CN106095529A - A kind of carrier wave emigration method under C RAN framework - Google Patents

Abstract

The invention discloses a kind of carrier wave emigration method under C RAN framework, described method includes: occur in resource pool that on carrier Virtual machine, CPU and internal memory comprehensive resources utilization rate are more than 80%, or less than 20%, triggers carrier wave emigration；Determine the whole user tasks needing to need to move out on the carrier Virtual machine of certain customers' task and the underload moved out on the carrier Virtual machine of overload；According to correlated variables, based on migrating time delay and two constraintss of resource, with energy consumption with migrate cost as target, define mathematic(al) representation, founding mathematical models；Solve mathematical model based on the max-min ant system improved, obtain the migration system of selection of different user task.The present invention is conducive to balancing the load of each carrier Virtual machine, sufficiently make use of the process resource of each carrier wave, reduces energy consumption, be advantageously implemented load balancing.

Description

A kind of carrier wave emigration method under C-RAN framework

Technical field

The invention belongs to communication technical field, particularly relate to a kind of carrier wave emigration method under C-RAN framework.

Background technology

Cloud radio access network (cloud-radio access network, C-RAN) is based on centralization process, association Deal with, the green wireless access planar network architecture of real-time cloud computing architecture.Traditional base station is all to dispose by " maximum processing capability ", Do not consider the dynamic variation characteristic of Network, i.e. tidal effect, cause utilization rate of equipment and installations low, and cause electric power to provide A large amount of wastes in source.The BBU baseband pool using centralized deployment can tackle tidal effect, opens dynamic carrier shift function After, the overall usage quantity of carrier wave can be saved, improve the utilization rate of carrier wave, and utilize dynamically distribution baseband pool to process resource, carry Rise systematic entirety energy power dissipation ratio.Existing think that carrier wave emigration i.e. undertakes about the method for carrier wave emigration under C-RAN framework The migration of the virtual machine of carrier processing, selects to be suitable for according to the triggering type of physical node and the load characteristic of virtual Multi-Mode Base Station The virtual Multi-Mode Base Station migrated, moves to it by the virtual machine being used for carrier processing on a station server by online migration pattern On his server, for load balancing；Or by the online migrating technology of virtual machine, physics clothes of carrier processing will be undertaken Business device (physical machine) moves to another station server by exchange network；When the carrier Virtual machine on server all moves to it After his server, in that context it may be convenient to it is overhauled or the attended operation such as upgrading, or be turned off reaching energy-saving and emission-reduction Effect.

Existing about under C-RAN framework carrier wave emigration method it is considered that determine from the load state of physical node be No migration, balance is the load of physical server, does not take into full account the load state of each carrier Virtual machine on it.Work as tide The when that nighttide effect occurring, it may appear that the number of users of the carrier service of some communities is too much, process resource not enough, and other one The number of users of a little cell serves is little, and resource is mostly in idle state, can not be very only by the mode migrating virtual machine Good ensures that resource pool each carrier Virtual machine interior load is in the state of balance, and the utilization rate of carrier wave is relatively low, and overall energy consumption is relatively High.

Summary of the invention

It is an object of the invention to provide a kind of carrier wave emigration method under C-RAN framework, it is intended to solve existing relevant In under C-RAN framework, the method existence of carrier wave emigration can not well ensure resource pool by the way of migrating virtual machine, each carries The load of ripple virtual machine is in the state of balance, and the utilization rate of carrier wave is relatively low, the problem that overall energy consumption is higher.

The present invention is achieved in that a kind of based on the carrier wave emigration method under C-RAN framework, and the method relates to mobile logical Letter field and computer realm, the carrier wave emigration method under described C-RAN framework comprises the following steps:

Step one, occurs in resource pool that on carrier Virtual machine, CPU and internal memory comprehensive resources utilization rate are more than 80%, or is less than 20%, trigger carrier wave emigration；

Step 2, determines the certain customers' task needing to move out on the carrier Virtual machine of overload and underload The whole user tasks moved out are needed on carrier Virtual machine；

Step 3, according to correlated variables, based on migrating time delay and two constraintss of resource, with energy consumption and migration cost be Target, defines mathematic(al) representation, founding mathematical models；

Step 4, solves mathematical model based on the max-min ant system improved, obtains the migration of different user task System of selection.

Further, the carrier Virtual machine of described overload need the certain customers' task moved out specifically instigate carrier wave The communication services such as the resource utilization of the virtual machine voice of part user more than 80%, data, video.

