Research Article An Optimization Algorithm for Multipath Parallel Allocation for Service Resource in the Simulation Task Workflow

Transcription

1 e Scientific World Journal, Article ID 50757, pages Research Article An Optiization Algorith for Multipath Parallel Allocation for Resource in the Siulation Task Workflow Zhiteng Wang, Hongjun Zhang, Rui Zhang, Yong Li, 2 and Xuliang Zhang PLA University of Science & Technology, Nanjing 20007, China 2 Nanjing Artillery Acadey, Nanjing 200, China Correspondence should be addressed to Zhiteng Wang; zhitengwang@sina.co Received 9 October 203; Accepted 25 Noveber 203; Published 2 May 204 Acadeic Editors: W. Sun, G. Zhang, and J. Zhou Copyright 204 Zhiteng Wang et al. This is an open access article distributed under the Creative Coons Attribution License, which perits unrestricted use, distribution, and reproduction in any ediu, provided the original work is properly cited. oriented odeling and siulation are hot issues in the field of odeling and siulation, and there is need to call service resources when siulation task workflow is running. How to optiize the service resource allocation to ensure that the task is coplete effectively is an iportant issue in this area. In ilitary odeling and siulation field, it is iportant to iprove the probability of success and tieliness in siulation task workflow. Therefore, this paper proposes an optiization algorith for ultipath service resource parallel allocation, in which ultipath service resource parallel allocation odel is built and ultiple chains coding schee quantu optiization algorith is used for optiization and solution. The ultiple chains coding schee quantu optiization algorith is to extend parallel search space to iprove search efficiency. Through the siulation experient, this paper investigates the effect for the probability of success in siulation task workflow fro different optiization algorith, service allocation strategy, and path nuber, and the siulation result shows that the optiization algorith for ultipath service resource parallel allocation is an effective ethod to iprove the probability of success and tieliness in siulation task workflow.. Introduction With the developent of web technology, web service technology, SOA, grid, and cloud coputation, there is a great revolution in odelling and siulation field. A fraework of odelling and siulation based on siulation service, that is, service oriented odelling and siulation, is being fored. oriented siulation task fraework fro service oriented idea depends on inforation grid infrastructure with unified siulation description, access, and share standard to coplete siulation task with the way of dynaic creation and running in the siulation resource interlink and share environent. Copared with traditional siulation fraework, one of the service oriented odeling and siulation fraework ain characteristics is supporting dynaic creation and running siulation application and dynaic integration and running siulation in network by calling siulation service resource according to the siulation task workflow need. In service oriented ilitary odeling and siulation field, it is iportant to iprove the probability of success and tieliness in siulation task workflow. Many researchers have done lots of research on how to iprove the probability of success and tieliness in service coposition in any stages. Soe researchers try to iprove the probability of siulation task workflow work success instead of failed service [ 3], but this ethod does not consider the tielessness of workflow. Therefore, it will produce extra tie charge when the called service cannot respond or akes a istake. Artificial intelligence (AI) planner is also applied to solve service coposition cobinatorial optiization proble [4 7]. But for the reason that it could iensely increase coupling between AI planner and other procedures when the special AI planner is used to solve autoatic service coposition proble, it will lead to liiting its web coposition application filed. In addition, there are any differences in AI planner for applicability and designs perforance, and this reason will result in biased service coposition schee and larger coposition charge. Yajuan Song proposes a blend web service coposition schee based on buffer pool [8], in which dynaic service coposition schee is saved in buffer pool and user can use the coposition schee as well as use predefined workflow. But the coposition algorith should

2 2 The Scientific World Journal be iproved. ZHANG Bo investigates a service coposition algorith based on sub-web service, in which the ultiinput and ultioutput paraeter of web service is divided into sub-web service with ulti-input and single output paraeters. In this way, it reduces dependencies of a Web service for ultiple outputs with higher coposition efficiency. In this way, we reduce dependencies of a Web service on ultiple-output with higher coposition efficiency; however, there exists redundancy in the coposition route. Genetic algorith (GA) and particle swar optiization (PSO) [9] are typical evolution algorith in service coposition optiization filed, and PSO is ore effective than GA with advantages such as too few arguents specified and the faster convergence speed. However, the local particle in the swar is easy to be regarded as global best position leading to quickly converging to local best value. Zheng et al. [0]andZengetal. []useobjectiveprograingethodtofindthebestpath