CN112256415B - Micro cloud load balancing task scheduling method based on PSO-GA - Google Patents
Micro cloud load balancing task scheduling method based on PSO-GA Download PDFInfo
- Publication number
- CN112256415B CN112256415B CN202011118823.6A CN202011118823A CN112256415B CN 112256415 B CN112256415 B CN 112256415B CN 202011118823 A CN202011118823 A CN 202011118823A CN 112256415 B CN112256415 B CN 112256415B
- Authority
- CN
- China
- Prior art keywords
- micro
- task
- cloud
- clouds
- population
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 23
- 230000004044 response Effects 0.000 claims abstract description 31
- 239000002245 particle Substances 0.000 claims description 14
- 230000035772 mutation Effects 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 6
- 230000001149 cognitive effect Effects 0.000 claims description 3
- 230000003247 decreasing effect Effects 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 3
- 230000005012 migration Effects 0.000 abstract description 7
- 238000013508 migration Methods 0.000 abstract description 7
- 238000009826 distribution Methods 0.000 description 8
- 239000011159 matrix material Substances 0.000 description 8
- 230000008569 process Effects 0.000 description 3
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 238000013500 data storage Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000009827 uniform distribution Methods 0.000 description 2
- 229920002430 Fibre-reinforced plastic Polymers 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 239000011151 fibre-reinforced plastic Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/48—Program initiating; Program switching, e.g. by interrupt
- G06F9/4806—Task transfer initiation or dispatching
- G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
- G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5083—Techniques for rebalancing the load in a distributed system
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention relates to a micro cloud load balancing task scheduling method based on PSO-GA, which comprises the following steps: step S1, initializing micro-cloud set parameters; s2, calculating the local task response time of all the micro clouds; s3, sorting the micro clouds according to the local task response time of the micro clouds, and dividing the micro cloud set into an overload set and an underload set; and S4, acquiring an optimal micro cloud task scheduling scheme by adopting an improved PSOGA algorithm based on constraint conditions. The method and the device can effectively reduce the response time of the migration task and improve the dispatching efficiency of the micro cloud load task.
Description
Technical Field
The invention belongs to the technical field of micro clouds, and particularly relates to a PSO-GA-based micro cloud load balancing task scheduling method.
Background
In recent years, with the continuous development of mobile communication technology and internet of things, the number of mobile devices is increasing in blowout type. Through various applications on mobile devices, people's lives are increasingly more convenient. However, it is difficult to meet the user's demands for low latency and high reliability due to limited computing power, battery and data storage capacity of the mobile device itself. The cloud in the mobile cloud computing has rich computing resources and data storage space, and can efficiently process task requests of application programs on the mobile device, so that the problems of limited computing resources and battery energy consumption of the portable mobile device are solved. However, migrating tasks on the mobile device to the remote cloud can cause long communication delays, which can degrade the QoS experience of the user.
But the advent of micro cloud technology has solved this problem. A micro cloud is a group of clusters of computers that are close to a user and connected through a wireless network. In the three-layer MCC framework (mobile device, micro cloud and remote cloud respectively), the micro cloud is in the middle layer, so that the mobile device can access the micro cloud through a wireless network with lower delay, and further real-time interaction response is achieved. Moreover, when the micro cloud is in a loaded state, the delay tolerant task can be migrated to the remote cloud for execution. The deployment of the micro cloud infrastructure is similar to the deployment of wireless access points, and many researchers discuss how to deploy micro clouds in public wireless metropolitan area networks so that the public can enjoy the efficient services provided by the micro clouds. Because of the large population density in metropolitan areas, mobile users are not starved for access to the clouds.
However, one major problem that the micro-clouds need to handle is how to distribute task requests that mobile users migrate so that the load between the micro-clouds can be balanced. It is common practice to distribute the task requests of users to the nearest clouds for execution, but when the number of users in one area is excessive, the corresponding clouds will be loaded too much. If a large number of user task requests are suddenly accepted on the micro cloud, the response time of the task becomes long, which affects the experience of the user.
