Inernaonal Journal of Compuer Trends and Technology (IJCTT) Volume 18 Number 6 Dec 2014 Even Based Projec Schedulng Usng Opmzed An Colony Algorhm Vdya Sagar Ponnam #1, Dr.N.Geehanjal #2 1 Research Scholar, Dep. of Compuer Scence & Technology,Sr Krshnadevaraya Unversy, Ananhapuram 2 Assocae Professor & Head Dep. of Compuer Scence & Technology,Sr Krshnadevaraya Unversy,Ananhapuram. ABSTRACT The need for compuer-aded ools for Task Schedulng and Resource share/allocaon n Sofware Indusry s ncreasng. The radonal mehod uses boh Even-Based Scheduler (EBS) and An Colony Opmzaon (ACO) o ackle boh he problems n Projec Plannng. Due o hs sgnfcance and challengesof projec plannng, here s a fas growng need for mprovngvaluablecompuer years. The exsng approaches ypcally consder human resource allocaon and ask schedulng as wo dsnc acves. The exsen models also have he predcon ha each person can smply be alloed o a sngle ask a only once due o hs projec are no compleed on me and lacks proper plannng and schedulng n he projec. The problem of ask pre-empon exss n he prevous models. The exsng sysem also suffers from he problem of allocang he same ask for dfferen group of employees n dfferen perods. ACO solve he problem of projec schedulng bu does no consder he employee allocaon marx. The ACO s no a sasfacory model o solve he problem of projec schedulng. INTRODUCTION: Projec managemen s undoubedly an ulzaon of knowledge, sklls, ools and echnque o solve projec schedulng problem. Research no developng effecve compuer aded echnques for plannng sofware projecs s valuable and challengng for sofware engneerng. No he same as projecs n oher felds, sofware projec developmen s naurally a human cenrc acvy.[1] Sofware developmen organzaons ofen fgh delver projecs n me, whn budge keepng he requred qualy. One possble explanaon for he problem s poor sofware projec managemen and, especally, nadequae projec schedulng and neffecve eam saffng. [2] Saffng a ool projec s edous acvy. Manager s nended o choose beween he eam of employees, here can be chance many combnaons. [1] Objecves Movaed by real-world suaons, an array of objecves for projec schedulng have already been suded nsde he leraure. Now we wll jus brng ha up whn hese ndvduals, some objecves can be regardng me, smply because hey concern emporary usage of renewable and doubly consraned resources, whereas ohers â o cos, because conend wh nake nonrenewable and doubly consraned resources. Boh knds usually represen conflcng objecves, snce shorenng he processng me oucomes n ncreasng he resource consumpon, and vce versa decreasng he execuon cos (n regards o he resources consumed) lenghens hs projec duraon. To formulae a reamen projec, hs projec manager has go o esmae hs ask workload and prce and deermne he consrucon projec schedule and resource allocaon. Sofware projec asks requre employees wh dfferen sklls, and skll profcency of employees sgnfcanly nfluences he effcency of projec execuon as shown n Fg 1. Idenfy Acvy ISSN: 2231-2803 hp://www.jcjournal.org Page280 I Fg 1. Basc Flow char of Job Schedulng Allocang employees o he bes fed acves s demandng for sofware planmanagers, and hr allocaon has become an essenal par n sofware projec plannng accordng o he experences and sklls of he employees. Projec managemen approaches usually consder hr allocaon and ask schedulng as wo dsnc acves and leave he job of hr allocaon o be performed by projec managers ndvdually, leadng o nsuffcen poor managemen and resource allocaon effcency. Man assesn sofware developmen are humans n conras