Further, the virtual resource needed for the user task moved out is processed under described correlated variables includes w consensus standard Amount Task_i(vr_c,vr_m), wherein vr_cAnd vr_mThe calculating needed for the user task moved out on carrier wave virtual machine i under w consensus standard Stock number and memory source amount；

The energy consumption of carrier Virtual machine is its CPU and internal memory energy consumption sum, and mathematic(al) representation is as follows:

$power\left({\mathrm{vm}}_i\right)=E_{cpu,{\mathrm{vm}}_i}+E_{ram,{\mathrm{vm}}_i};$

Wherein

WithIt is respectively carrier Virtual machine vm_iCPU and memory usage, α_cpu、γ_cpu、α_ramAnd γ_memFor The specific constant of model.

Further, described migration cost mathematic(al) representation is as follows:

Costmig(Task(vm_i),vm_j)=C_migD(vm_i,vm_j)；

Wherein, Task (vm_i) it is carrier Virtual machine vm_iThe upper task of needing to migrate, D (vm_i,vm_j) it is from virtual machine vm_iArrive Virtual machine vm_jTopology distance, C_migFor moving costs parameter, C_migIt is defined as follows:

f+M(Task(vm_i))/B(vm_i,vm_j)；

Wherein, f is fixed constant, M (Task (vm_i)) it is carrier Virtual machine vm_iOn the internal memory shared by task to be migrated, B (vm_i,vm_j) it is virtual machine vm_iAnd vm_jBetween transmission bandwidth；

Further, described mathematical model is defined as follows:

And

$s.t.\left\{\begin{array}{cc}\underset{j=1}{\overset{m}{\Σ}}{x}_{ij}{\mathrm{Task}}_j\left({\mathrm{vr}}_c\right)\≤{\mathrm{vm}}_{ic},& 1\≤i\≤n,j\∈M\\ \underset{j=1}{\overset{m}{\Σ}}{x}_{ij}{\mathrm{Task}}_j\left({\mathrm{vr}}_m\right)\≤{\mathrm{vm}}_{im},& 1\≤i\≤n,j\∈M\\ Tmig\left({\mathrm{Task}}_j\right)\≤T_{th},& j\∈M\\ \underset{i=1}{\overset{n}{\Σ}}{x}_{ij}=1,& j\∈M\end{array}\right.$

Wherein mathematic(al) representation Tmig (Task_j)≤T_thFor task immigration delay constraint condition, Tmig (Task_j) it is carrier wave Task immigration time delay on virtual machine j, T_thFor the largest tolerable time delay of user in communication, it is defined as 20ms；

Mathematic(al) representationWithFor resource constraint, expression is moved The resources requirement of shifting task is less than the available volume of resources of purpose carrier Virtual machine, vm_icAnd vm_imIt is respectively carrier Virtual machine i Amount of available computational resources and memory source amount；

N is the number of carrier Virtual machine total in resource pool, and M is the carrier Virtual machine set needing migration task, and m is collection Close carrier Virtual machine number in M,Expression task finally can only move on a virtual machine.

Further, the max-min ant system of described improvement is to change the definition of volatility coefficient in pheromone formula Entering, the form after improvement is as follows:

${\mathrm{\γ}}_{ku}(t+1)={\[(1-\mathrm{\ρ}(t\left)\right){\mathrm{\γ}}_{ku}\left(t\right)+{{\mathrm{\Δ\γ}}_{ku}}^{best}\]}_{{\mathrm{\γ}}_{\min }}^{{\mathrm{\γ}}_{\max }},$

Wherein, volatility coefficient ρ (t) of pheromone updates as follows:

Wherein, t is iterations, γ_kuFor migrating the task k selection carrier Virtual machine u pheromone as purpose virtual machine, Δγ_ku ^bestFor the pheromone increment size of optimal solution, γ_maxAnd γ_minFor the upper bound and the lower bound of pheromone, t_maxFor greatest iteration time Number, f (t-1) is that value arrives between 0 by normalization and the suitability degree function of weighting factor method definition by two optimization aim Between 1, ρ_inFor the initialization value of ρ, ρ_minFor the minima of ρ, n₁For the real number between 0 to 1.

