedividesitwillgivesusthatabout6.38deviceswouldbeusingper userin2020.And the number would bychangingrapidlyandwill reach500 billion devices by2025.The IoTdevicemaybe anything which can share, process, generate data, and have a connectivity feature is called IoT device [2]. As when these heterogeneousdeviceswillbeconnectedtheywillproduceahugeamountofdata,tohandlethislargenumberofdataweneed a large capacity of storage, bandwidth and computational power. Keeping all the requirements of IoT environment into consideration the organizations developed the infrastructure known as cloud computing. American psychologist and computerscientistJ.C.R.Licklider(JosephCarlRobnettLicklider)developedtheideaofcloudcomputingintheearly1960s.In aneffortto connectpeople anddata globally,he workedon theARPANet(AdvancedResearchProjectAgency Network)and introduced the Cloud Computing concept that is so well known today. [3] .IN cloud computing the data centres are geographically centralized, and the IoT devices taking services from these data cloud. As the data centres are located at differentfixedlocations,tohandleallthedataofIoTapplicationwhichisinzetabytesmakessomeoftheIoTapplicationslow ,thereforemanagingandprocessing suchahugedatawasachallengingandfacingalotofissueswiththeapplicationswhich needs a real time processing such as heath ,traffic monitoring transportations and wearables ,due to these challenges the organisationmoveintothenewnetworkinfrastructurewhichovercomestheloopholesofthecloudcomputingwhichiscalled theFogcomputing.

FogComputingwasfirstinitiated byCiscotoextendthecloudcomputingto the edgeofthenetwork [4].Thus enablesa new breedofapplicationsandservices.themotivationbehindtheFogcomputingwasthatitshouldnotincurwiththeloopholes which was in cloud computing, and it should reduce the burden of processing in cloud computing ,it should reduce the high latency which the cloud takes for processing the data ,so that the applications which needs the real time responses will operatesmoothly.Itshouldalsoreducethedistributedenvironmentalcomplexity[5].Thefogdeviceswouldbeinstalledcloser totheenddevicessothatthemostof thedataprocessingwouldbedoneatthesedeviceswhichwillinturnwillreducesthe burdenoncloudcomputingandwillreducesthebandwidth,computingandstorageofthecloud[6].toadoptthesefeaturesin fog computing architecture most of the IoT applications can use fog computing services, like health care centres ,traffic monitoringsystems,smartgridsandwearablesetc.

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

Alikeofcloudcomputingfogisadecentralizedinfrastructure,wherefognodescanjoinandleavethenetworkarbitrary[7].As the fog nodes are closer to the IoT end devices by this fog computing reduces the latency which was a challenge in cloud computing,apartfromthisfognodescangivealocalstorageandcanalsosomeofthecomputationlocallywhichcanenhance the efficiency of applications which needs the real time response, by giving these services closer distance fog cannot replace thecloudasthefogdevicesarecontainedbyresourceslikestoragecapacity,processpower,andenergy.Therefore,thefogis a supplementtocloudcomputationratherthana replacementforit.Theconnection betweenfogandcloudisimportant[8]. fognodesgivesrealtimeprocessingandpushesthedataintothecloudforstorage,asmostoftheprocessisdoneatfognodes the only useful data is send towards the cloud for storage, which will use the overall bandwidth ,processing and storage efficiently.

Fromtheabovediscussion,Wehavestatedthatfogisacomplementtocloudcomputingratherthanareplacementforit.,but itcanjustleveragetheefficiencytothetimesensitiveapplications,thecomputationaloperationsoftimesensitiveapplications are carried out at the fog layer, through the resources which are available at fog nodes .by doing this the latency of the applicationsisreducedandthebandwidthissaved.

Itwilluselessenergybyusingfogcomputing.Infogcomputing,thedesignwillutilisethepotentialandresourcespresentin fogdevices to the greatest extent possible; but,ifwe require efficient resources thatare notpresentinthe fog environment, wewilltransfertothecloud,withassociatedhighercosts.Continuedusageoffogresourceswillresultinlowerprices,shorter waittimes,higherlevelsofsecurityandconfidentiality,lessnetworktraffic,andgreenercomputing.

Inthisarchitecture the foglayer nodeswill provide, resourcestotheIoTdevices [9][10].byproviding these resources inan efficientwaytheresourceschedulerandmanagercomeintopicture,itwouldbetheresponsibilityoftheschedulerwhichwill provide the resources in an efficient manner. When IoT devices submit the requests to the fog nodes ,the scheduler will providetheapproximateresourcestothetasks[11]Afterassigningthejobs,thefognodeswilleitherbeovercrowdedorunder loaded.[12][13]soweshouldhave adynamicandefficientloadbalancingalgorithmswhichwillincreasetheoverallefficiency ,performanceandreducethepowerenergyconsumption.

Different optimization strategies have been suggested in order to solve the scheduling algorithms in a fog cloud setting. Researchers have categorised optimization approaches in a number of ways, with deterministic and stochastic optimization being one of them.Indeterministic optimization,the data for the given issue are precisely known. But occasionally,the data cannotbeknownwithabsolutecertaintyforanumberofreasons.Astraightforwardmeasuringerrorcanbetoblame.Another reason is that some statistics describe future knowledge, which makes it impossible to know with certainty. Stochastic optimizationisusedinoptimizationunderuncertaintywhentheuncertaintyisbuiltintothemodel.