satisfied with condition by building the objective function with the condition and the probability of success and tie requisition. The shortcoings of which ay not coplete the workflow requests. Xiaohao et al. propose parallel allocation algorith for service [2], in which one node starts any services to iprove the probability of success and tie requests in the siulation task workflow. The ethod gets good result but there exists exception probability in large-scale siulation task. Therefore, based on service parallel allocation in [2], this paper extends the schee and proposes an optiization algorith for ultipath service resource parallel allocation schee to further iprove the probability of siulation task workflow work success and tie and solve service scheduling proble. The following section will investigate the systeatic structure of siulation task body fro two perspectives, thatis,thereferenceodelandthefraeworkstructureof siulation task body. In service oriented ilitary odel and siulation field, it is iportant to ensure the tieliness and the probability of success in the siulation task workflow or it will greatly affecttheworkflowworkefficiencyandevenleadtoworkflow which cannot coplete the task. In fact, it could iprove the tieliness and the probability of success by the optiization of the service coposition in siulation task workflow. However, any optiization algoriths have deficiencies in solving this proble. For this reason, this paper proposes an optiization algorith for ultipath service resource parallel allocation for service resource to iprove the tieliness and the probability of success in the siulation task workflow. 2. The Description of Coposition Proble 2.. The Concept of Scheduling and Siulation Task Workflow Fraework. The concept of workflow originated fro the organization of production and office autoation field, in which work was separated into well-defined task, role, and the work is executed and onitored by specific rule and process to iprove work probability and reduce production costs and raise enterprise copetitive power. In ilitary service oriented odel and siulation field, workflow idea is introduced to arrange the role produce specific event under the siulation plot rule in siulation task. The event of role is related to siulation service agent, which is a service set and could call the siulation service by siulation service bus. The specific siulation service in service sets is eventual execution unit, and the purpose of service scheduling is to find the best path to allocate the siulation service in service sets to specific event of role in siulation plot. It can be shown in Figure. The scheduling is to build a apping relation fro siulation event sets to siulation service sets by the scheduling algorith. Its purpose is to choose the best service coposition path under the siulation task liited condition to effectivelycopletethesiulation.inthefollowingpart,a siulation task exaple of artillery firepower attacking is described. In this siulation task plot, artillery firepower attackingstepisasfollows. () The artillery needs to get eney s position by radar search service. (2) The artillery needs to get eney s global situation by situation analysis service. (3) The artillery needs to prepare for firing by firing preparation service. (4) The artillery needs to fire allocation schee by fire allocation service. In order to iprove tieliness and the probability of success in siulation task, the probability of success and response tie should be constrained. In the first phase, the longest echo tie (LET) of radar service should be less than 0 seconds and the success call probability (SCP) should be ore than 98%. In the second phase, for situation analysis service and firing preparation service, LET is required less than 30 seconds and SCP is required ore than 98%. In the third phase, for situation analysis service and firing preparation service, LET is required less than 30 seconds and SCP is required ore than 98%. For the whole of siulation task, LET is less than 40 seconds and SCP is required ore than 98%.ItcanbeshowninFigure The Description of Scheduling Model. Siulation task plot is used to arrange role copletion action in specific tie and position and it is iportant easure for workflow service scheduling. In fact, service scheduling is to optiize the siulation service coposition, and service scheduling odel can be siply described as the relation of siulation service coposition optiization according to the siulation plot. Therefore, service scheduling optiization odel can be described as: P = {S,P,C} wherein S is representative of the service sets according to the event in plot, S={S,S 2,S 3,...,S n }; P is representative of the apping fro the event relation to siulation service relation; C is representative of the constrained condition in the whole process The Optiization of Coposition Path. The service coposition path is selected one by one fro siulation

3 The Scientific World Journal 3 Workflow engine Event 4 Event Event 2 Role 4 Event 5 Role Role 2 Event 3 Role 5 Role 3 agent agent 2 agent 3 agent 4 agent n 2n 3n 4n 5n Siulation service bus Figure : Siulation task workflow running fraework. Radar search service Situation analysis service Firing preparation service Fire allocation service LET <0 LET <30 LET<5 SCP >98% SCP >98% SCP >98% LET <40 SCP >98% Figure 2: The service coposition odel under condition.

