CN104915557B - A kind of cloud method for allocating tasks based on Bi-objective ant group algorithm - Google Patents
A kind of cloud method for allocating tasks based on Bi-objective ant group algorithm Download PDFInfo
- Publication number
- CN104915557B CN104915557B CN201510305822.5A CN201510305822A CN104915557B CN 104915557 B CN104915557 B CN 104915557B CN 201510305822 A CN201510305822 A CN 201510305822A CN 104915557 B CN104915557 B CN 104915557B
- Authority
- CN
- China
- Prior art keywords
- mrow
- virtual machine
- task
- ant
- cloud
- 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.)
- Expired - Fee Related
Links
Abstract
The present invention relates to a kind of cloud method for allocating tasks based on Bi-objective ant group algorithm, the cloud Task Allocation Problem of two-objective programming optimization is considered for solving, this method is that the visibility in ant group algorithm chooses the calculation relevant with corresponding object function by different probability, the cloud Task Allocation Problem for considering biobjective scheduling is solved to reach, compared with prior art, solution procedure is simple and with preferable deadline and load balancing.
Description
Technical field
The present invention relates to field of cloud calculation, distributed more particularly, to a kind of cloud task based on Bi-objective ant group algorithm
Method.
Background technology
Ant group algorithm is applied to the solution of cloud Task Allocation Problem, mainly there is two kinds.One kind is asked for single object optimization
Topic, another is to be directed to multi-objective optimization question.
In application ant colony optimization for solving cloud Task Allocation Problem, people are usually calculated using for single goal based on ant colony
The cloud method for allocating tasks of method, this method the visibility meter of ant group algorithm count in consider single-objective problem optimization, but this
The method of sample can not take into account multiple object functions.
And for the cloud method for allocating tasks based on ant group algorithm of multi-objective optimization question, use and calculated by ant colony
The method that the method that method is combined with other theories, such as ant group algorithm are combined with Pareto Optimum Theory, although such
Method can solve the cloud Task Allocation Problem for multi-objective optimization question, have and preferably solve performance, but be to solve for
Journey is comparatively more complicated.
The content of the invention
To solve above the deficiencies in the prior art, the invention provides a kind of cloud task point based on Bi-objective ant group algorithm
Method of completing the square, the solution procedure of this method is simple and with preferable deadline and load balancing.
To realize above goal of the invention, the technical scheme of use is:
A kind of cloud method for allocating tasks based on Bi-objective ant group algorithm, the cloud of two-objective programming optimization is considered for solving
Task Allocation Problem, the two-objective programming includes task completion time function f1 and system load balancing function f2, described double
Target ant group algorithm comprises the following steps:
S1. the quantity of cloud task is set as n, and the Imaginary Mechanism that ant can select is into set allowedk, the quantity of ant
For antNum, the numbering of ant represents that iterations is represented with t with antid, and total iterations is times, iterations t
< times;
S2. iterations t=0, initialization ant colony and pheromones are initialized, the pheromones of initialization are represented by following formula:
Wherein τij(0) pheromones of initialization are represented, i represents the sequence number of task, and j represents the numbering of virtual machine,Table
Show task i to the virtual machine j initial information element relevant with task completion time function f1,Expression task i is to virtual machine
The j initial information element relevant with system load balancing function f2, w1+w2=1;
S3., numbering antid=0 ant is defined as to first ant in the t times iteration, then determined by calculating
The virtual machine j of cloud task i distribution, cloud task is assigned on virtual machine j, then virtual machine j is excluded into allowedk, and according to
The codomain that system is produced is updated for the random number r of (0,1) to visibility, specific as follows:
Wherein ηij(t) it is the visibility after updating,Task i to virtual machine j and task are complete during for antid=0
Into the relevant visibility of function of time f1,Task i is to virtual machine j and system load balancing function during for antid=0