Thekeycontributionsofthisstudyarehighlightedasfollows:

•Introducingavariousoptimizationtechniquesbasedonmetaheuristicoptimizationalgorithmsinfogcloudenvironment

• Classification of many methods that have been proposed on the fundamentals of job, task, workflow, and resource scheduling.

•Discussingandcomparingtheexistingschedulingalgorithmsonthebasisofperformanceandefficiency.

The remainder of the document is structured as follows: The concept of scheduling algorithms is covered in Section 2. The ConceptsandMethodsofOptimizationareprovidedinSection3.Section4presentsareviewofrelatedresearchonscheduling algorithms.DiscussionandcomparisoninSection5.ConclusionandResearchGapsinSection6 CitationsforSection7.

2. CONCEPTS

This section outlines some key terms used in this research and provides a brief explanation of the logical justifications that driveresearchtowardfogcomputation.

As IoT has changed very fast ,it is now present in everywhere in our surroundings stated from universities ,vehicles ,homes ,health care centres smart grids ,traffic monitoring systems and wearables etc and is in fact increasing continuously[14][15]

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

,buttheseIoTdeviceshavenottoomuchcapacitytostorelargedataandcannotperformcomplexcomputationaloperations, because as if they will perform by its own they will consume energy very soon ,therefore there is a need to offload the task whereneedsaintenseresourceoutsthehigherlevellayerswithmoreresources[16]thereiswherefogandcloudcomeinto picture[17][18].

Fog is a decentralized computational scheme which is between IoT devices and cloud .before existence of fog computation architecture, IoT devices were taking services directly from cloud ,but Fog computing is being used, which raises serious concernsbecausethecloudwasunabletomeettheneedsofthemajorityofIoTapplicationsbecauseclouddatacentreswere geographicallycentralised,whichdramaticallyreducedQoS[19][20].AccordingtothedefinitionoffogprovidedbyCisco,fog is a geographically dispersed computer architecture that provides a pool of resources at the network's edge that are not dependent on the cloud and are used for elastic computation, archiving, and computation [21]. As a result, the goal of fog computing is to locally offer cloud like services to serve IoT applications, particularly delay sensitive applications. To accomplishthesegoalsandenhancetheeffectivenessofIoTdevicesandfogcomputing

Therearefivetypesofschedulingmechanismsas:

1. Resourcescheduling.

2. Taskscheduling. 3. Resourceallocation. 4. Workflowscheduling.

5. Jobscheduling.

2.1 Resource scheduling

Theobjectiveoftheresourceschedulingistoprovidethebestmachineresourcesforthecustomerstoobtainbestscheduling goal,Whenschedulingresources,thescheduleraimstoincreaseresourceutilisation,cutdownondelays,andimproveservice quality.QoS[23]

2.2 Task scheduler:

Thegoaloftaskschedulingistodistributeasetofjobsamongtheavailablefognodesinordertosatisfythequalityofservice QoSrequirementwhileoptimizingtaskexecutionandtransmissiontimes[24].

2.3

Resource allocation:

The goal of resource allocation is to match the available resources to client needs over the internet in a systematic manner. [25].

2.4 Workflow scheduling:

The goal of workflow scheduling is to assign computing resources with various processing powers to work flow application jobsinordertoreducemaketimeandcost[26].Anapplication'spriceincludesitscomputational,data transfer, andstorage costs.

2.5 Job scheduling

Thegoaloftaskschedulingistogiveacollectionofjobstotheleastamountoffog resourcessothattheycanbecompletedin theleastamountofCPUtime.

3. OPTIMIZATION

Inthesimplesenseoptimizationisa procedureofmaximizingorminimizingtheobjectivefunctionproblem,sofora simple function like optimization is a simple task, and can be done in an easy way. but in case a function is a nonlinear multimodal

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

function,multivariatelikethetaskschedulinginfogcomputing[]theoptimizationisnotaneasytask,thisproblemmayleadto many challenges and cannot be solved using the traditional methods. For those problems, efficient algorithms needed to be proposed.

Optimization Algorithms.

There have been numerous optimization algorithms developed, and one of the categories is as deterministic or stochastic. Efficientsearchmethodsarerequiredtohandletheoptimizationissuesoffogclouds.Deterministicalgorithmsarethosethat operatemechanicallydeterministically,withouttheuseofchance,andwillalwaysarrivetothesameconclusioniftheybegin from the same starting point. Examples of this type of algorithm are hill climbing and downhill. However, stochastic algorithms,whichmightachievevariousconclusionpositionsevenwhileutilizingthesamestartingpoint,havesomeelement ofrandomness.ExamplesofstochasticalgorithmsaregeneticandPSOalgorithms.