4 4 The Scientific World Journal S S 2 S 3 S 4 coposition path sets n 2n 3n 4n Figure 3: coposition path choosing. service sets to for a service coposition path to coplete siulation task. It can be shown in Figure 3. Fro Figure 3,wecanseethatanypathscanbeselected by the ethods of exhaustion, intelligence algorith, and so forth to copile statistics quality of service and get the service coposition path. However, these ethods cannot ensure that the path is the best choice, because these ethods are only considering the ideal condition without considering unexpected exception in practice. 3. The Strategy of Multipath Parallel Allocation for When the siulation service is called, various probles ay occur in the siulation task workflow during run tie and the task ay not be copleted. In order to solve this proble, any researchers have taken a variety of approaches such as failed service instead, recall service again, and reselection of the path of service coposition. The ethod of failed service instead is to replace the siulation service which is unable to continue to provide the service in process of the siulation task, and then another appropriate siulation service which has the sae function in the service sets is called to coplete the service function which the exception siulation service should coplete. It can be shown in Figure 4. The ethod of recall service is to recall service when the siulation service cannot coplete the appropriate service function and surpass axiu response tie. This ethod can be applied when the siulation service in service sets is not enough or it requires a higher liitation of siulation to use. Because the tie expense for this ethod is sall, if we recall the service again, it will restore the function proptly, so the siulation task workflow ay restore the work in the shortest tie. The ethod of reselection of the path of service coposition is to choose a service coposition path again under the constrainedconditionandthenrunthesiulationtaskagain. This ethod s tie expenses are large. Because it eans the siulation service in original path should coplete the task againinthenewservicecopositionpath,sothisethod is used in the situation in which both of the expiration substitution ethod and recall service ethod are unable to coplete the siulation task. In fact, although these ethods ay reduce the influence to soe extent on the siulation task workflow when the siulation service akes the istake, it also takes tie expenses as price. This is not peritted for real tie siulationtask,soitneedstouseaorenibleethodtosolvethis proble. In order to iprove the probability of success and tieliness in siulation task workflow, this paper proposes an optiization algorith for ultipath service resource parallel allocation to handle service failure to be called in the service coposition process, and each path is a parallel distribution services allocation plot. In the practice, various traditional ethods such as failed service instead and recall service can be used synthetically. For exaple, two paths are selectedandeachnodehastwoparallelservicesineachpath. ItcanbeshowninFigures5(a) and 5(b). This ethod is siultaneously calling any paths, and in each path siulation service siultaneously is called with any parallel services in service sets. The shortest response tie of the first path in a nuber of paths can be selected as the ain ipleentation of the path, in which anyservicecanbeselectedasaainservicewhenithas theshortestresponsetie.ifotherainserviceisfailed, other parallel service will be selected as ain service. The difference fro traditional instead ethods is the parallel services running, and the workflow work cannot be effected

5 The Scientific World Journal 5 coposition path S S 2 S 3 S sets n 2n 3n 4n Figure 4: Failed service instead sketch. even if one service is exception. Therefore, it nearly does not spend other expenses. In particular, because any paths run siultaneously, if the service in the ain path produces an unexpected result, the service in another path can also run instead of it in tie to effectively iprove probability of success and tieliness for siulation task. 4. The Model of Multipath Parallel Allocation Suppose there exist n agents in accordance with n service sets in workflow. S i (i =,2,3,...,n) represents the service in accordance with the ith service agent. S ij (j =,2,3,...,) represents the ith service agent in accordance with services in siulation service bus and the j represents the jth service in services called. Suppose the probability of success service S ij is f(i, j) and its copletion tie follows a noral distribution under ean as u ij and variance as σ 2 ij. When the workflow siultaneously call service in S i,itsprobability can be shown in the following forula: Suc (i) = ( f (i, j)). () M i = Min(t i,,t i,2,...,t i, ) is used to represent tie distribution function in accordance with the ith service agent, where t i,j is the ith service in services called by service agent. It is noral distribution and can be shown in the following forula: T i (t, j) = Suc (M i t) = Suc (M i t) = Suc [ (t i.j t)] [ ] = = = = Suc (t i,j t) [ Suc (t i,j <t)] [ T i,j (t)] t [ e (t u i,j) 2 /2σ 2 i,j dt]. 