wh bg resources,machnee n sofware projecs can normally be allocaed whn a flexble way han hose n manufacurng projecs or consrucon. Problem Defnon Idenfy Acvy Dependences Creae Projec Chars Esmae Resources Allocae People To Acves Sofware projecs are ndvduals nensve acon and requre employees wh varyngsklls. Assgnng employees owards he bes-fed asks and recrung allocaon has come o be an essenal par n sofware projec organzng.. Because of he mporance and dffculy of sofware projec organzng, here exss a growng necessy of developng effecve compuer aded ools for sofware projec organzng n recen mes. The presen approaches usually regard ask schedule and hr allocaon as wo separaed acons. The mos presen models may also have he dea ha each employee earns he person he rgh o be graned o jus one ask smulaneously resulng from hs projec aren' compleed n me and lacks proper organzng and schedule among he projec. Ths assumpon reduces he modfablly of resource allocaon n sofware projec organzng. RCPSP- Resource Consraned Projec Schedule Challenge In he mos general form, RCPSP asks some fundamenal: Handed a mulude of acons, a group of resources, along wh a measuremen of performance, so whch s he bes mehod o assgn he resources ono he acons n a way ha he performance s maxmzed? whch s he bes sraegy o assgn he resources no he acons a specfc mes so ha
Inernaonal Journal of Compuer Trends and Technology (IJCTT) Volume 18 Number 6 Dec 2014 each and every par of he resrcons are sasfed and of course he bes objecve measures are generaed? RCPSP can easly be hough as follows: parcular acons ha needs o be execueda mulude of resources wh whch o carry ou hevacves,specfc resrconswhch wll have o be sasfed a group of objecves wh whch ough o be acheved [1] SOFTWARE PROJECT SCHEDULING PROBLEM (SPSP) SPSP s ypcally a challenge of dscoverng a parcular schedule for geng a sofware projec o ensure he precedence and resource resrcons are sasfed as well as havng he fnal projec cos beng nclusve of personal wagesand projec duraon s mnmzed. And addonally o for he wages and sklls of employees, SPSP also akes workload and requred sklls of each and every ask no mnd, so SPSP s perfec and capable o descrbe he mporan sofware projec schedule. Alhough SPSP s closes o RCPSP, usually here are some dfferences beween SPSP and RCPSP. Frs, here's an exra objecve o be opmzed n SPSP beyond he projec duraon mnmzaon objecve n RCPSP. Second, employees los of possble sklls are classfed as he major resource n SPSP whle here are a number of sors of resources n RCPSP.[2][6] SPSP assocaed wh he resource-consraned projec schedule challenge (RCPSP) whch ams o locae a specfc schedule ha mees he precedence and resource requremens whle reducng he projec duraon. A genec algorhm s among he sochasc search echnques and 's been successfully appled n many search, opmzaon, and machne learnng ssues. Opmzed schedule ssues mgh be solved usng GAs.[7] In accordance o smplfcaons of naural evoluonary processes, genec algorhms run on a populaon of soluons nsead of one soluon and employ heurscs for nsance selecon, crossover, and muaon develop beer soluons. Colorn, Dorgo and Maezzo [1] developed ACO approach n 1991 based on he fac ha real ans are able o fnd he shores pah beween her nes and he source of food. Ths s done usng pheromone rals, whch ans depos whenever hey ravel, as a form of ndrec communcaon. Colorn, Dorgo and Maezzo [1] desgned arfcal ans, whch represens soluons and he collecve nellgence of ans are ransformed no useful opmzaon echnques ha fnd applcaon. Dorgo and Blum [2] revewed he convergence requremens of ACO, connecons beween ACO algorhms and cross enropy mehods. The nfluence of search bas on he workng of ACO algorhms s dscussed n Dorgo and Blum [2]. Mohamed [3] developed ACO algorhms wh sngle an, and also wh fve ans. The model s used for schedulng resource consraned projecs. In he fve ans ACO algorhm, each an fnds a soluon n all eraons and uses he bes-found soluon so far developed for he pheromone updae. Projec schedulng problem (RCPSP) model are some of he projec managemen echnques, ha are appled n sofware projec plannng[base]. The man reason s ha, dfferenly from oher projecs, a sofware projec s a people-nensve acvy and s relaed resources are manly human resources [8]. The human resource allocaon s a complex ask. Assgnng employee o he mos suable ask s challengng. Technques lke PERT and CPM lack he consderaon of resource allocaon and schedulng models lke he RCPSP do no consder he allocaon of employees wh varous sklls [14]. In We-Neng Chen s work [14] he consders he employee allocaon problem. The paper nroduces a new mehod EBS for represenng allocaon of human resources. ACO s used for plannng problem. Even based scheduler s a represenaonal scheme. I s he combnaon of ask ls [3] and employee allocaon marx [4]. The ask ls defnes he prores of asks o consume resources, and he planned employee allocaon marx specfes he orgnally planned workload assgnmens [14]. In hs way, he represenaon akes boh he ssues of ask schedulng and resource allocaon no accoun. The EBS regards he begnnng me of he projec, he me when resources are released from any fnshed ask, and he me when employees jon or leave he projec as evens. To generae an acual meable, he EBS adjuss he workload assgnmens of employees a evens and resource conflc s solved accordng o he prory defned by he ask ls. In Chang s work, Genec Algorhm for Projec Managemen [4], consders a genec algorhm for projec plannng. The schedulng of asks and he allocaon of resource n projecs s an exremely hard problem. Even hough we have an opmal soluon he changng condons wll affec. Brue force exhausve or branch-and-bound search mehods canno cope wh he complexy nheren n fndng sasfacory soluons o asss projec managers. In exsng projec managemen (PM) echnques, commercal PM ools, and research prooypes n no effcen n compuaonal capables and only provde passve projec rackng and reporng ads. Projec managers mus make all major decsons based on her ndvdual nsghs and experence and mus buld he projec daabase o record such decsons and produce repors n varous formas such as Gan or Per chars. Marco Dorgo proposed a new opmzaon echnque called An Colony Opmzaon (ACO). In hs paper he auhors nroduced a new compuaonal paradgm called he An Sysem, a vable new approach o sochasc combnaoral opmzaon. The man characerscs of hs model are posve feedback, dsrbued compuaon, and he use of a consrucve greedy heursc. Posve feedback accouns for rapd dscovery of good soluons, dsrbued compuaon avods premaure convergence, and he greedy heursc helps fnd accepable soluons n he early sages of he search process. The algorhms hey developed are models derved from he sudy of real an colones and are called An algorhms. Lu and Wang [6] developed a flexble model for handlng he opmzaon of schedulng problems n lnear consrucon projecs nvolvng dfferen objecves and resource argumen asks. The model s suggesed for he schedulng of consrucon acves of hgh-rse buldng or brdge projecs. Janxng and Cang [7] demonsraed he use of an colony opmzaon algorhm o solve he dynamc ISSN: 2231-2803 hp://www.jcjournal.org Page281