MetaheuristicAlgorithmswithstochasticcomponentsareoftenreferredtoasaheuristicinthepast,asrecentliteraturetends to refer to them as metaheuristic since all nature inspired algorithms proposed for optimization is called meta heuristic algorithmsaccordingtogloversconvention[].Metaheuristicalgorithmsmaybethoughtofasamasterstrategythatdirectsand changesotherheuristicstoyieldanswersbeyondthosethataretypicallygeneratedinahuntforlocaloptimality.Literally,the heuristic means to find out to discover the solution by hit and trial approach. There is a particular trade off that all metaheuristicalgorithmsemployinordertofindsolutionstorandomizationissuesinanacceptableperiodoftime,butthere isnoassurancethatthealgorithmswillalwayssucceedindoingso.Exploitationandexploration,alsoknownasintensification anddiversification,aretwo keyelementsof every metaheuristicalgorithm.Exploitationisthesearch'slocal emphasis,while explorationinvolvesproducingavarietyofanswerstoexplorethesearchspaceglobally.

Different Meta heuristic approaches for optimization in fog cloud environment.

3.1 PSO Algorithm (Particle Swarm Optimization)

KennedandEarhartcreatedPSO,anevolutionarycomputationalmethod,in1995.Giventhatthealgorithmwasinfluencedby aflockofbirds,itisa Metaheuristic.Beingameta heuristicalgorithm,thePSOstartswithapopulationofrandomsolutions, assigns a random velocity to each random solution, and then sends the possible solutions, known as particles, flying around theproblemspace.Eachparticlerecordsitscoordinatesintheissuespaceateachstage,alongwiththebestsolution(fitness). Ithassofarsucceeded.Theglobalversionoftheparticleswarmoptimizertracksanotherbestvalueinadditiontothefitness value,whichisalsosavedandisknownaspbest.

1.Intheissuespace,initializethepopulation(array)ofparticleswithrandompositionsandspeedsonad dimensionalgrid.

2.Assessthefitnessfunctionandgoalfunctioninddimensions.

3.Ifthepresentpositionind dimensionalspaceis preferable,updatethepbestandpbestlocationandcomparetheparticles fitnessevaluationwiththeparticlespbest.

4.Ifthenewevaluationoffitnessisbetterthanthepopulation'spriorbest,updatethegbest.

5.Modifytheparticle'slocationandvelocityinaccordancewithEquations(1)and(2).

Vi(t+1)=wvi(t)+c1r1(pi xi(t))+c2r2(pg xi(t)). …………………….1

x(t+1)=x(t)+vi(t) …………………………………………………………2

6.Repeatstep2untilarequirementissatisfied.

In the velocity equation of PSO there are some parameters which users can set according to the problem statement and different

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

3.2 GENETIC ALGORITHM

JohnHollandinitiallypresentedgeneticalgorithms[1],whicharesearchandoptimizationalgorithmsbasedonthe principles of natural evolution, in 1970. By replicating the development of species through natural selection, genetic algorithms also carry out the optimization procedures. Typically, a genetic algorithm consists of two steps. The first step is choosing an individual to produce the next generation, and the second is using crossover and mutation procedures to manipulate the chosenpersontoproducethefollowinggeneration[2].Whichindividualsarechosenforreproductionandhowmanychildren eachselectedpersonproducesaredeterminedbytheselectionprocess.Thefundamental tenetofselectionstrategyisthata person'slikelihoodofhavingchildrenincreaseswiththeirlevelofcompetence.

Usingageneticalgorithm,apopulationofstartingpeopleisevolved.

3.2.1 Selection operation: is to select elitist individuals as a parents in current population which will be used for further generation,fortheselectionprocedure thefitnessvaluesareusedasacriteriatojudgewhetherindividualsareelitist,various approachesareusedtoselecttheindividualsegBoltzmannselection,tournamentselection,roulettewheelselection.

3.2.2 Crossover operation: The cross over operator may create two new children from two parent strings by copying two selected bits from each parent. In a genetic algorithm, the generation of successors is governed by a series of operators that recombineandalterselectedmembersofthecurrentpopulation.

3.2.3 Mutation operation: In addition to the recombination operation, which creates offspring by fusing elements of two parents,thereisanothertypeofoperatorthatcreatesoffspringfromasingleparent,particularlythemutationoperator,which createssmallrandomchangestobitstringsbycrossingasinglebitatrandomandchangingitsvariousvaluesaftercrossover.

3.3 MONKEY ALGORITHM

The basic monkey algorithm, which has four processes (initialization process, climb process, watch jump process, and somersault process), was the inspiration for the monkey algorithm, which is an improved version of the basic algorithm. In thisimprovedversion,afifthprocesscalledrandomperturbationisadded.Thealgorithmsfunctionas

3.3.1 Initialization process: Since M denotes population size (number of monkeys), it creates a random position for each monkey,withithbeingasfollows:Xi=(i=1,2,3,4,5...M)foreachmonkeyasxi=(x1,x2.....,xm)(1)

Afterthatthepositionofeachmonkeyisevaluatedinobjectivefunction.

3.3.2 Climb process:inthisthemonkeychangesitspositionfrominitialpositionstepbystepthatcanmakeimprovementsin theobjectivefunction,Thefollowingequationdescribesthelengthofthemonkey'sstridewhileshiftingpositions: ∆xij {

Whereaispositivenosteplengthintheclimbprocesswitha=10 3,xijisupdatingthelocationofthemonkey(j=1,2,3,4,...,n).

3.3.3 Watch jump process: This procedure inspects each monkey's position following the climb. It determines whether or nottheirpositionhasreachedthetopbyhavingeachmonkeyscantheareaforapositionthatishigherthantheirpresentone. Iftheyfindone,theywillleapfromit;ifnot,itimpliesthattheirpositionhasnotreachedthetop.

