CN109492872A - Dynamic workflow scheduling method based on genetic algorithm - Google Patents
Dynamic workflow scheduling method based on genetic algorithm Download PDFInfo
- Publication number
- CN109492872A CN109492872A CN201811194823.7A CN201811194823A CN109492872A CN 109492872 A CN109492872 A CN 109492872A CN 201811194823 A CN201811194823 A CN 201811194823A CN 109492872 A CN109492872 A CN 109492872A
- Authority
- CN
- China
- Prior art keywords
- workflow
- individual
- dynamic
- population
- topological structure
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
Abstract
The dynamic workflow scheduling method based on genetic algorithm that the invention discloses a kind of, the target of the dispatching method is the average cost in each period of Optimization Work stream in the case where meeting total execution time-constrain of the execution cycle constraint of maximum specified by user and workflow a cycle.Since executive mode is that dynamic is changeable to workflow under the realization environment of cloud computing, so the present invention stream topologies being likely to occur all to dynamic workflow construct, the stream topologies that each may occur are corresponded to by establishing a series of subgraphs, and Holistic modeling is carried out with probabilistic model, to comprehensively consider the characteristic of the dynamic time-varying of workflow, and the period expense that workflow executes in dynamic environment is optimized using genetic algorithm, to improve the execution efficiency of workflow.
Description
Technical field
The present invention relates to cloud computing and intelligent algorithm technical fields, and in particular to a kind of dynamic based on genetic algorithm
Workflow schedule method.
Background technique
Cloud computing is by the virtual aggregation to a large amount of computing resources and shares, and realization provides a user various on demand
Calculating service, therefore can satisfy growing big data process demand.In order to further increase cloud computing system to big
How rationally, efficiently the management and processing capacity of data dispatched the resource of cloud computing and flexibly provided to realize to user
The service of calculating is the key that improve cloud computing system performance.
In cloud computing environment, the calculating volume of services undertaken by the substantial amounts of cloud computing resources, cloud system also phase
When huge, the use state of cloud network also moment variation, therefore dynamic time variation is that had in cloud computing system operation
Important feature.In application cloud computing processing big data calculating task, a kind of common taskings mode is workflow.
Workflow defining is the specific task sequence of a completion complex target.Normally, workflow can pass through directed acyclic graph
(DAG) form provides, and the node of figure indicates individual task, and between the directed edge expression task between node it is preferential about
Beam relationship.However, the control stream topological structure of a workflow is fixed and invariable in existing workflow schedule model,
It is to be provided by single DAG.In practical applications, the control flow structure of workflow may also have the selections such as IF-THEN point
Branch, control stream topological structure also have the characteristic of dynamic time-varying, how during workflow schedule to consider cloud environment
With the dynamic time-varying characteristics of Work-flow control topology, thus further increase workflow schedule system dynamic, when changing environment in
Availability, new challenge is proposed to workflow schedule method.
With the development of optimisation technique, such as the novel meta-heuristic intelligence computation method of genetic algorithm is complex optimization
The solution of problem provides new effective means.Genetic algorithm is the evolutionary phenomena for simulating nature biotechnology and one kind for proposing
Random mode optimization method, it has caused extensive concern since the 1960s proposes, and has been successfully applied
In the application in numerous scientific and engineering fields.Jakimovski proposes a kind of utilization genetic algorithm optimization grid work flow
Method, however the characteristics of for that may include multiple stream topological structures in dynamic workflow, which can not directly be carried out
Optimization.
Summary of the invention
The purpose of the present invention is to solve drawbacks described above in the prior art, provide a kind of dynamic based on genetic algorithm
State workflow schedule method.