Disclosure of Invention
Therefore, the invention aims to provide the micro-cloud load balancing task scheduling method based on the PSO-GA, which can effectively reduce the response time of migration tasks and improve the micro-cloud load task scheduling efficiency.
In order to achieve the above purpose, the invention adopts the following technical scheme:
a micro cloud load balancing task scheduling method based on PSO-GA comprises the following steps:
step S1, initializing micro-cloud set parameters;
step S2, calculating the local task response time of all the micro clouds
S3, sorting the micro clouds according to the local task response time of the micro clouds, and dividing the micro cloud set into an overload set and an underload set;
and S4, acquiring an optimal micro cloud task scheduling scheme by adopting an improved PSOGA algorithm based on constraint conditions.
Further, the micro-cloud aggregation parameters comprise the service rate, the number of servers and the task arrival rate of each micro cloud i and the network delay among the micro clouds.
Further, the step S4 specifically includes:
step S41 random initialization P num Individual, and check whether the location of the generated individual satisfies the condition;
step S42, iterating the population, and obtaining global extremum gbest of the population after N iterations
And S43, obtaining an optimal task scheduling strategy under the corresponding micro-cloud set division according to the global extremum gbest of the population.
Further, the step S42 specifically includes:
i) After the inertial weight omega is calculated, updating the position and the speed of the individual, and checking the calculated position and speed;
ii) selecting a part of individuals from the population to execute mutation operation according to the mutation probability, and checking the positions of the individuals;
iii) Calculating fitness values corresponding to all individuals;
iv) updating the individual extremum of the particle with the global extremum of the population.
Further, the calculating the fitness value corresponding to all the individuals specifically includes:
i) By solving for X k Updating the task arrival rate λ of all the micro clouds using the following equation;
ii) task latency to compute an overloaded micro-cloudAnd task latency of underloaded micro cloud +.>Time delay T with network net (j)；
iii) And taking the maximum task response time in the micro cloud set at the moment as the fitness value of the individual.
Further, the updating of the particle position and the particle velocity is specifically:
updating the location and speed of the individual is accomplished using the following,
V k (t+1)＝ωV k (t)+c 1 r 1 (pbest k -X k (t))+c 2 r 2 (gbest-X k (t)) (17)
X k (t+1)＝X k (t)+V k (t+1) (18)
wherein t represents the number of iterations required by the population, c 1 And c 2 R is the learning factor of the population 1 And r 2 Is a random number between (0, 1), and ω is an inertial weight; v (V) k (t) is recorded as a memory term representing the velocity V of the last iteration k (t) for the current iteration speed V k The effect of the magnitude and direction of (t+1); c 1 r 1 (pbest k -X k (t)) is recorded as a self-cognition item, is a vector from the current position to the self-optimal position of the individual, and represents that the movement of the individual is somewhat from own experience; c 2 r 2 (gbest-X k (t)) is noted as a population cognitive term, which is a vector of individuals from a current location to an optimal location in the population;
and checking the individual according to the following formula;
equation (13) indicates that the task flow of any one of the set of overloaded clouds i to redirect to other underloaded clouds must be less thanEqual to its task arrival rate lambda i The method comprises the steps of carrying out a first treatment on the surface of the Equation (14) represents that the total task flow received by any one of the clouds j in the underrun cloudlet set must be less than the total service rate of its server minus its own task arrival rate lambda j 。
Further, in the updating of the particle position and velocity, a linear decreasing strategy of inertia weight is added to the formula (17), as follows
ω＝ω max -t·(ω max -ω min )/N，ω∈[ω min ，ω max ] (19) 。
Further, the updating the individual extremum of each particle and the global extremum of the population specifically includes:
wherein equation (20) represents obtaining the self-optimal position of the kth individual since the t-th iteration, the arg min function being used to return the self-position X with the smallest fitness value from the 1 st to the t-th iteration; equation (21) represents the optimal position in the population in the t-th iteration, and the arg min function is used to obtain the position X corresponding to the individual with the smallest fitness value in the population.