Inernaonal Journal of Compuer Trends and Technology (IJCTT) Volume 18 Number 6 Dec 2014 problem of resource schedulng n group projec managemen. 1. ACO has good graph-based search problem solvng capably by splng ask and dsrbung employee dedcaon o ha asks. Consrucon graph s generaed based on ha and naurally SPP Problem s convered no graph-based search problem. 2. ACO suppors o use heursc nformaon o ncrease he search ably of ans. There are oal sx heurscs are used n SPPP-ACS, ncludng oal dedcaon of employee, allocaon dedcaon and mporance of asks.aco gves sasfacory soluon as compare o oher. The ACO wll develop an opmzed plan, n he form of marx, from all he eraons. And from ha plan he EBS wll develop schedule based on evens. When an unceran even occurs he remanng resource wll be calculaed, also he remanng asks o complee. And agan a new schedule wll be developed accordng o. I can also consder uncerany a he sarng phase. 1. ACO model: To solve NP-hard problem of projec managemen ACO s frs used. I s used o manage he asks of projec relaed o precedence and resource consrans. To make a schedule need o fnd order of asks whch sasfy ask precedence consran and generae ask ls. ACO solve he problem of projec schedulng bu does no consder he employee allocaon marx. The ACO s no a sasfacory model o solve he problem of projec schedulng. 2. Employee allocaon model: In hs model he problem of how o assgn employees o dfferen asks s fnd ou. The man objecve s o mnmze he number of consran volaons or o mnmze projec duraon and cos. In hs he sofware projec plannng have o assume ha he ask can be done or conduced by an unlmed number of employees and an employee can be assgned o an unlmed number of asks a a me, whch s usually no possble. 3. Mul-skll schedulng model: The model s same as he ACO for ask schedulng and regards dfferen combnaons of employees as dfferen alernave modes for he mplemenaon of a ask. Ths model solve problem of employee allocaon marx as well as ask schedulng hs model only consder he allocaon of employee o ask a only one me, bu hey does no consder he pre-empon of ask. Ths model reduces he allocaon flexbly of he human resources. In hs model f one of he employee s busy n oher ask ha me he whole eam has o wa ll he employee s released. Ths s he drawback of hs model and reduces he effcency of he projec. 4. The me lne based model: The new echnque ha combnes boh he human resource allocaon and ask schedulng s a me lne based model. Ths model generaes he me lne axs o soluon represenaon and makes a possble plan. Ths model has wo drawbacks. Frs, assgn workload perod by perod nsead of ask by ask and second, he plans produced by hs model may assgn wo compleely dfferen groups of employees o he same ask n dfferen perods. Projec plannng s par of projec managemen, whch s o use he schedules o plan and subsequenly repor progress whn he projec envronmen. The purpose of projec plannng s o denfy he scope of he projec, esmae he work nvolved, and creae projec schedule. Projec plannng begns wh requremens ha defne he sofware o be developed. The projec plan descrbes he asks ha wll lead o compleon of he projec. In he proposed mehod a praccal and effecve approach for he ask schedulng and human resource allocaon problem n sofware projec plannng wh an an colony opmzaon (ACO) algorhm s developed. The underlyng dea of ACO s he ans depos a specal chemcal called pheromone on he pah hey ravel hrough whle hey search for food. The pheromone s he communcaon medum beween he ans and by sensng