2πσ i,j The success probability of siulation service S i in the ith siulation in the waiting tie of d i is as follows: SucT i (d i,i)= Suc (i) T i (t, j) = [ [ [ [ 0 ( f (i, j))] ] ( 2πσ i,j (2) d i e (t u i,j) 2 /2σ 2 i,j dt)]. 0 (3)

6 6 The Scientific World Journal S S 2 S 3 S 4 coposition sets n 2n 3n 4n (a) S S 2 S 3 S 4 coposition n 2n 3n 4n sets (b) Figure 5: (a) The first path for ultiple service parallel allocation. (b) The second path for ultiple service parallel allocation. Suppose n events according to n services in a service coposition path. Let =, 2,..., n,where i (i =,2,...,n) represents the nuber of parallel called services in this node. Suppose siulation service bus could support A i ( i A i ) service for the ith node, and the syste expense in the parallel running i servicescanbeshownbythe following forula: C i ( i ) =e ( i/a i ) 2. (4) Therefore, the whole process of parallel service running expense can be shown as C () = n i= e ( i/a i ) 2. (5) The ultiate optiization goal is to select proper which is satisfied with constraint condition for service tie and probability, and the service running expense is the lowest. Suppose the weighting for the ith service in the whole service

7 The Scientific World Journal 7 is W i, n i= W i =and the ultiate optiization odel can be represented as Min (c ()) = Min W i e ( i/a i ) 2 n i= s.t. SucT j (SucT (d, ),SucT 2 (d 2, 2 ) SucT n (d n, n )) > SCP j 0<T j (d,d 2,...,d n )<LET j. 5. Scheduling Optiization Based on Quantu Optiization Algorith Infact,theprobleofultipleservicesparallelallocationis service resource scheduling proble which is a NP proble. Many ethods have been proposed to solve this proble: GA, PSO, and so forth. However, there always exist any disadvantages in partial convergence or slow optiization. Quantu coputation is based on the principal concepts of the quantu theory [3, 4]. Nuerous researchers have devoted increasing interests to quantu coputation, a novel interdisciplinary field that covers quantu echanics and inforation science [5 24]. This paper tries to use a quantu optiization algorith with ultiple chain coding schees to solve this proble and ainly use fourchain quantu-inspired evolutionary algorith (FCQIEA) to solve. 5.. Expanded Encoding Method for Quantu Chroosoe. Fro [24] we can know that ultiple chains can be required in Figure 6. Fro Figure 6, we can get(7) as follows: φ>=[cos θ, cos θ sin θ, sin θ sin θ]t. (7) To describe the quantu dynaics behavior objectively, coprehensively, and unabiguously, we can use a new angle φ (0 <φ<π), which is called supporting role, to replace θ and obtain vector sin θ as follows: [sin φ sin θ, cos φ sin θ, cos θ] T. (8) Equation (8) also satisfies the noralization condition. In fact, (8) also corresponds to the three-chain encoding ethod: p i = sin φ i sin θ i cos φ i sin θ i cos φ i sin φ in sin θ ij cos φ in sin θ ij cos φ ij (6). (9) Likewise, we can obtain vector sin φ sin θ by adding the supporting role β to for the four-chain encoding ethod as follows: [cos β cos φ sin θ, sin β sin φ sin θ, cos φ sin θ, cos θ] T, cos β i sin φ i sin θ i cos β ij sin φ ij sin θ ij sin β p i = i sin φ i sin θ i sin β ij sin φ ij sin θ ij. cos φ i sin θ i cos φ ij sin θ ij cos θ i cos θ ij (0) C sinθ D A θ θ B cosθ φ> Figure 6: Scheatic diagra of the qubit probability aplitude decoposition. We can obtain four optial solutions, which are expressed as follows: p i =(cos β i sin φ i sin θ i,...,cos β in sin φ in sin θ in ) p i2 =(sin β i sin φ i sin θ i,...,sin β in sin φ in sin θ in ) p i3 =(cos φ i sin θ i,...,cos φ in sin θ in ) p i4 =(cos θ i,...,cos θ in ). () With the sae principle,we can get N+chains coding schee as follows: p i = (sin θ n ) i (sin θ n ) i (sin θ 2 ) i (sin θ ) i (cos θ n ) i (sin θ n ) i (sin θ 2 ) i (sin θ ) i. (cos θ 3 ) i (sin θ 2 ) i (sin θ ) i (cos θ 2 ) i (sin θ ) i (cos θ ) i (sin θ n ) ij (sin θ n ) ij (sin θ 2 ) ij (sin θ ) ij (cos θ n ) ij (sin θ n ) ij (sin θ 2 ) ij (sin θ ) ij. (cos θ 3 ) ij (sin θ 2 ) ij (sin θ ). ij (cos θ 2 ) ij (sin θ ) ij (cos θ ) ij (2) 5.2. Solution Space Transforation. In the quantu evolution progress, all qubits have liited values within to.