Visibility relevant f2,Task i to the virtual machine j energy relevant with task completion time function f1 during for antid > 0
Degree of opinion,Task i to the virtual machine j visibility relevant with system load balancing function f2 during for antid > 0, q " is
First setting value of system;
S4. after all cloud tasks are assigned by order numbering antid=0 ant, point that the ant draws is calculated
Measurement S (t) with scheme, then carries out pheromones local updating, wherein calculating the allocative decision measurement S (t) that the ant draws
Process it is specific as follows:
S1(t) measurement of the allocative decision relevant with task completion time function f1, S are represented2(t) represent and system load
The measurement of allocative decision relevant balance function f2;
S5. make antid=antid+1 and step S3~S4 is repeated by the corresponding ant of the numbering, then repeat
S5 is until antid >=antNum;
S6. the measurement for the allocative decision that each ant draws in the t times iteration, Cong Zhongxuan are resulted in by step S5
Select the measurement of optimal allocative decision and be named as Sg1(t) global information element renewal, is then carried out;
S7. t=t+1 and repeat step S3~S6 operation are made, step S7 is repeated until t >=times;
S8. the measurement for the optimal distributing scheme that each iteration is obtained is resulted in by step S7, is obtained according to each iteration
The measurement of the optimal distributing scheme obtained is selected optimal allocative decision and exported;
The optimum distributing scheme that system is exported according to Bi-objective ant group algorithm is allocated to cloud task.
Preferably, in step S3, ant determines that the virtual machine j of cloud task i distribution process is specific as follows:
Wherein q is the random number that the codomain that system is produced is (0,1), q0For the second setting value of system.
Preferably, determine that the process of the virtual machine of distribution is specific as follows by roulette algorithm:
First calculate the select probability that kth ant distributes cloud task i to virtual machine j:
α is pheromones heuristic factor, and β is visibility heuristic factor;Then it is the random of (0,1) that system, which produces a codomain,
Number q ', does cumulative probability statistics, which cumulative probability random number q ' falls in by cloud task i to each virtual machine select probability
In, the corresponding virtual machine of the cumulative probability is the virtual machine for determining distribution.
Preferably, in step S3, when cloud task is assigned on virtual machine, if allowedkIt is for empty set, then again initial
Change allowedk, then cloud task is assigned to allowedkIn virtual machine on.
Preferably, in step S4, the detailed process for carrying out pheromones local updating is as follows:
τij(t+1) it is the pheromones updated, ρ is pheromones volatility coefficient,
Preferably, in step S6, the detailed process for carrying out global information element renewal is as follows:
τij(t+1) it is the global pheromones updated,
Compared with prior art, the beneficial effects of the invention are as follows:
The method that the present invention is provided is that the visibility selection in ant group algorithm has with corresponding object function by different probability
The calculation of pass, the cloud Task Allocation Problem for considering biobjective scheduling is solved to reach, the solution procedure of this method it is simple and
With preferable deadline and load balancing.
Brief description of the drawings
Fig. 1 is the flow chart of cloud method for allocating tasks.
Fig. 2 is the flow chart for updating visibility.
Fig. 3 is the flow chart for the measurement for calculating the allocative decision that the ant draws.
Fig. 4 is the deadline contrast schematic diagram of three algorithms.
Fig. 5 is the unbalanced degree standard deviation contrast schematic diagram of three algorithms.
Embodiment
Accompanying drawing being given for example only property explanation, it is impossible to be interpreted as the limitation to this patent;
Below in conjunction with drawings and examples, the present invention is further elaborated.
Embodiment 1
The method that the present invention is provided is carried out before careful elaboration, the present embodiment is retouched to cloud task and virtual machine first
The mode of stating is illustrated, and cloud task is described as follows:
Task={ TaskId, Length, ExCpuMips, ExMem, ExBw }
Wherein TaskId is the numbering of cloud task;Length is the length of cloud task;ExCpuMips is that cloud task is expected
Cpu instruction execution speed demands;ExMem is the memory size demand that cloud task is expected;ExBw is that the band that cloud task is expected is roomy
Small demand.