The purpose of the present invention can be reached by adopting the following technical scheme that:
A kind of dynamic workflow scheduling method based on genetic algorithm, the dispatching method include:
S1, the stream topologies being likely to occur all to dynamic workflow construct, by establish a series of subgraphs come
Each stream topologies that may occur is corresponded to, and carries out Holistic modeling with probabilistic model, process is as follows:
S101, all topological structure sum n that may be present in workflow are determined；
S102, workflow directed acyclic graph is rewritten into subgraph set { Φ1,Φ2,…,Φn, wherein ΦiIt is dynamic to represent this
A kind of control stream topological structure being likely to occur in state workflow；
S103, foundation { p1,p2,…,pnMaking by Probability Sets, wherein piIt represents dynamic workflow and takes ΦiControl stream topology knot
Structure executes the probability of work, piIt is executed in record information according to workflow in m history before this and takes ΦiControl stream topology knot
The times N of structureiAnd it is calculated, m >=500, it may be assumed that
And have:
S2, model is optimized, wherein optimization aim is to find one group of workflow schedule mode K, so that workflow
The desired value that the expense executed in a dynamic environment expends
It minimizes, wherein K.C (Φj) refer to that scheduling K flows topological structure Φ in controljUnder required expense.
Further, the process optimized in the step S2 to model is as follows:
S201, the crossing-over rate px of initialization algorithm, aberration rate pm and Population Size parameter popsize, and generate the first generation
Population, the coding mode of each individual is in population
K(k1,k2,…,kn)
Wherein, kiIt indicates the task T of workflowiIt is matched to corresponding cloud computing serviceIt executes；
The adaptive value of each individual, the mode of evaluation are as follows in S202, evaluation population:
Topological structure Φ is flowed according to each possible control of dynamic workflowj, calculate separately each solution K and open up at this
Flutter structure ΦjUnder required execution time K.T (Φj) and charge costs K.C (Φj), if for all topological structures
{Φ1,Φ2,…Φn, there is K.T (Φj)≤Deadline, i.e., the execution time of workflow can under all topological structures
Meet and complete time limit Deadline defined in user, then the adaptive value K.fitness for solving K is calculated by following formula:
If scheduling K is controlled in stream topological structure in one or more K.T (Φj) > Deadline cannot expire
The foot user-defined completion time limit, then the adaptive value of solution K is set as the upper limit MAX of adaptive value；
S203, by the way of tournament selection, selected from previous generation population popsize individual；
S204, crossover operation is carried out to group of new generation elected；
S205, to selection, intersect after obtained population further progress mutation operation of new generation
S206, to it is above intersect, each individual of the obtained population of new generation of mutation operation, according to step S202's
Mode carries out adaptive value evaluation；
The smallest optimum individual of adaptive value replaces Current generation kind in S207, the population of new generation obtained with aforesaid operations
The worst individual of group；
S208, terminate to optimize if current iteration number is beyond the maximum number of iterations of algorithm setting, otherwise return to step
Rapid S203 is continued to execute.
Further, the step S203 process is as follows:
Two individuals are arbitrarily selected from population, compare the size of the adaptive value K.fitness of the two individuals, and
Therefrom selection K.fitness lesser one enters among new population, repeats as procedure described above popsize times
Enter the individual of next-generation population to popsize.
Further, the step S204 process is as follows:
For each individual, determine the need for carrying out crossover operation to the individual according to crossover probability px；
For needing to carry out the individual of crossover operation, these individuals are matched two-by-two in a random fashion, establishing pair
Two individuals are respectivelyWithThen the two are individual as follows
Carry out crossover operation:
1. generating one 1 to the random integers p between n-1 as crossover location；
2. the two individuals are obtained two new individuals as single point crossing according to position pWith
3. replacing its parent individual for obtained new individual is intersected；
For not needing the individual intersected, stayed in population of new generation relaying continuation of insurance.
Further, the step S206 process is as follows:
For each individual, determine the need for carrying out mutation operation to the individual according to mutation probability pm；
If necessary to carry out mutation operation, then K (k is arbitrarily selected1,k2,…,kn) in a kran, wherein ran is 1 to arrive
Random integers between n, and by kranIt is set as gatheringIn any one；
If you do not need to carrying out mutation operation, then keeps the individual constant, continue with next individual.
Further, the described crossing-over rate px=0.7, the aberration rate pm=0.1, the Population Size parameter
Popsize=50.
The present invention has the following advantages and effects with respect to the prior art:
1, all control stream topological structures that dynamic duty flows down are considered, therefore it is lower dynamic adequately to solve cloud work
The characteristic of state time-varying.