3.3.4. Somersault process: They will locate locations near the barycentre of all monkeys' existing positions, which is describedasa pivot,andmonkeys will tumblealongthedirectionleadingtothepivotthankstothismethodforfindingnew positions(searchingdomain).

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

4 RELATED WORKS

Task or Workflow scheduling is an NP hard issue in a distributed computer system, as is well known [27]. To ensure that programmesrunasefficientlyandautomaticallyaspossible,schedulingmustbedone.Thereareseveralstudiesthatarenow available on task or workflow scheduling in a distributed context, with task scheduling under cloud computing platforms receivingthegreatestattention[28][29].

The author in [35] has implemented an efficient algorithm that can minimize the energy consumption in IoT workflow on heterogeneous fog computing architecture in this work the integer linear programming model is built and designed a performance effective algorithm based on the model. It takes into account the two types of edge nodes, of which one has greater performance and the other uses less power. The experiment's findings demonstrate that, when compared to the LongestTimeFirst(LTF),IntegerLinearProgramming(ILP),andRandomAlgorithm,theEMSalgorithmmayprovidethebest energyusage.

[38].InthispapertheauthorproposedAParticleSwarmOptimization basedHeuristicforSchedulingWorkﬂowApplications in Cloud Computing Environments it takes into account both computational cost and data transmission cost, the heuristic function optimizes the cost of the task resources mapping based on the solution given by the particle swarm optimization technique,themappingoptimizestheoverallcostofcomputingandworkflowapplication.

[30].TheauthorofImprovedParticleSwarmOptimizationBasedWorkflowSchedulingdescribesourschedulingsolutionfor Workflowapplicationsthat isbased on IPSOand primarilyfocuses on Workflowscheduling in a cloud fog environment. The originalPSO'schangeininertiaweightisemployedasanuniquenonlinear decreasingfunctioninIPSO.Thegoalistobalance andadapteachparticle'scapacitytoseekduringthesearchprocess.Finally,takingintoaccounttheactualissueofWorkflow applicationsinacloud fogenvironment,aschedulingmethodisdeveloped.

[33].Theproposedalgorithm,calledthe"beelifealgorithm," isa typeofjobschedulingalgorithm. Itsgoal istofind the best waytodistributea setoftasksamongavailablefogcomputingnodesinordertoachievethebestpossibletrade off between CPUexecutiontimeandallocatedmemoryneededbyfogcomputingservicessetupbymobileusers.Whencomparedtoother algorithmslikethegeneticalgorithm(GA)andPSOparticleswarmalgorithm,itsefficiencyisimprovedintermsofexecution timeandtheallocatedmemory.Inthisalgorithm,theyusedtwoperformancemetricstoevaluateCPUexecutiontimeandthe totalamountofmemoryneededtocompleteexecutioninafogenvironment.

[39] forbusinessworkflowsthatrequirealotofparallelinstancesand areinstance intensive,suggestanearlyidealdynamic priorityscheduling(DPS)technique.[31].TheHHalgorithm,orhyper heuristicalgorithm,usesmachinelearningtechniques to choose, combine, produce, or match several sample heuristics in order to solve computational search problems. The suggestedapproachlocatestheidealresponsetotheworkflowschedulingissues.IncomparisontoPSO,ACO,andSA,thisHH algorithmprovidesthebestaverageenergyusage.Byprovidingresourcestouserswithpreciselimits,thisstrategyshortens simulationtimeandsavesenergy.Italsogivesusersmorecontroloverresourceallocation.

[36]. the goal of this [paper] is to decrease the average response time and to optimize resource utilization efficiently by scheduling tasks and managing fog nodes that are available. The author proposed a novel bio inspired hybrid algorithm in which the resources are managed according to the incoming requests of the users. ABIHA is a hybrid of modified particle swarmoptimizationandmodifiedcatswarmoptimizationMCSO.InthisalgorithmMPSOalgorithmscheduledthetasksamong thefognodesandthehybridofMPSOandMCSOaremanagingtheresourcesatthefogdevicelevel.

[34]. the suggested methodology, called Dynamic Energy Efficient Resource Allocation, is the method for distributing the workload among fog infrastructure. The tasks that the user submits are handled by the task manager and passed to the resourceinformationproviderRIP,whichwillregistertheresourcesforthetasks.Thethirdstepisresourcescheduler,which collectsinformationaboutthetasksfromtaskmanagerandalsogathersinformationabouttheresourcesfromRIP.Tasksand resourcesarethensenttotheresourceengine,whichassignstaskstotheresourcesaspersortedlist,resourceloadmanager examinestheresourcestatusduringexecutionandtransfersthestatustoresourcepowermanagerwhichinturnmanagesthe resourceon/offpowerstatus.Theresultoftheproposedalgorithmiscompared with theDRAMresultshowsthatitreduces theenergyconsumptionandcomputationalcostby8.67%and16.77%.