he concenraon of pheromone, ohers ans follow he pah o fnd he food. An ACO algorhm works by dspachng a group of arfcal ans o buld soluons o he problem eravely. ACO algorhm s he repeaed execuon of hree man procedures, Soluon consrucon, Pheromone Managemen and Daemon acons. An Colony Opmzaon The suggesed sraegy s defned by wo mos mporan feaures. Frs, a descrpon scheme made from acvy ls along wh a planned employee allomen marx ogeher wh a novel even based scheduler akes shape. I allows he modelng of resource conflc and acvy preempon. Second, an ACO sraegy s suggesed snce shows successful applcaon o vared combnaoral maxmzaon ssues. ACO bulds soluons nsde a sep-by-sep manner ha wll acually make d ans o plan he crcal asks far back as possble and hen o assgn he consrucon projec asks o approprae employees wh requred sklls. The suggesed mehod effcenly manages employees usng an employee daabase and also denfes asks ulzng a Acvy Precedence Graph whch defnes han a acvy are only able o sar when all of ha drec predecessor asks have fnshed. Hence he plannng objecve among he suggesed sraegy s promsng. The suggesed sysem wll reduce overall projec cos,resources are resourcefully employed n hs ask as well as a new echnque for resolvng he sofware program projec plannng problem I ceranly wll decrease he wo basc ssues n sofware projec managemen ha mgh be acvy schedulng and employee allomen. They provde he clear dea for me schedulng and resource allomen and s gong o lessen he manual effor. The suggesed sysem ulze he resources effcenly and allowng he employers o fnalze anyone acvy among he gven me. I furnshes one of he bes way o solve hs acvy schedulng and employee allomen ssues n sofware projec managemen process. Reckonng on he worker experse, allocang employee o ceran acvy resource perm he consrucon projec o become compleed promply. By evaluang employees work me ha s undoubedly /hour salary for normal effor and overme he expense of hs projec can possbly be mnmzed. The suggesed sysem helps Projec manager n ISSN: 2231-2803 hp://www.jcjournal.org Page282
Inernaonal Journal of Compuer Trends and Technology (IJCTT) Volume 18 Number 6 Dec 2014 allocang projecs o Team leaders and n s place for Team leaders for allomen of acvy o eam members. I ads Daa Forma: allocae employee o perform n overmes o deal wh her asks as well as havng he suggesed algorhm s able o *** yeld beer plans wh lower coss, sronger workload assgnmens decreases he dmensons of he reques space n fle wh basedaa : mf7_.bas conras o oher exsen approaches. AOC ALGORITHM 1. Inalze all parameers.e. Q0, Ngen, Nan whch are used n ACO. These parameers are used o evaluae he mporance of Heursc nformaon and hsory, whch also adjus he pheromone updang, balance he behavor of ans. Nan s number of ans and Ngen s number of generaons. 2. Inalze all pheromone value as 0. 3. Each an selecs her own pah for fndng soluon. Each an selec nex node as per selecon scheme and fll he marx. When ravellng of ans s compleed Soluon marx s consruced. 4. By usng fness funcon evaluaes he qualy of soluon, also calculae he cos, duraon of projec and overwork for ha projec. 5. Compare he soluons and selec he bes one, updae he pheromone value. 6. Repea he procedure ll condon s sasfed. Generally ermnaon condon s deermned by fx number of generaon. 