Carrier wave emigration method under the C-RAN framework that the present invention provides, by using the cell carrier upper part of load weight The task immigration that family is submitted to is on the lighter cell carrier of duty factor, it is to avoid competing to resource of user on the carrier wave of load weight Strive, it is ensured that the service quality of user, and when the number of users that certain cell carrier services for a long time is little, can be by by this load The user task that ripple is corresponding all moves to an other carrier Virtual machine thus closes this carrier Virtual machine, reduces energy consumption.

The present invention passes through founding mathematical models, solves the migration system of selection obtaining different user task.Moved by task The mode moved, is conducive to balancing the load of each carrier Virtual machine, sufficiently make use of the process resource of each carrier wave, reduce Energy consumption, improves the service quality of user, improves the load balancing degrees that resource pool is overall simultaneously.

Accompanying drawing explanation

Fig. 1 is the carrier wave emigration method flow diagram under the C-RAN framework that the embodiment of the present invention provides.

Fig. 2 is the topology diagram that the C-RAN framework that the embodiment of the present invention provides downloads wavelength-division bunch.

Fig. 3 is the user task schematic diagram at bunch internal migration of embodiment of the present invention offer.

Fig. 4 be the embodiment of the present invention provide user task bunch between migrate schematic diagram.

Fig. 5 is based on improving resource pool energy consumption pair before and after the migration that obtains of MMAS (max-min ant system) optimized algorithm Than figure.

Fig. 6 is based on improving load balancing degrees pair before and after the migration that obtains of MMAS (max-min ant system) optimized algorithm Than figure.

Detailed description of the invention

In order to make the purpose of the present invention, technical scheme and advantage clearer, below in conjunction with embodiment, to the present invention It is further elaborated.Should be appreciated that specific embodiment described herein, only in order to explain the present invention, is not used to Limit the present invention.

The present invention by will the task immigration submitted to of the cell carrier upper part user of load weight to lighter little of duty factor On district's carrier wave, it is ensured that QoS of customer, and when the number of users that certain cell carrier services for a long time is little, can pass through will The user task that this carrier wave is corresponding all moves to an other carrier Virtual machine thus closes this carrier Virtual machine, reduces energy Consumption.By founding mathematical models, solve the migration system of selection obtaining different user task.

Below in conjunction with the accompanying drawings the application principle of the present invention is explained in detail.

As it is shown in figure 1, the carrier wave emigration method under the C-RAN framework of the embodiment of the present invention comprises the following steps:

Occur in S101, resource pool that on carrier Virtual machine, CPU and internal memory comprehensive resources utilization rate are more than 80%, or be less than 20%, trigger carrier wave emigration；

S102, determine the certain customers' task and the load of underload needing to move out on the carrier Virtual machine of overload The whole user tasks moved out are needed on ripple virtual machine；

S103, according to correlated variables, based on migrating time delay and two constraintss of resource, with energy consumption with migrate cost as mesh Mark, defines mathematic(al) representation, founding mathematical models；

S104, max-min ant system based on improvement solve mathematical model, obtain the migration choosing of different user task Selection method.

Below in conjunction with specific embodiment, the application principle of the present invention is further described.

The embodiment of the present invention provides the carrier wave emigration method under C-RAN framework, comprises the following steps:

Occur in S1, resource pool that on carrier Virtual machine, CPU and internal memory comprehensive resources utilization rate are more than 80%, or be less than 20%, trigger carrier wave emigration.

It should be noted that carrier Virtual machine described in step S1 specifically refers to the carrier processing resource for a community The virtual machine formed after virtualization.The configuration of each carrier Virtual resource is different according to the average portfolio in community.

CPU and the internal memory comprehensive resources utilization rate of carrier Virtual machine specifically refer to:

${\mathrm{vm}}_{util}=\frac{1}{2}(\frac{v_{cutil}}{{\mathrm{vm}}_{cpu}}+\frac{v_{mutil}}{{\mathrm{vm}}_{mem}}),$

Wherein, v_cutilAnd v_mutilCPU and memory source usage amount, vm for carrier Virtual machine_cpuAnd vm_memEmpty for carrier wave The CPU of plan machine and memory source configuration.

S2, determine the certain customers' task and the carrier wave of underload needing to move out on the carrier Virtual machine of overload The whole user tasks moved out are needed on virtual machine.

It should be noted that the user task in step S2 specifically refer to business such as voice that cell carrier processes in real time, Data, video traffic etc..

Need it is further noted that need the part moved out to use on the carrier Virtual machine of overload described in step S2 Family task specifically instigates the task of the resource utilization of the carrier Virtual machine part user more than 80%.