8 8 The Scientific World Journal Thus, we need to transfor all qubit values fro unit space I n = [,] n to space Ω of the continuous optiization proble () by using linear transforation. Each gene value corresponds to an optiization variable in the solution space. If thejth qubit on chroosoe p i is [x 4 ij,x3 ij,x2 ij,x ij ], then the corresponding variables in the solution space are coputed as follows: X j i = 0.5 [b j ( + x ij )+a j ( x ij )] X j i2 = 0.5 [b j ( + x ij )+a j ( x ij )] X j i3 = 0.5 [b j ( + x ij )+a j ( x ij )] X j i4 = 0.5 [b j ( + x ij )+a j ( x ij )], (3) where i =, 2, 3,..., and j =, 2, 3,..., n. Thus, each chroosoe aps to four approxiate solutions of the optiization proble. u 34 = [sin Δθ (cos φ cos Δφ sin φ sin Δφ)] T, u 4 =u 42 =0, u 43 = sin Δθ cos φ, u 44 = cos Δθ, U [ cos β sin φ sin θ sin β sin φ sin θ [ cos φ sin θ ] [ cos θ ] cos (β + Δβ) sin (φ + Δφ) sin (θ+δθ) sin (β + Δβ) sin (φ + Δφ) sin (θ+δθ) =. cos (φ + Δφ) sin (θ+δθ) [ ] [ cos (θ+δθ) ] (4) 5.3. Quantu Chroosoe Update. Considering that quantu chroosoes are present in the colony and we can obtain 4 approxiate solutions by solution space transforation, we can then copute the fitness of these approxiate solutions and define the solution with the axiufitnessasthecurrentoptiusolutioninthequantu evolution progress. The chroosoe corresponds to the current optiu solution called the optiu chroosoe. By coputing the fitness, we can obtain both optiu solution and optiu chroosoe and subsequently update the colony by using the quantu rotation gate to obtain the optial solution. In this updated process, the new optiu chroosoecanbeproducedsuchthatthecolonycanlikely evolve. The present study proposes the quantu rotation gate U to update the individual qubit as follows: U= [ u u 2 u 3 u 4 u 2 u 22 u 23 u 24 [ u 3 u 32 u 33 u 34 ], [ u 4 u 42 u 43 u 44 ] u =[cos Δβ cos Δφ (cos Δθ T sin Δθ cos θ )], sin θ T sin Δθ cos θ u 2 =[sin Δβ cos Δφ ( cos Δθ)], sin θ u 3 =[sin Δφ cos Δθ (sin β sin Δβ cos β cos Δβ)] T, u 4 =[cos φ sin Δφ (cos β cos Δβ sin β sin Δβ)] T, u 2 =[ sin Δβ sin Δφ (cos Δθ + T sin Δθ cos θ )], sin θ T sin Δθ cos θ u 22 =[ cos Δβ sin Δφ (cos Δθ + )], sin θ u 23 =[cos Δφ cos Δφ (sin β cos Δβ + cos β sin Δβ)] T, u 24 =[cos φ cos Δφ cos Δφ (sin β cos Δβ + cos β sin Δβ)] T, sin Δφ u 3 = cos β, u 32 =0, u 33 = cos Δφ cos Δθ, 5.4. Mutation Operation. Quantu nongate is applied to exchange the probability aplitudes to avoid local optial solution in a certain qubit as follows: [ 0 θ θ ][cos ]=[sin ]. (5) 0 sin θ cos θ Such influence as expressed in (5) can be considered as the phase utation of a qubit, in which θ is utated to (π/2) θ.inthiscase,aquantunongatev is proposed to utate the quantu as follows: V= [ tan β cotφ cotθ tan β cotφ cotθ 0 0 [ 0 0 tan φ cotθ 0 ]. [ tan θ] (6) 5.5. The Procedure of FCQIEA. It can be suarized as follows. Step (initialize the population). Let the current generation t=0; generate an initial population Q(t) = {q t,qt 2,...,qt }, which has individual qubits. Set the agnitude of the rotational angle Δβ = β 0, Δφ = φ 0,and Δθ = θ 0, respectively. Set p as the utation and Max gen as the axiu generation. Step 2 (transfor the solution space). Four approxiate solutions in each chroosoe are transfored fro the unit space I n = [,] n to the solution space Ω of the continuous optiization proble (); thus, the set of approxiate solution X(t) can be obtained. Step 3 (copute the fitness). By coputing the fitness of 4 approxiate solutions, obtain the best solution BestX in the current solution and the best chroosoe BestC in the current chroosoe. Store BestX as the global optiu solution GX and store BestC as the global optiu chroosoe GC.