Compared with the prior art, the invention has the following beneficial effects:
the method and the device can effectively reduce the response time of the migration task and improve the dispatching efficiency of the micro cloud load task.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a graph showing task response times for 3 algorithms in different configurations in accordance with one embodiment of the present invention;
FIG. 3 is a run-time of 3 algorithms in different configurations in an embodiment of the invention.
Detailed Description
The invention will be further described with reference to the accompanying drawings and examples.
In this embodiment, the micro cloud provider has established K micro clouds {1,2,., K } in an area, both deployed in a fixed location with the wireless access point, and the micro clouds are connected through a wireless network, so that data transmission can be performed between any two micro clouds. It is assumed that an application on a mobile device can be dynamically arbitrarily divided into multiple tasks while the application is running, and that these tasks can be migrated to any one of the K clouds for execution (the mobile user's task will preferentially migrate into the nearby clouds).
Modeling the micro cloud by using an M/M/n queuing model, wherein M/M/n represents that time intervals of sequential arrival of task flows migrated by users are exponential distribution, task service time on the micro cloud is exponential distribution and a single micro cloud has n servers [16] . That is, it indicates that the micro cloud i has n i Station server, and service rate of each server is mu i Where i e {1,2,..k }. Furthermore, each micro cloud i has an important parameter, namely the task arrival rate lambda i . It refers to the average number of tasks that arrive per unit time. For each cloudlet i, sampling from normal distribution N (5, 2) > 0 to determine cloudlet service rate μ i And samples from a poisson distribution with a mean value of 3 to determine the number n of micro cloud servers i . Task arrival rate lambda due to micro cloud i i Cannot exceed its total service rate mu i ·n i Otherwise, the queuing waiting time of the micro cloud i is infinitely prolonged, so that the task arrival rate lambda i By distributing 0 < N (15, 6) < mu from normal i ·n i -0.25 samples for designation. The average waiting time of the tasks in each micro cloud i consists of two parts, namely task service time and task queuing time. Calculating task latency T using equation (1) i This time represents the average task latency of the cloudlet i at a task arrival rate λ.
The formula (2) is an Erlang C formula, and is used for calculating the probability that a task cannot be immediately processed in the micro cloud and needs to wait according to the number n of servers and the traffic intensity ρ.
The task arrival rate λ will be different on each micro cloud due to the randomness of the user's service requests. The task arrival rate λ of some cloudlets may be relatively large such that the cloudlets are in an overload state, resulting in increased task latency. And the task arrival rate lambda of other micro clouds is smaller, so that most service resources of the micro clouds are in an idle state. Let the cloudlet redirect some of its tasks to any other cloudlet and use f (i, j) to represent the task flow that cloudlet i redirects to cloudlet j, with the following restrictions on f (i, j):
where equation (3) indicates that the task flow f (i, j) of any of the clouds i through j is the negative of the task flow of the clouds j through i, and that the task flow of any of the clouds through i is zero. Equation (4) indicates that the sum of all task flows sent or received by all the micro clouds is equal to zero, i.e., the task flows between the micro clouds are conservative. Equation (5) shows that the sum of task flows sent by the micro cloud i must be less than or equal to the task arrival rate lambda of the micro cloud i i (here the task stream received by the mini-cloud i is ignored).