7. Selec and dsplay he bes soluon whose cos and duraon s less. Even-Based Schedulng A schedule s a lsng of a projec's mlesones, acves, and delverables, usually wh nended sar and fnsh daes. The proposed work combnes he ask ls represenaon and he employee allocaon marx represenaon so ha boh he problems of ask schedulng and human resource allocaon are addressed. Sep 1: Inalze he number of avalable human resources Sep 2: Fnd he ask Sep 3: If he planned workng hours s no greaer han he remanng workng hours of he -h employee, assgn planned workng hours of he projec o he number of workng hours of he -h employee for he ask j Sep 4: Else, he number of workng hours of he -h employee for he ask j s se o he remanng workng hours of he -h employee a. Sep 5: Evaluae he compleon suaon of he ask a me Sep 6: If any ask s fnshed a me, se +1 as even Sep 7: Incremen nal value random generaor: 1431890842 *** projecs : 1 jobs (ncl. supersource/snk ): 32 horzon : 241 RESOURCES - renewable : 2 R - nonrenewable : 2 N - doubly consraned : 0 D *** PROJECT INFORMATION: pronr. #jobs rel.dae duedae ardcos MPM-Tme 1 30 0 33 0 33 *** PRECEDENCE RELATIONS: jobnr. 1 1 3 2 3 4 2 3 3 5 7 9 3 3 2 6 16 4 3 3 15 20 22 5 3 1 13 6 3 3 12 14 15 7 3 1 8 8 3 3 10 16 22 9 3 1 28 10 3 3 11 13 17 11 3 1 21 12 3 2 17 26 13 3 3 27 28 31 14 3 2 19 30 15 3 2 24 25 16 3 3 21 24 26 17 3 2 18 30 18 3 3 20 25 28 19 3 3 20 21 22 20 3 2 29 31 21 3 1 27 22 3 1 23 23 3 2 24 25 24 3 1 31 25 3 1 27 26 3 1 30 27 3 1 29 28 3 1 29 29 3 1 32 30 3 1 32 31 3 1 32 32 1 0 #modes #successors successors *** ISSN: 2231-2803 hp://www.jcjournal.org Page283
Inernaonal Journal of Compuer Trends and Technology (IJCTT) Volume 18 Number 6 Dec 2014 21 1 6 8 7 0 8 2 8 7 6 7 0 3 10 7 3 3 0 22 1 2 8 2 4 0 2 3 7 2 1 0 3 9 7 2 0 4 23 1 1 6 9 10 0 2 7 3 3 0 7 3 7 2 6 2 0 24 1 1 4 6 0 5 2 5 4 5 8 0 3 6 3 2 8 0 25 1 8 7 7 5 0 2 9 5 6 4 0 3 10 2 6 0 3 26 1 2 8 6 0 2 2 5 4 3 0 1 3 5 6 1 0 2 27 1 1 6 4 8 0 2 2 6 2 6 0 3 4 3 1 6 0 28 1 2 5 1 0 8 2 5 3 1 0 7 3 10 3 1 0 4 29 1 4 5 7 0 9 2 8 1 3 0 3 3 8 4 4 7 0 30 1 2 7 8 0 3 2 8 5 4 0 3 3 9 3 4 5 0 31 1 5 8 3 0 4 2 7 8 2 3 0 3 7 7 3 0 3 32 1 0 0 0 0 0 *** RESOURCEAVAILABILITIES: R 1 R 2 N 1 N 2 32 27 48 59 *** REQUESTS/DURATIONS: jobnr. mode duraon R 1 R 2 N 1 N 2 ------------------------------------------------------------------------ 1 1 0 0 0 0 0 2 1 1 8 2 0 3 2 6 8 2 0 2 3 8 6 2 0 2 3 1 2 3 7 3 0 2 4 2 6 3 0 3 5 2 6 0 9 4 1 4 9 8 0 10 2 7 7 5 0 6 3 9 3 4 0 4 5 1 4 7 6 0 3 2 5 7 5 0 3 3 8 5 5 0 1 6 1 1 9 8 7 0 2 4 9 8 5 0 3 9 9 6 3 0 7 1 4 9 5 0 4 2 6 9 4 7 0 3 7 9 4 6 0 8 1 1 8 9 6 0 2 5 7 8 5 0 3 9 5 8 4 0 9 1 5 6 6 0 7 2 7 2 4 0 5 3 7 2 2 0 6 10 1 7 8 9 0 10 2 9 7 7 0 9 3 10 6 6 1 0 11 1 2 9 8 6 0 2 6 7 6 0 1 3 9 3 3 4 0 12 1 1 2 3 6 0 2 3 1 3 0 1 3 10 1 2 5 0 13 1 1 5 10 0 6 2 1 4 10 0 8 3 3 2 9 0 6 14 1 3 7 4 0 3 2 5 6 4 4 0 3 10 5 3 0 2 15 1 2 7 3 0 6 2 6 6 3 6 0 3 10 5 2 3 0 16 1 4 10 10 5 0 2 6 9 9 0 5 3 8 7 9 3 0 17 1 3 8 4 8 0 2 5 7 4 7 0 3 9 6 4 5 0 18 1 4 5 7 5 0 2 5 5 7 4 0 3 8 3 6 3 0 19 1 4 4 7 0 3 2 4 4 6 7 0 3 7 3 6 5 0 20 1 1 9 5 0 10 2 3 8 4 0 10 3 10 7 4 0 10 The problem of employee allocaon s o assgn employees o suable asks so ha he asks can be done effcenly1]. Suppose m employees are nvolved n he projec, for he h employee (=1,2,3..;m) he followng arbues are consdered. s b =Basc salary for he employee per user defned me perod. ISSN: 2231-2803 hp://www.jcjournal.org Page284