2, in the process of genetic algorithm optimization performance indicator, optimal solution controls stream topological structure for each can
Row, improves the strong robustness of algorithm, and the present invention for being performance in industrial application is more preferable.
Detailed description of the invention
Fig. 1 is more control stream topological structure directed acyclic graphs modeling schematic diagram of dynamic workflow in the present invention；
Fig. 2 is the dynamic workflow scheduling method flow diagram in the present invention based on genetic algorithm.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
Member's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Embodiment
Workflow can be stated by a directed acyclic graph G=(V, A), the set V={ T of interior joint1,
T2,…,TnCalculating task in corresponding workflow, n is the number for the task that workflow is included, a directed edge (Ti,Tj)
Expression task TiAnd TjBetween priority constraint relationship, i.e. task TjIt can only be in his father's task TiIt could start to execute after the completion.
In the environment of cloud computing, each task can be realized by a variety of different cloud computing services, i.e. task TiCorresponding one
The relative cloud service of seriesWhereinIndicate that one kind can be used for realizing TiCloud computing service,
miIt is task TiThe sum of corresponding all available cloud services.The attribute of one cloud service can be with one group of binary group come table
ShowWherein,WithRespectively represent serviceThe execution time and expense.The mesh of workflow schedule
Mark is to find a kind of scheduling method K (k1,k2,…,kn), wherein kiExpression task TiByIt executes, makes these calculating tasks
It can be arranged into corresponding cloud computing service and execute under conditions of meeting the limitation of precedence constraint defined in directed acyclic graph,
So that the index of workflow can satisfy the demand of user, and make what is optimized required for user to be optimized.But in reality
In cloud computing application, due to cloud system and the time-varying dynamic characteristic applied is calculated, workflow often also has certain time-varying
Dynamic characteristic, therefore also to fully consider in Optimization Work stream scheduling problem the time-varying dynamic characteristic of workflow.
One, to the specific steps of dynamic workflow modeling:
(1) all topological structure sum n that may be present in workflow are determined；
(2) workflow directed acyclic graph is rewritten into subgraph set { Φ1,Φ2,…,Φn, wherein ΦiRepresent the dynamic
A kind of control stream topological structure being likely to occur in workflow；
(3) { p is established1,p2,…,pnMaking by Probability Sets, wherein piIt represents dynamic workflow and takes ΦiControl stream topological structure
Execute the probability of work, piIt is executed in record information according to workflow in 500 history before this and takes ΦiControl stream topology knot
The times N of structureiAnd it is calculated, it may be assumed that
And have:
Based on this modeling pattern, model of the present invention can fully consider to move possessed by workflow
State time-varying characteristics.It is worth noting that, in a model, and if only if scheduling solution to all control stream topology knots being likely to occur
Structure Φi∈{Φ1,Φ2,…ΦnRequired for execute the time can satisfy the customized execution time restriction of user
Deadline, which is feasible.The optimization aim of the model is to find one group of workflow schedule mode K, is made
Obtain the desired value that the expense that workflow executes in a dynamic environment expends
It minimizes, wherein K.C (Φj) refer to that scheduling K flows topological structure Φ in controljUnder required expense.