S3, according to correlated variables, based on migrating time delay and two constraintss of resource, with energy consumption with migrate cost as mesh Mark, defines mathematic(al) representation, founding mathematical models.

It should be noted that correlated variables described in step S3 includes processing the user task moved out under w consensus standard Required virtual resource amount Task_i(vr_c,vr_m), wherein vr_cAnd vr_mFor the user moved out on carrier wave virtual machine i under w consensus standard The amount of computational resources of task needs and memory source amount.

The energy consumption of carrier Virtual machine is its CPU and internal memory energy consumption sum, and mathematic(al) representation is as follows:

$power\left({\mathrm{vm}}_i\right)=E_{cpu,{\mathrm{vm}}_i}+E_{ram,{\mathrm{vm}}_i},$

Wherein

WithIt is respectively carrier Virtual machine vm_iCPU and memory usage.α_cpu、γ_cpu、α_ramAnd γ_memFor The specific constant of model.

Migrate cost mathematic(al) representation as follows:

Costmig(Task(vm_i),vm_j)=C_migD(vm_i,vm_j),

Wherein, Costmig (Task (vm_i),vm_j) it is carrier Virtual machine vm_iOn task task (vm_i) move to carrier wave Virtual machine vm_jUpper required migration cost, D (vm_i,vm_j) it is from virtual machine vm_iTo virtual machine vm_jTopology distance, C_migFor moving Move cost parameter.C_migIt is defined as follows:

f+M(Task(vm_i))/B(vm_i,vm_j),

Wherein, f is fixed constant, M (Task (vm_i)) it is carrier Virtual machine vm_iOn the internal memory shared by task to be migrated, B (vm_i,vm_j) it is virtual machine vm_iAnd vm_jBetween transmission bandwidth.

Need it is further noted that the mathematical model described in step S3 is defined as follows:

And

$s.t.\left\{\begin{array}{cc}\underset{j=1}{\overset{m}{\Σ}}{x}_{ij}{\mathrm{Task}}_j\left({\mathrm{vr}}_c\right)\≤{\mathrm{vm}}_{ic},& 1\≤i\≤n,j\∈M\\ \underset{j=1}{\overset{m}{\Σ}}{x}_{ij}{\mathrm{Task}}_j\left({\mathrm{vr}}_m\right)\≤{\mathrm{vm}}_{im},& 1\≤i\≤n,j\∈M\\ Tmig\left({\mathrm{Task}}_j\right)\≤T_{th},& j\∈M\\ \underset{i=1}{\overset{n}{\Σ}}{x}_{ij}=1,& j\∈M\end{array}\right.$

Wherein, mathematic(al) representation Tmig (Task_j)≤T_thFor task immigration delay constraint condition, Tmig (Task_j) for carrying Task immigration time delay on ripple virtual machine j, T_thFor the largest tolerable time delay of user in communication, it is defined as 20ms；

Mathematic(al) representationWithFor resource constraint, expression is moved The resources requirement of shifting task is less than the available volume of resources of purpose carrier Virtual machine, vm_icAnd vm_imIt is respectively carrier Virtual machine i Amount of available computational resources and memory source amount；

N is the number of carrier Virtual machine total in resource pool, and M is the carrier Virtual machine set needing migration task, and m is collection Close carrier Virtual machine number in M,Expression task finally can only move on a virtual machine.

S4, max-min ant system based on improvement solve mathematical model, and the migration obtaining different user task selects Method.

It should be noted that the max-min ant system improved described in step S4 refers to, wave in pheromone formula The definition sending out coefficient improves, and the form after improvement is as follows:

${\mathrm{\γ}}_{ku}(t+1)={\[(1-\mathrm{\ρ}(t\left)\right){\mathrm{\γ}}_{ku}\left(t\right)+{{\mathrm{\Δ\γ}}_{ku}}^{best}\]}_{{\mathrm{\γ}}_{\min }}^{{\mathrm{\γ}}_{\max }},$

Wherein, volatility coefficient ρ (t) of pheromone updates as follows:

Wherein, t is iterations, γ_kuFor migrating the task k selection carrier Virtual machine u pheromone as purpose virtual machine, Δγ_ku ^bestFor the pheromone increment size of optimal solution, γ_maxAnd γ_minFor the upper bound and the lower bound of pheromone, t_maxFor greatest iteration time Number, f (t-1) is that value arrives between 0 by normalization and the suitability degree function of weighting factor method definition by two optimization aim Between 1, ρ_inFor the initialization value of ρ, ρ_minFor the minima of ρ, n₁For the real number between 0 to 1.