Virtual machine is described as follows:
Vm={ VmId, CpuNum, CpuMips, Mem, Bw }
Wherein VmId is the numbering of virtual machine;CpuNum is the CPU numbers of virtual machine;CpuMips is that the Cpu of virtual machine refers to
Order performs speed;Mem is the memory size of virtual machine;Bw is the amount of bandwidth of virtual machine.
And when calculating visibility and pheromones, it is necessary first to cloud task and virtual machine are normalized, it is normalized
Process is specific as follows:
Cloud task i length is normalized by following mode:
During wherein TLeArr [i] represents that the task length of the cloud task i after normalization, maxLe are all cloud tasks
Maximum cloud task length, minLe is minimum cloud task length in all cloud tasks.Cloud task i ExCpuMips,
ExMem, ExBw normalization are similarly.Therefore cloud task i 4 normalization arrays are obtained:TLeArr, TCpArr, TMeArr and
TBwArr。
Virtual machine j computing capability normalization is as follows:
Vm [j] .Capacity=vm [j] .CpuNum × vm [j] .CpuMips
Virtual machine j Mem, Bw normalization are identical with the normalization mode of cloud task.Therefore 3 of virtual machine j are obtained
Normalize array:VCaArr, VMeArr and VBwArr.
On the basis of more than, then detailed narration is carried out to technical scheme, the method that the present invention is provided is used for
Solve consider two-objective programming optimization cloud Task Allocation Problem, wherein two-objective programming include task completion time function f1 and
System load balancing function f2, as shown in figure 1, Bi-objective ant group algorithm comprises the following steps:
S1. the quantity of cloud task is set as n, and the Imaginary Mechanism that ant can select is into set allowedk, the quantity of ant
For antNum, the numbering of ant is represented with antid, and iterations is represented with t (t is since 0), and total iterations is times,
Iterations t < times;
S2. iterations t=0, initialization ant colony and pheromones are initialized, the pheromones of initialization are represented by following formula:
Wherein τij(0) pheromones of initialization are represented, i represents the sequence number of task, and j represents the numbering of virtual machine,Table
Show task i to the virtual machine j initial information element relevant with task completion time function f1,Expression task i is to virtual machine
The j initial information element relevant with system load balancing function f2, w1+w2=1;
S3., numbering antid=0 ant is defined as to first ant in the t times iteration, then determined by calculating
The virtual machine j of cloud task i distribution, cloud task is assigned on virtual machine j, then virtual machine j is excluded into allowedk, and according to
The codomain that system is produced is updated for the random number r of (0,1) to visibility, specific as follows:
Wherein ηij(t) it is the visibility after updating,Task i to virtual machine j and task are complete during for antid=0
Into the relevant visibility of function of time f1,Task i is to virtual machine j and system load balancing function during for antid=0
Visibility relevant f2,Task i to the virtual machine j energy relevant with task completion time function f1 during for antid > 0
Degree of opinion,Task i to the virtual machine j visibility relevant with system load balancing function f2 during for antid > 0, q " is
First setting value of system;
In step S3, ant determines that the virtual machine j of cloud task i distribution process is specific as follows:
Wherein q is the random number that the codomain that system is produced is (0,1), q0For the second setting value of system.
Determine that the process of the virtual machine of distribution is specific as follows by roulette algorithm simultaneously:
First calculate the select probability that kth ant distributes cloud task i to virtual machine j:
α is pheromones heuristic factor, and β is visibility heuristic factor;Then it is the random of (0,1) that system, which produces a codomain,
Number q ', does cumulative probability statistics, which cumulative probability random number q ' falls in by cloud task i to each virtual machine select probability
In, the corresponding virtual machine of the cumulative probability is the virtual machine for determining distribution.
In such scheme, when cloud task is assigned on virtual machine, if allowedkFor empty set, then reinitialize
allowedk, then cloud task is assigned to allowedkIn virtual machine on.