Two, the specific execution step of algorithm
Algorithm mainly includes 8 following steps:
(1) the crossing-over rate px=0.7 of initialization algorithm, aberration rate pm=0.1 and Population Size parameter popsize=50,
And first generation population is generated, the coding mode of each individual is in population
K(k1,k2,…,kn)
Wherein, kiIt indicates the task T of workflowiIt is matched to corresponding cloud computing serviceIt executes；
(2) adaptive value of each individual in population is evaluated, the mode of evaluation is: can according to each of dynamic workflow
The control stream topological structure Φ of energyj, each solution K is calculated separately in topological structure ΦjUnder required execution time K.T (Φj)
With charge costs K.C (Φj).If for all topological structure { Φ1,Φ2,…Φn, there is K.T (Φj)≤
Deadline, i.e., the execution time of workflow, which can meet, under all topological structures completes the time limit defined in user
Deadline, then the adaptive value K.fitness for solving K are calculated by following formula:
If scheduling K is controlled in stream topological structure in one or more K.T (Φj) > Deadline cannot expire
The foot user-defined completion time limit, then the adaptive value of solution K is set as the upper limit MAX of adaptive value.The adaptive value for solving K is lower, table
The quality of the bright solution is more excellent；
(3) by the way of tournament selection, popsize individual is selected from previous generation population.I.e. algorithm is first from kind
Two individuals are arbitrarily selected in group, compare the size of the adaptive value K.fitness of the two individuals, and are therefrom selected
K.fitness lesser one enters among new population.Popsize times, which is repeated, according to above-mentioned steps obtains popsize
A individual for entering next-generation population；
(4) crossover operation is carried out to group of new generation elected.It is general according to intersecting firstly for each individual
Rate px determines the need for carrying out crossover operation to the individual.For needing to carry out the individual of crossover operation, algorithm is according to random
Mode by these individual match two-by-two.Two individuals of establishing pair are respectivelyWithThen the two individual carry out crossover operation as follows: 1. generate one 1 to n-1 it
Between random integers p as crossover location；2. the two individuals are obtained two new individuals as single point crossing according to position pWith3. by intersecting
The new individual arrived replaces its parent individual.For not needing the individual intersected, continuation of insurance will be relayed in population of new generation and stayed；
(5) to the population further progress mutation operation of new generation obtained after selection, intersection.Firstly for it is each each and every one
Body determines the need for carrying out mutation operation to the individual according to mutation probability pm.If necessary to carry out mutation operation, then appoint
Meaning selection K (k1,k2,…,kn) in a kran, wherein ran is 1 to the random integers between n, and by kranIt is set as gatheringIn any one.
(6) to each individual by the obtained population of new generation in above-mentioned steps (4)-(5), in the way of step (2)
Carry out adaptive value evaluation；
(7) the worst individual for the optimum individual replacement Current generation population found with current algorithm；
(8) terminate optimization algorithm if current iteration number is beyond the maximum number of iterations of algorithm setting, otherwise return to
Step (3)；
The flow chart of entire algorithm is as shown in Figure 2.
Algorithm in relation to dynamic workflow scheduling side is very rare, in existing document, the propositions such as only Yu
Deadline-MDP algorithm can be used in the solution of similar problems.Therefore, by the method for invention and Deadline-MDP algorithm into
Row compares.The present embodiment tests both methods with 10 examples.Wherein, preceding 3 examples are the dynamics in reality
Workflow issues, including e-Economic application problem, Neuscience application problem and e-protein dynamic duty flow problem.
Remaining example is then generated according to the library PSPLIB.Since Deadline-MDP is a kind of deterministic algorithm, it is limited by this, therefore
Its primary calculates can only provide a solution of problem.In order to by comparison result according to adequacy, by the method independent operating of invention
Obtain 100 results 100 times.In 10 all problems, what the dynamic workflow scheduling method based on genetic algorithm obtained
Average value is superior to the result that Deadline-MDP is obtained.In addition, (having 30 or more tasks in or in extensive problem
), even the worst result that the present invention obtains also is better than the result of Deadline-MDP.Generally speaking, institute of the present invention
The method of proposition can reduce the expense of dynamic workflow 10-20%, this demonstrate that the method for invention is effective.
In conclusion the invention proposes workflows in changing environment when a kind of dynamic based on genetic algorithm optimization to be averaged
The method of period expense.This method is statisticallyd analyze based on historical operational information, and is executed by circulation to control cloud work and flow down
Topological structure and its probability of happening.Work under changing environment when by this method obtaining that dynamic can be effectively improved to optimal solution
Make the efficiency of stream scheduling.
The above embodiment is a preferred embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment
Limitation, it is other it is any without departing from the spirit and principles of the present invention made by change, modification, substitution, combination, letter
Change, should be equivalent substitute mode, be included within the scope of the present invention.