Because the cloudlet needs to transmit over the wireless network while redirecting some of the task flows to other cloudlets, this will create a corresponding network delay. Assuming that all migrated packets are the same size, the network delay created by transmitting any packet between a pair of clouds is consistent. Using C.epsilon.R K×K To represent a network delay matrix, where c i，j Representing the network delay that a task passes from microcloud i to microcloud j. The network delay c of the micro cloud is specified by sampling in normal distribution of 0.1-N (0.15,0.05) -0.2. Thus, the network delay required by the redirection task flow f (i, j) < 0 accepted by the micro cloud i is equal to-f (i, j) ·c i，j . The total network delay required by each micro cloud in receiving the redirection task is calculated using equation (6).
Finally, we use D (i) to represent the average task response time for executing all tasks on the mini-cloud i and calculate by equations (1), (6).
Wherein,,and the arrival rate of the remaining tasks of the micro cloud i is represented. Because a micro cloud may redirect a portion of its task flow to or receive a portion of its task flow from other micro clouds, it is necessary to recalculate its task arrival rate when calculating the task response time for each micro cloud.
In this embodiment, the micro cloud load balancing task scheduling problem is defined as follows. Fixation in citiesK micro clouds are deployed on the position, and each micro cloud i is provided with n i Station server, service rate of the server is mu i And the arrival rate of the task on the micro cloud i is lambda i (for convenience we will refer to the above parameters collectively by the symbol CLD). Our goal is to find a set of task flows f= { f (i, j) |i, j e {1, 2..once., K } (this task flow must meet the requirements of formulas (3), (4), (5)), through which all clouds are scheduled. This allows all of the clouds to minimize the maximum task response time D (i) in the clouds while they are in a load balanced state. Namely, formula (9):
in this embodiment, the micro clouds are first ordered according to their local task response time (the local task response time refers to the time it takes for all the micro clouds to process only local tasks without task distribution).
Then using a given micro-cloud p, divide the micro-cloud set into an overloaded micro-cloud set Vs and an underloaded micro-cloud set Vt with its local task response time (equations (10), (11)), and ensure that the task flow can only be redirected from the overloaded micro-cloud into the underloaded micro-cloud.
Using equation (12) as a representation of the solution, where X (i, j) represents the sum of the values from the cloudlet i ε V s To micro cloud j epsilon V t And X is a two-dimensional matrix of |vs|×|vt| (equation (12)).
V s ＝{i|T i (λ i )＞T p (λ p )}，i∈{1，2，...，K} (10)
V t ＝{j|T j (λ j )≤T p (λ p )}，j∈{1，2，...，K} (11)
For the two-dimensional matrix X (i, j), there are the following limitations:
wherein equation (13) indicates that any one of the set of overloaded clouds must have its task flow redirected to the other underloaded clouds less than or equal to its task arrival rate λ i The method comprises the steps of carrying out a first treatment on the surface of the Equation (14) shows that the total task flow received by any one of the clouds j in the underrun cloudlet set must be less than its server total service rate minus its own task arrival rate λ j 。
Referring to fig. 1, in this embodiment, a method for scheduling a micro cloud load balancing task based on a PSO-GA is provided, including the following steps:
step S1, initializing micro-cloud aggregation parameters, including the service rate, the number of servers, the task arrival rate and the network delay among the micro-clouds;
step S2, calculating the local task response time of all the micro clouds
S3, sorting the micro clouds according to the local task response time of the micro clouds, and dividing the micro cloud set into an overload set and an underload set;
and S4, acquiring an optimal micro cloud task scheduling scheme by adopting an improved PSOGA algorithm based on constraint conditions.
The specific procedure of the PSOGA algorithm employed is shown in algorithm 1.
In this embodiment, the step S4 specifically includes:
step S41: random initialization P num Individual, and check whether the location of the generated individual satisfies the condition;
step S42: iterating the population, and obtaining a global extremum gbest of the population after N iterations
Step S43: and obtaining an optimal task scheduling strategy under the corresponding micro-cloud set division according to the global extremum gbest of the population.
Preferably, in this embodiment, an initial population of PSOGA algorithms is constructed.
First, related parameters including population size Pnum, maximum number of iterations N, etc. need to be set.