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

[40]. in this paper the author has proposed an algorithm called DAPSO (Dynamic adaptive particle swarm optimization algorithm)ToimprovethefunctionalityofthefundamentalPSOmethod,theyhaveoptimize theparametersruntimeoftasks byreducingthe makespan of a particulartask set and ithasalsotakeninto considerationthe resource utilization,which is themaximizingthealgorithmisneedtocloudcomputingtoscheduletheindependenttask,inthispapertheauthorhastaken theamalgamation ofCROOKOOsearchalgorithmandDAPSOandproposedtheMDAPSO,inthisalgorithm,Theyhavefounda solutiontothePSOaffinityproblemwheremoreinertiaweighthelpswithglobalsearchesandlessinertiaweighthelpswith localsearches.

[41].In this paper they developed a task scheduling technique for IoT requested in cloud fog environment ,this algorithm is dependon amodifiedartificial basedoptimization(AEO).themodificationisanattempttoenhancetheexploitationabilityof AEOtogettheoptimalsolutionfortheproblemsunderinvestigation,thealgorithmsiscalled asAEOSSA,thismodification is developedusingtheoperatorsofthesalpswarmalgorithm(SSA),theproposedalgorithmsiscomparedwithother proposed algorithmofmetaheuristiconthebasesofperformancemetricandtheproposedalgorithmshowsbetterresultsascompared totheother methodsaccordinglytotheperformancemetricslikemakespantime,throughputetc.

[42].ThisproposedalgorithmsolvesthejobschedulingproblemsincloudcomputingbymodifyingthebasicPSOapproach,in thisalgorithmtheyhaveremovedsomecomponentsofthebasicPSOalgorithmandanalyze theeffectofthatcomponentsin performanceofalgorithm,theyremovedthepersonal memoryterm fromthevelocityupdateformula andreinforcethecase forPSObeingmostlyreliantonsocialinteractionratherthanpersonalexperience.

ThevelocityequationofmodifiedbasicPSOapproachis:

Vi(t+1)=WVi(t)+C2X2(pg xi(t))

Ratherthan

Vi(t+1)=WVi(t)+C1R1(pi xi(t))+C2X2(pg xi(t))

5 DISCUSSION AND COMPARISON:

Thissectionwillgivebriefcomparisonsofvariousalgorithmsoffogcomputingandwewilldiscusssomeimportantinsightsof theworkflowalgorithms,resourceallocation,taskscheduling,andjobschedulingalgorithmsTable1showsthecomparisonof differentalgorithms onthe basisofvarious evaluation parameterslikeenergyconsumption,cost, responsetime,make span, bandwidthutilization,memoryallocatedanddatatransmissioncost.Fromthegiventablemostofthealgorithmstriedtosolve theproblemofmakespan,energyconsumption,andcostbutonlyafewhavetakentheresponsetime, memoryallocationand datatransmissioncostintoconsideration.

Intable2wehavecomparedthevariousalgorithmsonthebasisofenvironmenti.e. forwhichinfrastructurethealgorithms are built may be fog or cloud and which algorithm is what type whether job scheduling, resource allocation, workflow scheduling,ortaskscheduling,andfromwhichalgorithmsareperformingbetterwhencomparedwiththeexistedonce.

In table 3 we have given the simulation techniques of each algorithm whether ifogsim , MATLAB, CloudSim, or c++ and for each algorithm, we have figured out some advantages and disadvantages which will give us the picture of the future work .advantagesarediscussedonthebasisofdifferentparametersofevaluationwhencomparedwithotheralgorithms.

As a result, workflow scheduling in fog is a crucial study field that has to be pursued. The fog plays a significant role in the implementation of IoT scheduling algorithms. According to the process scheduling algorithms, new approaches to maximize make span, cost, energy consumption, resource usage, or bandwidth are required. Due to the fact that time sensitive IoT applications employ fog computing. The overall execution time (make span) that the jobs require is a crucial element. Additionally, because the fog has limited resources, energy wastage and battery life have become more of a problem. By employing an effective strategy for data scheduling, these measurements may be preserved in the fog infrastructure.. Other metricsandapplications,suchasdynamictaskscheduling,periodictasks,taskmigrationbetweenedgenodes,heterogeneous fognodes,applicationswithsoftorharddeadlines,virtualmachinemigration,andenergyconsumptioninfognodes,mustbe takenintoaccountbytheresearchersinorderfortheirnewproposedschedulingalgorithmtobeeffective.

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

Table 1:Comparisonofvariousalgorithmsonthebasisofdifferentparameters

Algorithms Make Span Cost Bandwidth Utilization Response Time Energy Consumption

Allocated Memory Data Transfer Cost

HH YES YES NO NO YES NO NO

EMS NO NO NO NO YES NO NO

PSO NO YES NO NO NO NO YES

IPSO YES YES NO NO NO NO NO

BLA YES NO YES NO NO YES NO

ERA NO NO NO YES NO NO YES

MARKET YES YES NO NO NO NO NO

DEER NO YES NO NO YES NO NO

MDAPSO YES NO YES NO NO NO NO

AEOSSA YES NO NO NO NO NO NO

SPSO YES NO NO NO NO NO NO

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

Table-2: Comparisonofvariousalgorithmsonthebasisofenvironment,schedulingtypeandresultcompared

Refer ence Algorithm

Scheduling Type Parameter Taken Into Consideration Enviro nment Result Compared

[31] HH (Hyper heuristic) Workflow scheduling Make span, usageof energynetworkusage, cost