Claims (6)
1. a kind of dynamic workflow scheduling method based on genetic algorithm, which is characterized in that the dispatching method includes:
S1, the stream topologies being likely to occur all to dynamic workflow construct, and are corresponded to by establishing a series of subgraphs
Each stream topologies that may occur, and Holistic modeling is carried out with probabilistic model, process is as follows:
S101, all topological structure sum n that may be present in workflow are determined；
S102, workflow directed acyclic graph is rewritten into subgraph set { Φ1,Φ2,…,Φn, wherein ΦiRepresent the dynamic duty
A kind of control stream topological structure being likely to occur in stream；
S103, foundation { p1,p2,…,pnMaking by Probability Sets, wherein piIt represents dynamic workflow and takes ΦiControl stream topological structure is held
The probability of row work, piIt is executed in record information according to workflow in m history before this and takes ΦiControl stream topological structure
Times NiAnd it is calculated, m >=500, it may be assumed that
And have:
S2, model is optimized, wherein optimization aim is to find one group of workflow schedule mode K, so that workflow is in dynamic
The desired value that the expense executed under environment expends
It minimizes, wherein K.C (Φj) refer to that scheduling K flows topological structure Φ in controljUnder required expense.
2. the dynamic workflow scheduling method according to claim 1 based on genetic algorithm, which is characterized in that the step
The process optimized in rapid S2 to model is as follows:
S201, the crossing-over rate px of initialization algorithm, aberration rate pm and Population Size parameter popsize, and first generation population is generated,
The coding mode of each individual is in population
K(k1,k2,…,kn)
Wherein, kiIt indicates the task T of workflowiIt is matched to corresponding cloud computing serviceIt executes；
The adaptive value of each individual, the mode of evaluation are as follows in S202, evaluation population:
Topological structure Φ is flowed according to each possible control of dynamic workflowj, each solution K is calculated separately in the topological structure
ΦjUnder required execution time K.T (Φj) and charge costs K.C (Φj), if for all topological structure { Φ1,Φ2,…
Φn, there is K.T (Φj)≤Deadline, i.e., the execution time of workflow can meet user and determine under all topological structures
The completion time limit Deadline of justice, then the adaptive value K.fitness for solving K are calculated by following formula:
If scheduling K is controlled in stream topological structure in one or more K.T (Φj) > Deadline, i.e., be not able to satisfy user
The completion time limit of definition, then the adaptive value of solution K is set as the upper limit MAX of adaptive value；
S203, by the way of tournament selection, selected from previous generation population popsize individual；
S204, crossover operation is carried out to group of new generation elected；
S205, to selection, intersect after obtained population further progress mutation operation of new generation
S206, to it is above intersect, each individual of the obtained population of new generation of mutation operation, in the way of step S202 into
The evaluation of row adaptive value；
The smallest optimum individual of adaptive value replaces Current generation population most in S207, the population of new generation obtained with aforesaid operations
Poor individual；
S208, terminate to optimize if current iteration number is beyond the maximum number of iterations of algorithm setting, otherwise return step
S203 is continued to execute.
3. the dynamic workflow scheduling method according to claim 2 based on genetic algorithm, which is characterized in that the step
Rapid S203 process is as follows:
Two individuals are arbitrarily selected from population, compare the size of the adaptive value K.fitness of the two individuals, and are therefrom selected
It selects K.fitness lesser one to enter among new population, repeats popsize times and obtain as procedure described above
Popsize enters the individual of next-generation population.
4. the dynamic workflow scheduling method according to claim 2 based on genetic algorithm, which is characterized in that the step
Rapid S204 process is as follows:
For each individual, determine the need for carrying out crossover operation to the individual according to crossover probability px；
For needing to carry out the individual of crossover operation, these individuals are matched two-by-two in a random fashion, two of establishing pair
Individual is respectivelyWithThen the two individuals carry out as follows
Crossover operation:
1. generating one 1 to the random integers p between n-1 as crossover location；
2. the two individuals are obtained two new individuals as single point crossing according to position pWith
3. replacing its parent individual for obtained new individual is intersected；
For not needing the individual intersected, stayed in population of new generation relaying continuation of insurance.
5. the dynamic workflow scheduling method according to claim 2 based on genetic algorithm, which is characterized in that the step
Rapid S206 process is as follows:
For each individual, determine the need for carrying out mutation operation to the individual according to mutation probability pm；
If necessary to carry out mutation operation, then K (k is arbitrarily selected1,k2,…,kn) in a kran, wherein ran is 1 between n
Random integers, and by kranIt is set as gatheringIn any one；
If you do not need to carrying out mutation operation, then keeps the individual constant, continue with next individual.