Next, the initial population is randomly generatedSo as to iterate.
For individual X k X in (2) k (i) (i-th row representing two-dimensional matrix) according to uniform distributionTo select random numbers for filling.
If randomly generated individuals X k And (3) violating the formulas (13) and (14), selecting the relevant rows or columns of the two-dimensional matrix to randomly reduce until the individual meets the requirements.
Preferably, in this embodiment, the speed of the individual also needs to be initialized. For individual X k Velocity V of (3) k (i) (representing the ith row of the two-dimensional matrix) using uniform distribution To specify. Here, the uniformly distributed range of values selects 10% of the task arrival rate of the overload microcloud i.
Preferably, in this embodiment, the step S42 specifically includes:
i) After the inertial weight omega is calculated, updating the position and the speed of the individual, and checking the calculated position and speed;
ii) selecting a part of individuals from the population to execute mutation operation according to the mutation probability, and checking the positions of the individuals;
iii) Calculating fitness values corresponding to all individuals;
iv) updating the individual extremum of the particle with the global extremum of the population.
Preferably, in this embodiment, the calculating fitness values corresponding to all the individuals specifically includes:
i) By solving for X k Updating the task arrival rate λ of all the micro clouds using the following equation;
ii) task latency to compute an overloaded micro-cloudAnd task latency of underloaded micro cloud +.>Time delay T with network net (j)；
iii) And taking the maximum task response time in the micro cloud set at the moment as the fitness value of the individual.
In this embodiment, it is preferable that updating of the position and speed of the individual is achieved using the following formulas, and the individual is checked according to formulas (13), (14);
V k (t+1)＝ωV k (t)+c 1 r 1 (pbest k -X k (t))+c 2 r 2 (gbest-X k (t)) (17)
X k (t+1)＝X k (t)+V k (t+1) (18)
wherein t represents the number of iterations required by the population, c 1 And c 2 R is the learning factor of the population 1 And r 2 Is a random number between (0, 1), and ω is an inertial weight; v (V) k (t) is recorded as a memory term representing the velocity V of the last iteration k (t) for the current iteration speed V k The effect of the magnitude and direction of (t+1); c 1 r 1 (pbest k -X k (t)) is recorded as a self-cognition item, is a vector from the current position to the self-optimal position of the individual, and represents that the movement of the individual is somewhat from own experience; c 2 r 2 (gbest-X k (t)) is noted as a population cognitive term, which is a vector of individuals from a current location to an optimal location in the population;
preferably, in the updating of the particle position and the particle speed, an inertia weight linear decreasing strategy is added to a formula (17), and the formula is as follows:
ω＝ω max -t·(ω max -ω min )/N，ω∈[ω min ，ω max ] (19)
after the position and speed of the individual are calculated, the individual extremum of each particle and the global extremum of the population need to be updated. Equation (20) represents the acquisition of the self-optimal position of the kth individual since the t-th iteration, the arg min function being used to return the self-position X with the smallest fitness value from the 1 st to the t-th iteration. Equation (21) represents the optimal position in the population in the t-th iteration, and uses the arg min function to obtain the position X corresponding to the individual with the smallest fitness value in the population.
Finally, the mutation operator of the GA algorithm needs to set individuals in the population firstlyVariation probability R of (2) mut . When the population performs an iterative operation, if at [0,1]The internally generated random number rand_num is smaller than the variation probability R mut The current individual is selected to perform the mutation operation. The mutation process is as follows:
first, the data from the set 1,2, V s Selecting a subset { i } from the group consisting of 1 ，i 2 ，...，i p }，2≤p≤|V s I (I); from the set {1, 2., |v s Selecting a subset { j } from 1 ，j 2 ，...，j q }，2≤q≤|V t | a. The invention relates to a method for producing a fibre-reinforced plastic composite. Wherein the subset { i } 1 ，i 2 ，...，i p The row subscript of the two-dimensional matrix X corresponds to the subset j 1 ，j 2 ，...，j q The column subscripts of the two-dimensional matrix X. Next, p×q values for the corresponding locations in the feasible solution are reinitialized. Finally, checking whether the mutated individual meets the requirements of formulas (13) and (14).