[35] EMS (EnergyEfficient Scheduling)

[38] PSO (particleswarm optimization)

fog Particleswarm optimization,genetic algorithm,antcolony optimization,

Workflow scheduling Energyconsumption fog Longestfirsttime,ICP, randomalgorithm

Workflow scheduling Computationalcost,data transmissioncost Cloud Bestresource scheduling

[30] IPSO(improved particle swarmoptimization) Workflow scheduling Costandmakespan Fog cloud particleswarm optimization

[33] BLA(BeesLife Algorithm) Job scheduling Makespan,allocated memory fog particleswarm optimization,genetic algorithm

[36] NBIHA (NovelBio inspired HybridAlgorithm)

[32] ERA (EfficientResource Allocation)

Task scheduling andresource allocation

Makespan,response time,energy consumption

Resource allocation Responsetime,data transmissionexpense, andbandwidthuse

fog Firstcomefirstserve, SJE,MPSO

fog RDLB,ORT

[37] Market oriented hierarchicalStrategy) Workflow scheduling Cost,makespan,cputime Cloud PSO,geneticalgorithm

[34] DEER(DynamicEnergy EfficientResource Allocation)

[40] MDAPSO (Modified dynamicadaptiveparticle swarmoptimization)

[41] AEOSSA(Eco system basedandsalpswam algorithm)

Resource allocation Energyconsumption computationalcost fog DRAM

Task scheduling Make span,Resource utilization Cloud PSO

Task scheduling Make spanand Throughput Cloud fog AEO,SSA,PSO

[42] SPSO(Simplifiedparticle swarmoptimization) Job scheduling Make span Cloud GA,PSO

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

Table 3:AdvantagesandLimitationsofvariousalgorithmsandsimulatedtools

Reference Algorithm Simulated by Advantages

[31] HH ifogSim HHalgorithmimprovedtheaverage energyconsumptionascomparedto PSO,ACO,SAalgorithmby69.99 costtocost59,62%

[36] NBIHA ifogSim Firstcome,firstserved,shortest taskfirst,andmodifiedparticle swarmoptimizationmethodsare inferiortothesuggestedalgorithms intermsofexecutiontime,energy consumption,andaverageresponse time.

[35] EMS SPECCPU 2006 EMSalgorithmuseslessenergy thanthelongestjobfirst,integer linearprogramming,andrandom algorithms.

Limitations

InthisalgorithmVM(virtual machine)migrationconceptisnot address.

Ithaslowscalability.

Communicationcostisnotaddressed, theintendtousereinforcement learningtechniquestomanaging resourcesinfogIoTenvironment.

Nottakingdifferentnodesinto considerationittakenonlytwotypes ofnodes.

Noconsiderationisgiventotask migrationacrossedgenodes.Itdoes notconsiderperiodictaskwhich shouldbefinishedwithintheir periods

[30] IPSO MATLAB 2016a

Theoverallcompletiontimeand economicalcostisreducedas comparedtoPSO

[34] DEER CloudSim

IncomparisontoDRAM,theDEER iseffectiveatloadbalancingtocut downonenergyusageand computationalcostsby8.67percent and1.677percent,respectively. Thereisnooverheadforsorting.

Ittakesonlytwometricsinto considerationi.e.computationaland economicalcost.

Itgivesthetrade offbetweenmake spanandcost

Inthisproposedalgorithmthereisno faulttolerancemechanism implemented.

Highcomplexity

[33] BLA C++

Whentheproposedalgorithmis comparedwithparticleswarm optimizationandgeneticalgorithm itperformsbetterintermsof processingspeedandmemory allotted

Theproposedalgorithmistestedon smalldatasetandithaslow scalability.

Thedynamicschedulingtechniqueis notaddressed,andthetaskexecution reactiontimeislong.

[32] ERA CloudAnalyst

Theproposedalgorithmshows betteroptimizedwayofresource allocationwhencomparedwith RDLB,ORTandintermsofoverall responsetime,datatransfercost andbetterbandwidthutilizationin thefogcomputingenvironment

Resourcesareonlyallottedtousers whohaverequestedthempriorto processingbecausetheneedsforthe resourcesduringtheexecutionofthe requestarenotsatisfactory.

Itonlyappliestoinstancesthathave beenreserved. pooraccessibility.

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Itisreallycomplicated. andlengthyexecution

[38] PSO JSwarm package Thecomputationalcostand communicationalcostsavingsare3 timesmoreinPSOwhencompared withthebestresourceselection algorithmBRS.

Itgivesgoodworkloaddistribution ontoresources.

[40] MDAPSO CloudSim3.3 Resultsshowthatitoutperforms thedefaultPSO,ADPSO,andPSOCS algorithmsby38.63and25.30 percent,respectively.

[42] Simplified PSO CloudSim Easytoimplement

6. CONCLUSIONS

Itdoesnotworkwellinscheduling workflowofrealapplications.