S4. after all cloud tasks are assigned by order numbering antid=0 ant, calculated by method shown in Fig. 3
The measurement S (t) for the allocative decision that the ant draws, then carries out pheromones local updating, carries out the tool of pheromones local updating
Body process is as follows:
τij(t+1) it is the pheromones updated, ρ is pheromones volatility coefficient,
And the process for calculating the allocative decision measurement S (t) that the ant draws is specific as follows:
S1(t) measurement of the allocative decision relevant with task completion time function f1, S are represented2(t) represent and system
The measurement of allocative decision relevant load balancing function f2;
S5. make antid=antid+1 and step S3~S4 is repeated by the corresponding ant of the numbering, then repeat
S5 is until antid >=antNum;
S6. the measurement for the allocative decision that each ant draws in the t times iteration, Cong Zhongxuan are resulted in by step S5
Select the measurement of optimal allocative decision and be named as Sg1(t) global information element renewal, is then carried out, global information element renewal is carried out
Detailed process it is as follows:
τij(t+1) it is the global pheromones updated,
S7. t=t+1 and repeat step S3~S6 operation are made, step S7 is repeated until t >=times;
S8. the measurement for the optimal distributing scheme that each iteration is obtained is resulted in by step S7, is obtained according to each iteration
The measurement of the optimal distributing scheme obtained is selected optimal allocative decision and exported;
The optimum distributing scheme that system is exported according to Bi-objective ant group algorithm is allocated to cloud task.
In such scheme,It can be carried out by below equation
Calculate:
Otherwise
Wherein curGjRepresent the cloud set of tasks being currently assigned on virtual machine j.
Meanwhile, S1And S (t)2(t) be calculated as follows:
Wherein m is the quantity of virtual machine, unLoadjIt is virtual machine j unbalanced degree,It is averagely unbalanced degree.
While the superiority in order to which the method that the present invention is provided is better described, the present embodiment has also carried out specific emulation
Experiment, on cloud computing emulation platform CloudSim, what the present invention was provided " considers task completion time and system load balancing
The cloud task allocation algorithms of two object functions " and " the cloud task distribution for only considering task completion time based on ant group algorithm
Algorithm " and " the cloud task allocation algorithms for only considering system load balancing based on ant group algorithm " have carried out the contrast of performance.
The iterations times of wherein ant colony is set to 100, and ant quantity antNum is set to 100, pheromones heuristic factor α
1.0 are set to, visibility heuristic factor β is set to 5.0, and pheromones volatility coefficient ρ is set to 0.4, and kth ant distribution cloud task i is arrived
Q in virtual machine j mode0It is set to the coefficient w in 0.1, two-objective programming1It is set to 0.1, coefficient w2It is set to 0.9.
The quantity for setting virtual machine is 10, and the quantity of cloud task is incremented by every time by 10, and 10 until 100, have 10 kinds
Situation, 10 experiments is done to each case, then average.
Virtual machine parameter list and cloud task template parameter list are as shown in table 1, table 2:
The virtual machine parameter list of table 1
VmId | CpuNumber | Mips | Memory | Bandwidth |
0 | 2 | 2500 | 1024 | 2000 |
1 | 2 | 1000 | 1024 | 2000 |
2 | 1 | 1200 | 2048 | 2500 |
3 | 2 | 2000 | 512 | 1500 |
4 | 3 | 3000 | 2048 | 1500 |
5 | 3 | 2500 | 1024 | 2500 |
6 | 2 | 2500 | 1024 | 1000 |
7 | 4 | 1500 | 512 | 2500 |
8 | 2 | 1000 | 1024 | 2500 |
9 | 1 | 3000 | 1024 | 3000 |
The cloud task template parameter list of table 2
CloudletId | Length | ExCpuMips | ExMemory | ExBandwidth |
0 | 40000 | 150 | 128 | 250 |
1 | 30000 | 200 | 128 | 200 |
2 | 20000 | 100 | 64 | 200 |
3 | 50000 | 200 | 256 | 150 |
4 | 20000 | 250 | 64 | 300 |
5 | 30000 | 150 | 256 | 200 |
6 | 10000 | 250 | 128 | 150 |
7 | 25000 | 300 | 128 | 250 |
8 | 20000 | 250 | 256 | 200 |
9 | 15000 | 100 | 128 | 200 |
More than set experiment condition under, the deadline of three algorithms drawn to such as shown in table 3, Fig. 4.