6. the dynamic workflow scheduling method according to claim 2 based on genetic algorithm, which is characterized in that the friendship
Fork rate px=0.7, the aberration rate pm=0.1, the Population Size parameter popsize=50.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811194823.7A CN109492872A (en) | 2018-10-15 | 2018-10-15 | Dynamic workflow scheduling method based on genetic algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811194823.7A CN109492872A (en) | 2018-10-15 | 2018-10-15 | Dynamic workflow scheduling method based on genetic algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109492872A true CN109492872A (en) | 2019-03-19 |
Family
ID=65690596
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811194823.7A Pending CN109492872A (en) | 2018-10-15 | 2018-10-15 | Dynamic workflow scheduling method based on genetic algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109492872A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113158568A (en) * | 2021-04-23 | 2021-07-23 | 电子科技大学 | Near-field sparse array design method |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226759A (en) * | 2013-04-25 | 2013-07-31 | 中山大学 | Dynamic cloud workflow scheduling method based on genetic algorithm |
-
2018
- 2018-10-15 CN CN201811194823.7A patent/CN109492872A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226759A (en) * | 2013-04-25 | 2013-07-31 | 中山大学 | Dynamic cloud workflow scheduling method based on genetic algorithm |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113158568A (en) * | 2021-04-23 | 2021-07-23 | 电子科技大学 | Near-field sparse array design method |
| CN113158568B (en) * | 2021-04-23 | 2022-12-02 | 电子科技大学 | Near-field sparse array design method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Stützle et al. | A concise overview of applications of ant colony optimization | |
| Rosenberg et al. | Metaheuristic optimization of large-scale qos-aware service compositions | |
| CN107977740A (en) | A kind of scene O＆M intelligent dispatching method | |
| Qu et al. | Particle swarm optimization for the Steiner tree in graph and delay-constrained multicast routing problems | |
| Lalbakhsh et al. | An improved model of ant colony optimization using a novel pheromone update strategy | |
| Habibi et al. | A multi-objective optimization model for project scheduling with time-varying resource requirements and capacities | |
| CN103226759A (en) | Dynamic cloud workflow scheduling method based on genetic algorithm | |
| CN107656799B (en) | Workflow scheduling method considering communication and computing cost under multi-cloud environment | |
| CN108664330A (en) | A kind of cloud resource distribution method based on change neighborhood search strategy | |
| CN103279818A (en) | Method for cloud workflow scheduling based on heuristic genetic algorithm | |
| Pooranian et al. | Hybrid metaheuristic algorithm for job scheduling on computational grids | |
| Antevski et al. | Resource orchestration of 5G transport networks for vertical industries | |
| Zhou et al. | A game-theory approach for job scheduling in networked manufacturing | |
| Li et al. | Task scheduling of computational grid based on particle swarm algorithm | |
| CN109492872A (en) | Dynamic workflow scheduling method based on genetic algorithm | |
| Mortazavi et al. | A discrete cuckoo search algorithm for reliability-aware energy-efficient iot applications multi-service deployment in fog environment | |
| Ren et al. | Learning automata-based data aggregation tree construction framework for cyber-physical systems | |
| Lopes et al. | Algorithm based on particle swarm applied to electrical load scheduling in an industrial setting | |
| CN110362378A (en) | A kind of method for scheduling task and equipment | |
| Jian et al. | Cloud based evolution algorithm with feedback control for emergency logistic | |
| Ebrahim et al. | Privacy-aware load balancing in fog networks: A reinforcement learning approach | |
| Glišović | Comparison of a fuzzy genetic and simulated annealing algorithm approach for project time-cost tradeoff | |
| Horng et al. | Integrating ant colony system and ordinal optimization for solving stochastic job shop scheduling problem | |
| Zambelli et al. | A predictive control approach for long term hydrothermal scheduling | |
| Cao et al. | A service process optimization method based on model refinement |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190319 |
|
| WD01 | Invention patent application deemed withdrawn after publication |