9 The Scientific World Journal 9 running costs Probability of success FCQIEA GA Global tie constrained PSO Figure 7: Coparison of the optiization result with different algoriths. Step 4 (set t=t+). Update and utate Q(t ).Calculate the new population Q(t). Step 5. Transfor the solution space again and obtain a set of approxiate solution X(t). Step 6. By coputing the fitness of Q(t), deterine the current optiu solution BestX and the current optiu chroosoe BestC.If fit(bestx) < fit(gx),then update the current optiu solution BestX = GX; at the sae tie, update the current optiu chroosoe BestC = GC to avoid population degradation. Otherwise, let GX = BestX and GC = BestC so that the algorith approaches the global optiu solution. Step 7. If the algorith does not converge and if t < Max gen, then go back to Step 4 until the algorith becoes convergent or until t>max gen. 6. Siulation Experient and Analysis 6.. Coparisons of the Different Optiization Result. Under the hypothesis condition as Figure 2, four events are in accordance with service as S, S 2, S 3,andS 4, and the specific constraint conditions can be shown as in Table. In order to copare the different optiization results between different optiization algoriths, FCQIEA, GA, and PSO, optiize the service scheduling, respectively. In both of FCQIEA and GA, the generation set is 00 and intersection probability and utation set are 0.6 and 0., respectively.thenuberofcodingchainsis4,andaxiu rotate angle set Δθ = 0.04π. In PSO, regulatory factor set w = 0.5 and study factor set C = 0.7 and C 2 =.2. In order to avoid the influence of rando factors, we perfored the calculation 0 ties in each algorith and obtained the average value as the optiization result. The global constraint tieofeachservicenodeisaddedwith5secondsasbaseunit; the optiization result can be shown in Figure Waiting tie (s) Multipath services parallelallocation strategy allocation strategy Best service selection strategy Failed service instead strategy Figure 8: The probability of success of different service allocation strategy. Task execution costs Multipath service parallel Failed service Best service parallel allocation allocation instead selection Figure 9: Coparison of task execution costs with different service allocation strategy. Fro Figure 7,obviously,FCQIEAhasbestperforance copared with GA and PSO and its service running costs ost rapid decline than other algoriths Coparisons of the Allocation Strategy. The different service allocation strategy, that is, selection best service, failed service instead, services parallel allocation, and ultipath services parallel allocation, is copared under the constraint condition in Table.The failed service is produced randoly with probability of %. The failed service instead strategy is that the service can be called failure when waiting tie is beyond the 3 ties of δ (3 ties δ rule) service execution tie and it should select another service to replace the failed service. In service parallel allocation strategy, 20 percent of service sets are randoly selected as parallel execution service. In the ultipath service parallel allocation strategy,thenuberofpathset3and20percentofservice sets are randoly selected as parallel execution service. After long running tie, the relation between the probability of success and execution of different service allocation strategy is investigated. It can be shown in Figure 8.