The deadline table of 3 three algorithms of table
From chart above, with the increase of number of tasks, the deadline of three algorithms gradually increases.During based on completing
Between (only consider deadline) cloud task allocation algorithms deadline preferably, (considered based on deadline and load balancing
Deadline and load balancing) cloud task allocation algorithms (using the present invention method) although sometimes than based on load balancing
The deadline of the cloud task allocation algorithms of (only considering load balancing) is big, but generally speaking, deadline ratio is based on load
Cloud task allocation algorithms are small in a balanced way, and the deadline of the cloud task allocation algorithms based on load balancing is worst.Therefore, based on complete
Cloud task allocation algorithms (using the method for the present invention) into time and load balancing have the preferable deadline.
And the unbalanced degree standard deviation of three algorithms is contrasted then as shown in table 4 and fig. 5.
The unbalanced degree standard deviation table of 4 three algorithms of table
From chart above, with the increase of number of tasks, the unbalanced degree standard deviation of three algorithms becomes larger.It is based on
The unbalanced degree standard deviation of the cloud task allocation algorithms of deadline is worst, the cloud task point based on deadline and load balancing
Unbalanced degree standard deviation with algorithm (using the method for the present invention) is taken second place, and the cloud task allocation algorithms based on load balancing
Unbalanced degree standard deviation is best.Therefore, the cloud task allocation algorithms based on deadline and load balancing are (using the side of the present invention
Method) there is preferable unbalanced degree standard deviation.
In summary, what the present invention was provided " considers the cloud of two object functions of task completion time and system load balancing
Task allocation algorithms " have preferable deadline and preferable load balancing.
Obviously, the above embodiment of the present invention is only intended to clearly illustrate example of the present invention, and is not pair
The restriction of embodiments of the present invention.For those of ordinary skill in the field, may be used also on the basis of the above description
To make other changes in different forms.There is no necessity and possibility to exhaust all the enbodiments.It is all this
Any modifications, equivalent substitutions and improvements made within the spirit and principle of invention etc., should be included in the claims in the present invention
Protection domain within.
Claims (6)
1. a kind of cloud method for allocating tasks based on Bi-objective ant group algorithm, considers that the cloud of two-objective programming optimization is appointed for solving
Business assignment problem, the two-objective programming includes task completion time function f1 and system load balancing function f2, and its feature exists
In:The Bi-objective ant group algorithm comprises the following steps:
S1. the quantity of cloud task is set as n, and the Imaginary Mechanism that ant can select is into set allowedk, the quantity of ant is
AntNum, the numbering of ant represents that iterations is represented with t with antid, and total iterations is times, iterations t<
times;
S2. iterations t=0, initialization ant colony and pheromones are initialized, the pheromones of initialization are represented by following formula:
<mrow>
<msub>
<mi>&tau;</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mn>0</mn>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msub>
<mi>w</mi>
<mn>1</mn>
</msub>
<msubsup>
<mi>&tau;</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
<mn>1</mn>
</msubsup>
<mrow>
<mo>(</mo>
<mn>0</mn>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>w</mi>
<mn>2</mn>
</msub>
<msubsup>
<mi>&tau;</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
<mn>2</mn>
</msubsup>
<mrow>
<mo>(</mo>
<mn>0</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein τij(0) pheromones of initialization are represented, i represents the sequence number of task, and j represents the numbering of virtual machine,Represent to appoint
Business i to the virtual machine j initial information element relevant with task completion time function f1,Expression task i is to virtual machine j's
The initial information element relevant with system load balancing function f2, w1+w2=1;
S3., numbering antid=0 ant is defined as to first ant in the t times iteration, then determines that cloud is appointed by calculating