Fig. 2 gives the topology diagram of the carrier wave sub-clustering under C-RAN framework, the carrier wave in resource pool with bunch form enter Line pipe is managed, and each bunch of carrier Virtual machine containing protocol type of the same race, each carrier Virtual machine is according to its community serviced Average traffic and have different resource distributions.

Fig. 3 and Fig. 4 gives the schematic diagram that user task migrates, and needs migrate to appoint when being calculated each by step S4 After purpose carrier Virtual machine selected by business, under network topology structure based on sub-clustering, in fact it could happen that bunch internal migration and bunch Between migrate two kinds of situations.Due to will be through multistage switching network across bunch migration, the migration cost of consuming be big, therefore, when bunch in When there is the suitable carrier virtual machine of acceptable migration task, in order to reduce migration cost, by selection bunch internal migration, otherwise, By by bunch between migration by the way of find suitable purpose carrier Virtual machine.First monitoring resource module monitors to bunch internal memory Carrier Virtual machine meet migrate trigger condition, then informing removal management module according to migrate task resources requirement and Migrate the constraints that time delay should meet, by setting up the migration system of selection of seismic responses calculated different user task.

Fig. 5 gives based on improving resource pool energy before and after the migration that obtains of MMAS (max-min ant system) optimized algorithm Consumption comparison diagram.Emulation arranges resource pool same kind carrier Virtual machine number configure by basic, normal, high 1:1:1, every kind of configuration bag Contain the carrier Virtual machine of equally distributed different resource utilization rate.After can be seen that migration, the total energy consumption of resource pool reduces, Because by the way of migration, it is achieved readjusting of task, turned off the carrier Virtual machine that resource utilization is low, alleviated simultaneously The user competition to resource on the carrier wave that load is high.When carrier Virtual machine number total in resource pool increases when, need The user task migrated increases therewith, and the energy consumption reduced by migration is the most more.

Fig. 6 gives based on improving load balancing before and after the migration that obtains of MMAS (max-min ant system) optimized algorithm Degree comparison diagram.Load balancing degrees is defined as:

$B=\sqrt{\frac{1}{n}*\underset{i=1}{\overset{n}{\Σ}}{(u_i-\stackrel{\‾}{u})}^2};$

u_iFor the resource utilization of carrier Virtual machine i,For the average of the resource utilization of all carrier Virtual machines, n is money The number of the carrier Virtual machine of the total protocol type of the same race opened in pond, source.The value of B is the least, the resource of each carrier Virtual machine Utilization rate value is the most close, and load balancing degrees is the best.By analogous diagram it can be seen that before Qian Yiing, the differently configured carrier Virtual of resource pool In the case of machine number, load balancing degrees is in a close level.After migration, the value of B reduces, this is because by moving Moving, the carrier Virtual office making resource utilization low closes, and by the carrier Virtual machine of its whole tasks and high load condition Partial task moves to resource utilization and is on the carrier Virtual machine under normal condition, makes the load being ultimately under opening The resource utilization of ripple virtual machine is in an approximately equalised state, improves load balancing degrees.

The foregoing is only presently preferred embodiments of the present invention, not in order to limit the present invention, all essences in the present invention Any amendment, equivalent and the improvement etc. made within god and principle, should be included within the scope of the present invention.

Claims (5)

1. the carrier wave emigration method under a C-RAN framework, it is characterised in that the carrier wave emigration method under described C-RAN framework Comprise the following steps:

Step one, occurs in resource pool that on carrier Virtual machine, CPU and internal memory comprehensive resources utilization rate are more than 80%, or is less than 20%, trigger carrier wave emigration；

Step 2, determines the certain customers' task and the carrier wave of underload needing to move out on the carrier Virtual machine of overload The whole user tasks moved out are needed on virtual machine；

Step 3, according to correlated variables, based on migrating time delay and two constraintss of resource, with energy consumption with migrate cost as mesh Mark, defines mathematic(al) representation, founding mathematical models；

Step 4, solves mathematical model based on max-min ant system, obtains the migration system of selection of different user task.

2. the carrier wave emigration method under C-RAN framework as claimed in claim 1, it is characterised in that the load of described overload Need on ripple virtual machine that the certain customers' task moved out specifically instigates the resource utilization of carrier Virtual machine more than 80% that The voice of certain customers, data, the communication service of video.