10 0 The Scientific World Journal Constrained condition Execution tie distribution Probability of success The nuber of parallel services The axiu waiting tie [5s,5s] unifor distribution A rando nuber fro 0.94 to Table : constrained condition. node S S 2 S 3 S 4 [0 s, 20 s] unifor [0 s, 25 s] unifor distribution distribution A rando nuber fro 0.94 to A rando nuber fro 0.92 to [2s,8s] unifor distribution A rando nuber fro 0.93 to The iniu probability of 98% 98% 98% success 98% W i The probability of success siulation task execution Waiting tie in task excution (s) Three paths for service parallel allocation One path for service parallel allocation Two paths for service parallel allocation Four paths for service parallel allocation Figure 0: Coparison of the effect for the probability of success siulation task with different paths. Different service execution costs can be shown in Figure 9. Fro analysis, it can be found that the probability of success of ultipath service parallel allocation strategy is obviously higher than other strategies; however, it also results in extra syste costs than other service allocation strategies The Effect of Probability of Success fro the Nuber of Paths. parallel allocation and ultipath services parallel allocation are used to execute siulation task and the nuberofpathsis,2,3,and4.thefailedserviceisproduced randoly with probability of %. After running in long tie, the relation between the probability of success and execution tie is investigated. It can be shown in Figures 0 and. Fro Figures 0 and, we can get the conclusion that we can iprove probability of success by adding the nuber of service paths in service parallel allocation, but it will not always get the best probability of success when increasing the execution costs Four paths Three paths Two paths One path Figure : Coparison of siulation task execution costs aong different paths. nuber of service paths, and there exist the best value. Fro the siulation experient, it can be seen that the probability of success for copletion task is iproved obviously with the nubers of 2 and 3, but when the nuber of path set is 4, the probability of success for copletion task is not iproved and the siulation execution costs are raised. Therefore, three paths are proper. 7. Conclusions In ilitary odeling and siulation field, it is iportant to iprove the probability of success and tieliness in siulation task workflow. This paper established the ultipath service parallel allocation optiization atheatical odel to investigate service scheduling optiization algorith in which FCQIEA is used. Through the siulation experient, this paper investigates the effect for the probability of success in siulation task workflow fro different optiization algorith, service allocation strategy, and path nuber, and the siulation result shows that the optiization algorith for ultipath service resource parallel allocation is an effective

11 The Scientific World Journal ethod to iprove the probability of success and tieliness in siulation task workflow. In the next step, ultipath service parallel allocation will be used in ilitary odeling and siulation field, and this theory can be iproved in practice. Conflict of Interests The authors declare that there is no conflict of interests regarding the publication of this paper. References [] D. König, N. Lohann, S. Moser, C. Stahl, and K. Wolf, Extending the copatibility notion for abstract WS-BPEL processes, in Proceedings of the 7th International Conference on World Wide Web (WWW 08), pp , Beijing, China, April [2] N. Salatge and J.