The virtual machine j of business i distribution, cloud task is assigned on virtual machine j, then virtual machine j is excluded into allowedk, and according to system
The codomain of generation is updated for the random number r of (0,1) to visibility, specific as follows:
<mrow>
<msubsup>
<mi>&eta;</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
<mo>&prime;</mo>
</msubsup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msub>
<mi>w</mi>
<mn>1</mn>
</msub>
<msubsup>
<mi>&eta;</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
<mrow>
<mo>&prime;</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>w</mi>
<mn>2</mn>
</msub>
<msubsup>
<mi>&eta;</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
<mrow>
<mo>&prime;</mo>
<mn>2</mn>
</mrow>
</msubsup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msubsup>
<mi>&eta;</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
<mrow>
<mo>&prime;</mo>
<mo>&prime;</mo>
</mrow>
</msubsup>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>&eta;</mi>
<mover>
<mi>&eta;</mi>
<mo>&CenterDot;&CenterDot;</mo>
</mover>
<mrow>
<mo>&prime;</mo>
<mo>&prime;</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mi>r</mi>
<mo>&le;</mo>
<msup>
<mi>q</mi>
<mrow>
<mo>&prime;</mo>
<mo>&prime;</mo>
</mrow>
</msup>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>&eta;</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
<mrow>
<mo>&prime;</mo>
<mo>&prime;</mo>
<mn>2</mn>
</mrow>
</msubsup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mi>r</mi>
<mo>></mo>
<msup>
<mi>q</mi>
<mrow>
<mo>&prime;</mo>
<mo>&prime;</mo>
</mrow>
</msup>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein ηij(t) it is the visibility after updating,During for antid=0 task i to virtual machine j's and task completion time
Visibility relevant function f1,Task i to virtual machine j is relevant with system load balancing function f2 during for antid=0
Visibility,For antid>Task i to the virtual machine j visibility relevant with task completion time function f1 when 0,For antid>Task i to the virtual machine j visibility relevant with system load balancing function f2 when 0, q " is the of system
One setting value;
S4. after all cloud tasks are assigned by order numbering antid=0 ant, the distribution side that the ant draws is calculated
The measurement S (t) of case, then carries out pheromones local updating, wherein calculating the mistake for the allocative decision measurement S (t) that the ant draws
Journey is specific as follows:
<mrow>
<msup>
<mi>S</mi>
<mo>&prime;</mo>
</msup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>w</mi>
<mn>1</mn>
</msub>
<msup>
<mi>S</mi>
<mn>1</mn>
</msup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mi>r</mi>
<mo>&le;</mo>
<msup>
<mi>q</mi>
<mrow>
<mo>&prime;</mo>
<mo>&prime;</mo>
</mrow>
</msup>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>w</mi>
<mn>2</mn>
</msub>
<msup>
<mi>S</mi>
<mn>2</mn>
</msup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mi>r</mi>
<mo>></mo>
<msup>
<mi>q</mi>
<mrow>
<mo>&prime;</mo>
<mo>&prime;</mo>
</mrow>
</msup>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
S1(t) measurement of the allocative decision relevant with task completion time function f1, S are represented2(t) represent and system load balancing
The measurement of allocative decision relevant function f2;
S5. make antid=antid+1 and step S3~S4 is repeated by the corresponding ant of the numbering, then repeatedly S5 is straight
To antid >=antNum;
S6. the measurement for the allocative decision that each ant draws in the t times iteration is resulted in by step S5, is therefrom selected most
The measurement of excellent allocative decision is simultaneously named as Sgl(t) global information element renewal, is then carried out;
S7. t=t+1 and repeat step S3~S6 operation are made, step S7 is repeated until t >=times;
S8. the measurement for the optimal distributing scheme that each iteration is obtained is resulted in by step S7, is obtained according to each iteration
The measurement of optimal distributing scheme is selected optimal allocative decision and exported;
The optimum distributing scheme that system is exported according to Bi-objective ant group algorithm is allocated to cloud task.
2. the cloud method for allocating tasks according to claim 1 based on Bi-objective ant group algorithm, it is characterised in that:Step S3
In, ant determines that the virtual machine j of cloud task i distribution process is specific as follows:
Wherein q is the random number that the codomain that system is produced is (0,1), q0For the second setting value of system, α be pheromones inspire because
Son, β is visibility heuristic factor.