3. the carrier wave emigration method under C-RAN framework as claimed in claim 1, it is characterised in that described correlated variables includes w Virtual resource amount Task needed for the user task moved out is processed under consensus standard_i(vr_c,vr_m), wherein vr_cAnd vr_mAssist for w The amount of computational resources of the user task needs moved out on view standard download ripple virtual machine i and memory source amount；

The energy consumption of carrier Virtual machine is its CPU and internal memory energy consumption sum, and mathematic(al) representation is as follows:

$power\left({\mathrm{vm}}_i\right)=E_{cpu,{\mathrm{vm}}_i}+E_{ram,{\mathrm{vm}}_i};$

Wherein

WithIt is respectively carrier Virtual machine vm_iCPU and memory usage, α_cpu、γ_cpu、α_ramAnd γ_memFor model Specific constant；

The migration cost mathematic(al) representation of task is as follows:

Costmig(Task(vm_i),vm_j)=C_migD(vm_i,vm_j)；

Wherein, Task (vm_i) it is carrier Virtual machine vm_iThe upper task of needing to migrate, D (vm_i,vm_j) it is from virtual machine vm_iTo virtual Machine vm_jTopology distance, C_migFor moving costs parameter, C_migIt is defined as follows:

f+M(Task(vm_i))/B(vm_i,vm_j)；

Wherein, f is fixed constant, M (Task (vm_i)) it is carrier Virtual machine vm_iOn the internal memory shared by task to be migrated, B (vm_i,vm_j) it is virtual machine vm_iAnd vm_jBetween transmission bandwidth.

4. the carrier wave emigration method under C-RAN framework as claimed in claim 1, it is characterised in that described mathematical model tool Body is defined as follows:

And

$s.t.\left\{\begin{array}{cc}\underset{j=1}{\overset{m}{\Σ}}{x}_{ij}{\mathrm{Task}}_j\left({\mathrm{vr}}_c\right)\≤{\mathrm{vm}}_{ic},& 1\≤i\≤n,j\∈M\\ \underset{j=1}{\overset{m}{\Σ}}{x}_{ij}{\mathrm{Task}}_j\left({\mathrm{vr}}_m\right)\≤{\mathrm{vm}}_{im},& 1\≤i\≤n,j\∈M\\ Tmig\left({\mathrm{Task}}_j\right)\≤T_{th},& j\∈M\\ \underset{i=1}{\overset{n}{\Σ}}{x}_{ij}=1,& j\∈M\end{array}\right.$

Wherein, mathematic(al) representation Tmig (Task_j)≤T_thFor task immigration delay constraint condition, Tmig (Task_j) it is carrier Virtual Task immigration time delay on machine j, T_thFor the largest tolerable time delay of user in communication, it is defined as 20ms；

Mathematic(al) representationWithFor resource constraint, represent to migrate and appoint The resources requirement of business is less than the available volume of resources of purpose carrier Virtual machine, vm_icAnd vm_imRespectively carrier Virtual machine i's can By amount of computational resources and memory source amount；

N is the number of carrier Virtual machine total in resource pool, and M is the carrier Virtual machine set needing migration task, and m is set M Middle carrier Virtual machine number,Expression task finally can only move on a virtual machine.

5. the carrier wave emigration method under C-RAN framework as claimed in claim 1, it is characterised in that the maximum of described improvement is Little ant system is to improve the definition of volatility coefficient in pheromone formula, and the form after improvement is as follows:

${\mathrm{\γ}}_{ku}(t+1)={\[(1-\mathrm{\ρ}(t\left)\right){\mathrm{\γ}}_{ku}\left(t\right)+{{\mathrm{\Δ\γ}}_{ku}}^{best}\]}_{{\mathrm{\γ}}_{\min }}^{{\mathrm{\γ}}_{\max }},$

Wherein, volatility coefficient ρ (t) of pheromone updates as follows:

Wherein, t is iterations, γ_kuSelect carrier Virtual machine u as the pheromone of purpose virtual machine, Δ for migrating task k γ_ku ^bestFor the pheromone increment size of optimal solution, γ_maxAnd γ_minFor the upper bound and the lower bound of pheromone, t_maxFor greatest iteration time Number, f (t-1) is that value arrives between 0 by normalization and the suitability degree function of weighting factor method definition by two optimization aim Between 1, ρ_inFor the initialization value of ρ, ρ_minFor the minima of ρ, n₁For the real number between 0 to 1.