Inernaonal Journal of Compuer Trends and Technology (IJCTT) Volume 18 Number 6 Dec 2014 Begn Inalzaon { jon, leave } =The me wndow when he employee s avalable for he projec. 1 2 5 { sk, sk,... sk } =The skll ls for he employee wh 5 ypes of sklls and skll. sk s he Profcency score of he h =1 -h an consrucs a soluon by frs buldng a ask ls and allocaon marx Local parameer updaon Local refnemen,ebs and objecve funcon evaluaon Le us suppose ha he h employee devoes hrs( s ) hours o he projec a he h monh ( hrs( s ) max ). If hrs( s ) s larger han he legal normal workng hours nw, mples ha he employee works overme for he projec. The salary sal for he h employee a he h monh s calculaed by 1) sal = bs hrs( s )* sh hrs( s ) nw 2) sal = b nh* s ( hrs( s ) nh)* s, nh< hrs( s ) max 3) s h o sal =, hrs( s ) max Suppose wh j s he number of workng hours of he h employee for j a he h monh d 1,2,3,4,5 Where s sk =-1 No >POPSIZE TRADITIONAL EXPERIMENTAL RESULTS Yes Check bes soluon and local muaon Global updaon Fnsh s h = Salary of he employee per hour normal work. s o =The salary for he employee s per-hour overme work. nw =Legal normal workng hours per monh. max =Maxmum possble workng hours per monh of he employee for he projec. Fg 1. Home Page of ACO based Even Schedulng ISSN: 2231-2803 hp://www.jcjournal.org Page285
Inernaonal Journal of Compuer Trends and Technology (IJCTT) Volume 18 Number 6 Dec 2014 Job Label :Projec 4 Job Successor Ls :[[Job-159]] Mean Value :383015 -Tes Value :-6.0 Job Info :[Job-159] Job ID :159 Job Type :SINK Job Label :Projec 4 Job Successor Ls :[] Mean Value :273991 -Tes Value :-7.0 Fg 3. Task relaed Daa Loadng Fg 4. Inal Job Schedulng Before ACO approach Fg 5. Task Vs Days To complee Job Successor Ls :[[Job-159]] Mean Value :146023 -Tes Value :-27.0 Job Info :[Job-158] Job ID :158 Job Info :[Job-45] 2014-12-23 11:30:00,950 [SwngWorker-pool-3-hread-1] DEBUG LS sep (505), me spen (12913), score (0/- 5247/-1427), bes score (0/-3769/-1199), acceped/seleced move coun (4/7), pcked move ([Allocaon-17] => [ExecuonMode-49]). Job ID :45 Job Label :Projec 1 Job Successor Ls :[[Job-53]] Mean Value :443562 -Tes Value :-52.0 Job Info :[Job-46] Job ID :46 Job Label :Projec 1 Job Successor Ls :[[Job-48], [Job-52], [Job-60]] Mean Value :470019 -Tes Value :-71.0 Job Info :[Job-47] 2014-12-23 11:30:00,950 [SwngWorker-pool-3-hread-1] DEBUG LS sep (506), me spen (12913), score (0/- 5208/-1427), bes score (0/-3769/-1199), acceped/seleced move coun (4/6), pcked move ([Allocaon-110] => 273). Job ID :47 Job Label :Projec 1 Job Successor Ls :[[Job-48], [Job-49], [Job-60]] Mean Value :265098 -Tes Value :-30.0 Job Info :[Job-48] Job ID :48 Job Label :Projec 1 Job Successor Ls :[[Job-55]] Mean Value :203489 -Tes Value :-10.0 Job Info :[Job-49] Job ID :49 Job Label :Projec 1 Job Successor Ls :[[Job-51], [Job-52], [Job-62]] Mean Value :333782 -Tes Value :-62.0 Job Info :[Job-50] Job ID :50 Job Label :Projec 1 Job Successor Ls :[[Job-55], [Job-56]] Mean Value :478423 -Tes Value :-76.0 ISSN: 2231-2803 hp://www.jcjournal.org Page286
Inernaonal Journal of Compuer Trends and Technology (IJCTT) Volume 18 Number 6 Dec 2014 CONCLUSION Job Info :[Job-51] Job ID :51 Job Label :Projec 1 Job Successor Ls :[[Job-53]] Mean Value :267179 -Tes Value :-57.0 JobName Mean Projec1 176043 Projec2 112043 Projec3 186096 Projec4 166023 Projec5 196023 Projec6 123123 Mean Access Tme Vs Projecs The man objecve of he hs paper s frs, he mehod akes advanage of ACO o solve he complcaed plannng problem, he second one mehod nroduces an even-based scheduler. Boh mehods have lmaon durng he projec plannng and allocaon. Expermenal resuls show ha he represenaon scheme wh he EBS s effecve n small arge asks, and he ACO algorhm manages o yeld beer plans wh hgh sasc- and mean access me and more sable workload assgnmens compared wh oher exsng approaches. A new mehod for solvng he sofware projec plannng problem