3. the cloud method for allocating tasks according to claim 2 based on Bi-objective ant group algorithm, it is characterised in that:Pass through wheel
Disk gambling algorithm determines that the process of the virtual machine of distribution is specific as follows:
First calculate the select probability that kth ant distributes cloud task i to virtual machine j:
Then system produces the random number q ' that a codomain is (0,1), and cloud task i to each virtual machine select probability is done into tired
Which cumulative probability product probability statistics, random number q ' falls in, and the corresponding virtual machine of the cumulative probability is to determine distribution
Virtual machine.
4. the cloud method for allocating tasks according to claim 2 based on Bi-objective ant group algorithm, it is characterised in that:Step S3
In, when cloud task is assigned on virtual machine, if allowedkFor empty set, then allowed is reinitializedk, then by cloud task point
It is fitted on allowedkIn virtual machine on.
5. the cloud method for allocating tasks according to claim 4 based on Bi-objective ant group algorithm, it is characterised in that:Step S4
In, the detailed process for carrying out pheromones local updating is as follows:
τij(t+1) it is the pheromones updated, ρ is pheromones volatility coefficient,
6. the cloud method for allocating tasks according to claim 4 based on Bi-objective ant group algorithm, it is characterised in that:Step S6
In, the detailed process for carrying out global information element renewal is as follows:
τij(t+1) it is the global pheromones updated,
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510305822.5A CN104915557B (en) | 2015-06-04 | 2015-06-04 | A kind of cloud method for allocating tasks based on Bi-objective ant group algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510305822.5A CN104915557B (en) | 2015-06-04 | 2015-06-04 | A kind of cloud method for allocating tasks based on Bi-objective ant group algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104915557A CN104915557A (en) | 2015-09-16 |
CN104915557B true CN104915557B (en) | 2017-09-01 |
Family
ID=54084618
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510305822.5A Expired - Fee Related CN104915557B (en) | 2015-06-04 | 2015-06-04 | A kind of cloud method for allocating tasks based on Bi-objective ant group algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104915557B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106776008A (en) * | 2016-11-23 | 2017-05-31 | 福建六壬网安股份有限公司 | A kind of method and system that load balancing is realized based on zookeeper |
CN106951059A (en) * | 2017-03-28 | 2017-07-14 | 中国石油大学(华东) | Based on DVS and the cloud data center power-economizing method for improving ant group algorithm |
CN108846623B (en) * | 2018-09-17 | 2021-02-19 | 安吉汽车物流股份有限公司 | Whole vehicle logistics scheduling method and device based on multi-target ant colony algorithm, storage medium and terminal |
CN109615057B (en) * | 2018-11-29 | 2023-02-28 | 山东大学 | Self-organizing task allocation method based on dynamic response threshold value in foraging of swarm robots |
CN112000485B (en) * | 2020-09-01 | 2024-01-12 | 北京元心科技有限公司 | Task allocation method, device, electronic equipment and computer readable storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103024048A (en) * | 2012-12-17 | 2013-04-03 | 南京邮电大学 | Resources scheduling method under cloud environment |
US8416869B2 (en) * | 2010-05-05 | 2013-04-09 | National Yunlin University Of Science And Technology | Peak-to-average power ratio reduction method for orthogonal frequency division multiplexing systems based on path finding |
CN103699446A (en) * | 2013-12-31 | 2014-04-02 | 南京信息工程大学 | Quantum-behaved particle swarm optimization (QPSO) algorithm based multi-objective dynamic workflow scheduling method |
CN103970609A (en) * | 2014-04-24 | 2014-08-06 | 南京信息工程大学 | Cloud data center task scheduling method based on improved ant colony algorithm |
CN104461739A (en) * | 2014-12-15 | 2015-03-25 | 中山大学 | Cloudsim platform based virtual machine batch deployment method |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8887072B2 (en) * | 2007-12-07 | 2014-11-11 | Orange | Method of managing a set of virtual environment control entities, device, and corresponding computer program product |