Itimplementationiscomplexasitthe combinationoftwoalgorithms

Itdidn’ttakenconsiderationmostthe performanceparameters

Theworkflowschedulinginfogcomputingisoneofthemostsignificantcontemporaryschedulingmethodsthatisthoroughly reviewed and analyzed in this study. The best scheduling methods are picked after a thorough reading and analysis of the majorityoftherecentpublicationsreleasedonschedulingalgorithmsinfogcomputing.Taskscheduling,resourcescheduling, resource allocation, job scheduling, and workflow scheduling are the five key categories used in this research to group the scheduling methods. From the comparison outcome, the majority of researches focused on energy usage and make span. However,otherfactorslikecost,reactiontime,andmemoryallocationarethemostimportantmetricsthatresearchersmust take into account in order to optimize the efficiency of IOT applications. Additionally, because the resources available to the foginfrastructurearefew,make spanandenergyusagearecrucialparametersinthefogcomputingschedulingalgorithms.

REFRENCES

[1] B.Z.AbbasiandM.A.Shah,"Fogcomputing:Securityissues,solutionsandrobustpractices,"2017 23rdInternational ConferenceonAutomationandComputing(ICAC),2017,pp.1 6,doi:10.23919/IConAC.2017.8082079.

[2] https://www.oracle.com/in/internet of things/what is iot/ [3] https://www.educba.com/history of cloud computing/

[4] Bonomi,Flavio&Milito,Rodolfo.(2012).FogComputinganditsRoleintheInternetofThings.ProceedingsoftheMCC workshoponMobileCloudComputing.10.1145/2342509.2342513.

[5] XiaoM, ZhouJ, LiuX, JiangM.A hybrid scheme for fine grained search and access authorization in fog computing environment.Sensors.2017;17:1423.

[6] TariqN, AsimM, AlObeidatF ,et al. The security of big data in fog enabled IoT applications including block chain: asurvey.Sensors.2019;19:1788.

[7] Mohanta, Bhabendu & Jena, Debasish & Panda, Soumyashree & Gountia, Debasis. (2019). Decentralized Secure Fog ComputinginCloud Fog IoTInfrastructureUsingBlockchain.

[8] Dastjerdi AV, GuptaH, Calheiros RN,Ghosh SK,Buyya R.Fog computing: principles, architectures, and applications. InternetofThings.Netherlands:Elsevier;2016:61 75.

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

[9] M.OguraandV.M.Preciado,‘‘Stabilityofspreadingprocessesovertimevaryinglarge scalenetworks,’’ IEEETrans.Netw. Sci.Eng.,vol.3,no.1,pp.44 57,Jan.2016.

[11] H. Yuan and M. Zhou, ‘‘Proﬁt maximized collaborative computation ofﬂoading and resource allocation in distributed cloud and edge computing systems, ’’IEEET rans. Automat. Sci. Eng., EarlyA ccess, Jul.14,2020,doi: 10.1109/TASE.2020.3000946.

[12] Y. Kong, M. Zhang, and D. Ye, ‘‘A belief propagation based method for task allocation in open and dynamic cloud environments,’’Knowl. BasedSyst.,vol.115,pp.123 132,Jan.2017.

[13] Li, Peng, Shengli Zhao, and Runchu Zhang, ‘‘A cluster analysis selection strategy for supersaturated designs,’’ Comput. Statist.DataAnal.,vol.54,no.6,pp.1605 1612,2010

72K.MatroukandK.Alatoun/InternationalJournalofNetworkedandDistributedComputing9(1)59 74

[14] Muralidharan S, Roy A, Saxena N. MDP IoT: MDP based interest forwarding for heterogeneous traffic in IoT NDN environment.FutureGenCompSyst.2018;79:892 9

[15] Gubbi J, Buyya R, Marusic S, Palaniswami M. Internet of Things (IoT): a vision, architectural elements, and future directions.FutureGenCompSyst.2013;29(7):1645 1660.

[16] DíazM,MartínC,RubioB.State of the art,challenges,andopenissuesintheintegrationofInternetofThingsandcloud computing.JNetworkCompAppl.2016;67:99 117.

[17] BottaA,DeDonatoW,PersicoV,PescapéA.Ontheintegrationofcloudcomputing andInternetof Things.In:2014 InternationalConferenceonFutureInternetofThingsandCloud.IEEE;2014,August:23 30

[18] Roca D, Milito R, Nemirovsky M, Valero M. Tackling IoT ultra large scale systems: fog computing in support of hierarchicalemergentbehaviors.In:FogComputingintheInternetof Things.Cham:Springer;2018:33 48.

[19] DastjerdiAV,GuptaH,CalheirosRN,GhoshSK,BuyyaR.Fogcomputing:principles,architectures,andapplications.In: InternetofThings.MorganKaufmann;2016:61 75.

[21] Cisco Systems. Fog computing and the Internet of Things: extend the cloud to where the things are. 2016. [Online]. Available:http://www.cisco.com/c/dam/en_us/solutions/trends/iot/docs/computingoverview.pdf

[22] Mahmud R, Srirama SN, Ramamohanarao K, Buyya R. Quality of experience (QoE) aware placement of applications in fogcomputingenvironments.JParallelDistribComp.2018:190 203.

[23] G.Li,Y.Liu,J.Wu,D.Lin,S.Zhao,Methodsofresourceschedulingbasedonoptimizedfuzzyclusteringinfogcomputing, Sensors(Basel)19(2019),2122.

[24] Z.Hao,E.Novak, S.Yi,Q. Li, Challengesandsoftwarearchitecture for fogcomputing, IEEE Internet Comput. 21(2017),44 53.