has been proposed n fuure work. The problem of ask pre-empon exss n he prevous models. The exsng sysem also suffers from he problem of allocang he same ask for dfferen group of employees n dfferen perods. ACO solve he problem of projec schedulng bu does no consder he employee allocaon marx. The ACO s no a sasfacory model o solve he problem of projec schedulng. Mean Access Tme Sasc - 250000 200000 150000 100000 50000 40 20 0-20 -40-60 0 Projec1 Projec2 Projec3 Projec4 Projec5 Projec Names Projec6 JobName Sasc- Projec1-23 Projec2-45 Projec3 0 Projec4 24 Projec5-45 Projec6 29 Sasc- Vs Sascal Parameer Projec1 Projec2 Projec3 Projec4 Projec5 Projec Names Projec6 Mean Sasc- REFERENCES 1. An Colony Opmzaon for Sofware Projec Schedulng and Saffng wh an Even-Based Scheduler, We-Neng Chen, Member, IEEE, and Jun Zhang, Senor Member, IEEE, IEEE TRANSACTIONS ON SOFTWARE ENGINEERING, VOL. 39, NO. 1, JANUARY 2013. 2. C. Wang, S. Lu, Resource consraned consrucon projec schedulng model for prof maxmzaon consderng cash flow. J. of Auo. n Con. 17 (2008) 966-974. 3. L. Dreze, J. Bllau, A projec schedulng problem wh labor consrans and me-dependen acves requremens. In. J. of Pro. Eco. 112 (2008) 217-225. 4. H. Abdalla, Hassan, Usng An colony opmzaon algorhm for solvng projec managemen problems. J. of Exp. Sys. wh App. 36 (2009)10004-10015. 4. S. Chrsodoulou, G. Ellnas, P.Aslan, Enropy-based schedulng of resource- consraned consrucon projecs. J. of Auo. In Con,.18 (2009) 919-928. 5. V. Thagarasu, T. Dev, Mul agen coordnaon n projec schedulng: Prory rules based resource allocaon. In. J. of Rec. Tr. n Eng. 1 (2009) 314-326. 6. Vahd Khodakaram, Norman Fenon, Marn Nel, Projec Schedulng:Imposed Approach o Incorporae Uncerany usng Bayesan Neworks (2007). 7. Carl K.Chang,Hsn-Y Jang,Yu D,Dan Zhu,Yuja Ge, Tme-lne based model for sofware projec schedulng wh genec lgorhms (2008) 8. Andrey Glaschenko, Anon Ivaschenko George Rzevsk,Per Skobelev, Mul-Agen Real Tme Schedulng Sysem for Tax Companes (2009) 9. M. Dorgo, V. Manezzo, A. Colorn,, An sysem: opmzaon by a colony of cooperang agens, IEEE Transacons on Sysems Man, and Cybernecs- par B: Cybernecs, vol.26, pp.29-41, 1996 10 R.-G. Dng and X.-H. Jng, Fve prncples of projec managemen n sofware companes, Projec Managemen Technology (n Chnese), vol.1, 2003. ISSN: 2231-2803 hp://www.jcjournal.org Page287
Inernaonal Journal of Compuer Trends and Technology (IJCTT) Volume 18 Number 6 Dec 2014 11 L.C. Lu and E. horowz, A formal model for sofware projec managemen, IEEE Transacons on sofware Engneerng, vol.15, no.10, pp. 1280-1293, 2001. 12 D. Merkle, M. Mddendorf, H. Schmeck, An colony opmzaon for resourceconsraned projec schedulng, IEEE Transacons on Evoluonary Compuaon, vol. 6, no. 4, pp. 333-346, 2002. P.vdya Sagar obaned hs M.C.A from Vsveswaraah Technologcal Unversy,Belgaum.Then he obaned hs M.Tech n Compuer Scence And Engneerng from Acharya Nagarjuna Unversy,Gunur and pursung PhD n Compuer Scence and Technology from Sr Krshnadevaraya Unversy,Ananapuram. He s a Professonal Member of ISCA. Hs specalzaons nclude sofware engneerng and sofware relably, web servces and neworkng. Dr.N.Geehanjal receved her PhD Degree from Sr Krshanadevaraya Unversy. Andhra Pradesh, Inda she s workng as Head, Deparmen of Compuer Scence & Technology, Sr Krshanadevaraya Unversy. Andhra Pradesh, Inda. She s a Professonal Member ACM. Her research neres ncludes Compuer Neworks, Cloud Compung, Sofware Engneerng, Programmng languages and Daa Mnng. ISSN: 2231-2803 hp://www.jcjournal.org Page288