-
2015
- 2015-06-04 CN CN201510305822.5A patent/CN104915557B/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8416869B2 (en) * | 2010-05-05 | 2013-04-09 | National Yunlin University Of Science And Technology | Peak-to-average power ratio reduction method for orthogonal frequency division multiplexing systems based on path finding |
CN103024048A (en) * | 2012-12-17 | 2013-04-03 | 南京邮电大学 | Resources scheduling method under cloud environment |
CN103699446A (en) * | 2013-12-31 | 2014-04-02 | 南京信息工程大学 | Quantum-behaved particle swarm optimization (QPSO) algorithm based multi-objective dynamic workflow scheduling method |
CN103970609A (en) * | 2014-04-24 | 2014-08-06 | 南京信息工程大学 | Cloud data center task scheduling method based on improved ant colony algorithm |
CN104461739A (en) * | 2014-12-15 | 2015-03-25 | 中山大学 | Cloudsim platform based virtual machine batch deployment method |
Non-Patent Citations (4)
Title |
---|
Research on Double Objective Optimization of Master Production Schedule Based on Ant Colony Algorithm;Wu Zheng-jia等;《2010 International Conference on Computational Intelligence and Security》;20101231;第200-204页 * |
基于负载均衡蚁群优化算法的云计算任务调度;张焕青灯;《微电子学与计算机》;20150531;第32卷(第5期);第31-35页 * |
基于资源状态蚁群算法的云计算任务分配;黄俊等;《计算机工程与设计》;20140930;第35卷(第9期);第3305-3309页 * |
蚁群算法研究评述;胡祥培等;《管理工程学报》;20081231;第22卷(第2期);第74-79页 * |
Also Published As
Publication number | Publication date |
---|---|
CN104915557A (en) | 2015-09-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104915557B (en) | A kind of cloud method for allocating tasks based on Bi-objective ant group algorithm | |
US10733026B2 (en) | Automated workflow selection | |
US20100293119A1 (en) | Systems and methods for parallel processing optimization for an evolutionary algorithm | |
CN107045456A (en) | A kind of resource allocation methods and explorer | |
CN103970609A (en) | Cloud data center task scheduling method based on improved ant colony algorithm | |
CN104023042B (en) | Cloud platform resource scheduling method | |
Peng et al. | A fast algorithm to find all-pairs shortest paths in complex networks | |
CN104104621B (en) | A kind of virtual network resource dynamic self-adapting adjusting method based on Nonlinear Dimension Reduction | |
CN106302170B (en) | A kind of resource allocation methods of wireless cloud computing system | |
CN106649391A (en) | Graph data processing method and apparatus | |
CN110297695A (en) | The system and method for variable bit rate limitation for shared resource access | |
CN109710406A (en) | Data distribution and its model training method, device and computing cluster | |
Tang et al. | Communication contention in APN list scheduling algorithm | |
CN105430074B (en) | Optimization method and system based on the distribution storage of the cloud data of data dependency and visit capacity | |
CN107704322A (en) | Ask distribution method and device | |
Oliker et al. | Efficient load balancing and data remapping for adaptive grid calculations | |
CN104618480A (en) | Cloud system source distributing method driven on basis of network link utilization rates | |
CN107329826A (en) | A kind of heuristic fusion resource dynamic dispatching algorithm based on Cloudsim platforms | |
CN104050189B (en) | The page shares processing method and processing device | |
CN110958192B (en) | Virtual data center resource allocation system and method based on virtual switch | |
CN101551749B (en) | Method and system of random test program generation and design verification method | |
CN104009904A (en) | Method and system for establishing virtual network for big data processing of cloud platform | |
CN115168058B (en) | Thread load balancing method, device, equipment and storage medium | |
CN116089083A (en) | Multi-target data center resource scheduling method | |
Xu et al. | Hybrid scheduling deadline-constrained multi-DAGs based on reverse HEFT |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170901 Termination date: 20180604 |
|
CF01 | Termination of patent right due to non-payment of annual fee |