-C. Fabre, Fault tolerance connectors for unreliable Web s, in Proceedings of the 37th Annual IEEE/ IFIP International Conference on Dependable Systes and Networks (DSN 07), pp. 5 60, June [3] J. Harney and P. Doshi, Speeding up adaptation of web service copositions using expiration ties, in Proceedings of the 6th International World Wide Web Conference (WWW 07), pp , May [4] E. Sirin, Cobining Description Logic Reasoning with AI Planning for Coposition of Web s, University of Maryland, College Park, Md, USA, [5] J.RaoandX.Su, Asurveyofautoatedwebservicecoposition ethods, in Proceedings of the st International Workshop on Seantic Web s and Web Process Coposition (SWSWPC 05),vol.3387,pp.43 54,July2005. [6] S.McilraithandT.C.Son, Adaptinggologforcopositionof seantic web services, in Proceedings of the 8th International Conference on Knowledge Representation and Reasoning, pp , Morgan Kaufann, Toulouse, France, [7] S. Narayanan and S. A. McIlraith, Siulation, verification and autoated coposition of web services, in Proceedings of the th International Conference on World Wide Web (WWW 02), pp , ACM, New York, NY, USA, May [8] S. Yajuan, Research on the Approaches for Web s Coposition, Jilin University, Changchun, China, 20. [9] X.Fan,C.Jiang,X.Fang,andZ.Ding, Dynaicwebservice selection based on discrete particle swar optiization, Journal of Coputer Research and Developent, vol.47,no.,pp , 200. [0] H. Zheng, W. Zhao, J. Yang, and A. Bouguettaya, QoS analysis for Web service coposition, in Proceedings of the IEEE International Conference on Coputing (SCC 09), pp , Bangalore, India, Septeber [] L. Zeng, B. Benatallah, A. H. H. Ngu, M. Duas, J. Kalagnana, and H. Chang, QoS-aware iddleware for Web services coposition, IEEE Transactions on Software Engineering,vol.30,no. 5, pp , [2] Y. Xiaohao, H. Dan, L. Xueshan, and L. Junxian, Parallel optiization ethod of service resource in ilitary inforation syste, Systes Engineering Theory & Practice, vol. 32, no. 9, pp , [3] D. Deutsch, Quantu theory, the Church-Turing principle and the universal quantu coputer, Proceedings of The Royal Society of London A: Matheatical and Physical Sciences, vol.400, no.88,pp.97 7,985. [4] M. A. Nielsen and I. L. Chuang, Quantu Coputation and Quantu Inforation, Cabridge University Press, st edition, [5] S.-Y. Li and P.-C. Li, Quantu genetic algorith based on real encoding and gradient inforation of object function, Journal of Harbin Institute of Technology, vol.38,no.8,pp , [6] P. C. Li and S. Y. Li, Quantu-inspired evolutionary algorith for continous spaces optiization based on Bloch coordinates of qubits, Neurocoputing, vol. 72, pp , [7] X. Shaohua, X. Chen, H. Xing, W. Ying, and L. Panchi, Iproved quantu genetic algorith with double chains and its application, Application Research of Copute,vol.26,no.7,pp , 200. [8] S. Li and P. Li, Quantu Coputing and Quantu Optiization Coputing Algorith, Harbin Polytechnical University, [9] K.-H. Han and J.-H. Ki, Quantu-inspired evolutionary algorith for a class of cobinatorial optiization, IEEE Transactions on Evolutionary Coputation,vol.6,no.6,pp , [20] A. D. S. Nicolau, R. Schirru, and A. Alvarenga de Moura Meneses, Quantu evolutionary algorith applied to transient identification of a nuclear power plant, Progress in Nuclear Energy, vol. 53, no., pp. 86 9, 20. [2] J. X. V. Neto, D. L. de Andrade Bernert, and L. dos Santos Coelho, Iproved quantu-inspired evolutionary algorith with diversity inforation applied to econoic dispatch proble with prohibited operating zones, Energy Conversion and Manageent,vol.52,no.,pp.8 4,20. [22] J.Xiao,J.Xu,Z.Chen,K.Zhang,andL.Pan, Ahybridquantu chaotic swar evolutionary algorith for DNA encoding, Coputers and Matheatics with Applications,vol.57,no.-2, pp ,2009. [23] J. Gu, M. Gu, C. Cao, and X. Gu, A novel copetitive co-evolutionary quantu genetic algorith for stochastic job shop scheduling proble, Coputers and Operations Research, vol. 37, no.5,pp ,200. [24] W. Zhiteng, Z. Hongjun, Z. Rui, X. Ying, and H. Jian, Quantu genetic algorith based on ulti-chain coding schee, Coputers & Operations Research, vol.29,no.26,pp , 202.

12 International Journal of Rotating Machinery Engineering Journal of The Scientific World Journal International Journal of Distributed Sensor Networks Journal of Sensors Journal of Control Science and Engineering Advances in Civil Engineering Subit your anuscripts at Journal of Journal of Electrical and Coputer Engineering Robotics VLSI Design Advances in OptoElectronics International Journal of Navigation and Observation Cheical Engineering Active and Passive Electronic Coponents Antennas and Propagation Aerospace Engineering International Journal of International Journal of International Journal of Modelling & Siulation in Engineering Shock and Vibration Advances in Acoustics and Vibration