[25] S.Agarwal,S.Yadav,A.K.Yadav,Anefficientarchitectureandalgorithmforresourceprovisioninginfogcomputing,Int. J.Inform.Eng.Electron.Bus.8 (2016), 48 61.

[26] R. Xu, Y. Wang, Y. Cheng, Y. Zhu, Y. Xie, A.S. Sani, et al., Improved particle swarm optimization based workflow scheduling in cloud fog environment, International Conference on Business Process Management, Springer, Cham, 2018,pp.337 347.

[27] Jennings,D.;Sabbagh,M.;Pfeﬀerer,C.;Fristch,A.;Russell,D.;Seibyl,J.;Marek,K.;Reininger,C.Aframework for resourceal location strategies in cloud computing environment. In Proceedings of the Computer Software & Applications ConferenceWorkshops,Munich,Germany,18 22July2011

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

[28] Wu, Z., Liu, X., Ni, Z., et al.: A market oriented hierarchical scheduling strategy in cloud workﬂow systems. J. Supercomput.63(1),256 293(2013)

[29] Verma,A.,Kaushal,S.:Ahybridmulti objectiveparticleswarmoptimizationforscientiﬁcworkﬂowscheduling.Parallel Comput.62,1 19(2017).

[30] Xu, Rongbin & Wang, Yeguo & Cheng, Yongliang & Zhu, Yuanwei & Xie, Ying & Sani, Abubakar & Yuan, Dong. (2019). Improved Particle Swarm Optimization Based Workflow Scheduling in Cloud Fog Environment: BPM 2018 International Workshops, Sydney, NSW, Australia, September 9 14, 2018, Revised Papers. 10.1007/978 3 030 11641 5_27.

[31] S. Kabirzadeh, D. Rahbari and M. Nickray, "A hyper heuristic algorithm for scheduling of fog networks," 2017 21st ConferenceofOpenInnovationsAssociation(FRUCT),2017,pp.148 155,doi:10.23919/FRUCT.2017.8250177.

[32] Agarwal, Swati et al. “An Efficient Architecture and Algorithm for Resource Provisioning in Fog Computing.” International Journal of Information Engineering and Electronic Business 8(2016):48 61.

[33] SalimBitam,SheraliZeadally&Abdelhamid Mellouk(2018)Fogcomputingjobschedulingoptimizationbasedonbees swarm,EnterpriseInformationSystems,12:4,373 397,DOI:10.1080/17517575.2017.1304579

[34] A.U.Rehmanetal.,"DynamicEnergyEfficientResourceAllocationStrategyforLoadBalancinginFogEnvironment,"in IEEEAccess,vol.8,pp.199829 199839,2020,doi:10.1109/ACCESS.2020.3035181.

[35] H. Y.WuandC. R.Lee,"EnergyEfficientSchedulingforHeterogeneousFogComputingArchitectures,"2018IEEE42nd Annual Computer Software and Applications Conference (COMPSAC), 2018, pp. 555 560, doi: 10.1109/COMPSAC.2018.00085.

[36] H.Rafique,M.A.Shah,S.U.Islam,T.Maqsood,S.KhanandC.Maple,"ANovel Bio InspiredHybridAlgorithm(NBIHA) for Efficient Resource Management in Fog Computing," in IEEE Access, vol. 7, pp. 115760 115773, 2019, doi: 10.1109/ACCESS.2019.2924958.

[37] Wu, Zhangjun & Liu, Xiao & Ni, Zhiwei & Yuan, Dong & Yang, Yun. (2011). A market oriented hierarchical scheduling strategyincloudworkflowsystems.TheJournalofSupercomputing.63.1 38.10.1007/s11227 011 0578 4.

[38] S. Pandey, L. Wu, S. M. Guru and R. Buyya, "A Particle Swarm Optimization Based Heuristic for Scheduling Workflow Applications in Cloud Computing Environments," 2010 24th IEEE International Conference on Advanced Information NetworkingandApplications,2010,pp.400 407,doi:10.1109/AINA.2010.31.

[39] Xu,Rongbin&Wang,Yeguo&Huang,Wei&Yuan,Dong&Xie,Ying&Yang,Yun.(2017).Near optimaldynamicpriority scheduling strategy for instance intensive business workflows in cloud computing. Concurrency Computation. 29. 10.1002/cpe.4167.

[40] Attiya, Ibrahim & Zhang, Xiaotong. (2017). A Simplified Particle Swarm Optimization for Job Scheduling in Cloud Computing.InternationalJournalofComputerApplications.163.34 41.10.5120/ijca2017913744.

[41] Al Maamari, Ali, and Fatma A. Omara. "Task scheduling using PSO algorithm in cloud computing environments." InternationalJournalofGridandDistributedComputing8.5(2015):245 256.

[42] Mohamed Abd Elaziz, Laith Abualigah, Ibrahim Attiya, Advanced optimization technique for scheduling IoT tasks in cloud fog computing environments, Future Generation Compute Systems,Volume 124,2021,Pages 142 154,ISSN 0167 739X,https://doi.org/10.1016/j.future.2021.05.026.

7.529 |

9001:2008 Certified Journal | Page1707

Impact Factor value:

ISO

Turn static files into dynamic content formats.

Create a flipbook