Example 1:
in this embodiment, the network delay between the micro clouds is proportional to the physical distance between them. The task arrival rate, service rate and network delay between the micro clouds of each micro cloud are all subjected to normal distribution, the number of the micro cloud servers adopts poisson distribution with the mean value of 3, and specific parameter settings are listed in table 1.
Table 1 experimental parameters
And then selecting two algorithms for comparison, namely a random migration algorithm and a greedy algorithm. In the random migration algorithm, m micro clouds with larger local task response time are randomly selected to carry out task migration, and the migrated task quantity is also randomly selected. And the greedy algorithm selects the micro cloud with the largest local task response time each time to migrate the task to the micro cloud with the smallest local task response time until the task response time of all the micro clouds is smaller than the given constraint. The above algorithms were all run on a server with a 2.20GHz Intel (R) i5 CPU and 8GiB RAM, and the 3 algorithms were each executed 10 times in each configuration environment, then the results were averaged 10 times.
Table 2 shows the maximum task response times of the PSOGA, stochastic migration, and greedy algorithms before and after task scheduling under different configurations. Figure 2 more intuitively reflects the minimum task response time that can be achieved by 3 algorithms in 6 configuration environments.
Table 2 maximum task response time before and after 3 algorithm scheduling under different configurations
It can be seen that the method of the present invention has a shorter task response time than the other two algorithms. This is because the PSOGA algorithm searches for the optimal solution in the problem domain, and once a better feasible solution is found, all individuals are continually drawn toward it, and finally the algorithm will obtain the optimal solution that can be found in the solution space.
The foregoing description is only of the preferred embodiments of the invention, and all changes and modifications that come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.
Claims (5)
1. The micro cloud load balancing task scheduling method based on PSO-GA is characterized by comprising the following steps of:
step S1, initializing micro-cloud set parameters;
s2, calculating the local task response time of all the micro clouds;
s3, sorting the micro clouds according to the local task response time of the micro clouds, and dividing the micro cloud set into an overload set and an underload set;
s4, acquiring an optimal micro cloud task scheduling scheme by adopting an improved PSOGA algorithm based on constraint conditions;
the step S4 specifically includes:
step S41 random initialization P num Individual, and check whether the location of the generated individual satisfies the condition;
step S42, iterating the population, and obtaining a global extremum gbest of the population after N iterations;
s43, obtaining an optimal task scheduling strategy under the corresponding micro-cloud set division according to the global extremum gbest of the population;
the step S42 specifically includes:
i) After the inertial weight omega is calculated, updating the position and the speed of the individual, and checking the calculated position and speed;
ii) selecting a part of individuals from the population to execute mutation operation according to the mutation probability, and checking the positions of the individuals;
iii) Calculating fitness values corresponding to all individuals;
iv) updating individual extremum of the particle and global extremum of the population;
the calculating of the fitness value corresponding to all the individuals specifically comprises the following steps:
i) By solving for X k Updating the task arrival rate λ of all the micro clouds using the following equation;
ii) task latency to compute an overloaded micro-cloudAnd task latency of underloaded micro cloud +.>Time delay T with network net (j)；
iii) And taking the maximum task response time in the micro cloud set at the moment as the fitness value of the individual.
2. The PSO-GA based micro cloud load balancing task scheduling method of claim 1, wherein the micro cloud set parameters include a service rate, a number of servers and a task arrival rate of each micro cloud i, and a network delay between the micro clouds.
3. The method for scheduling the micro cloud load balancing task based on the PSO-GA according to claim 1, wherein the updating of the particle position and the particle speed is specifically:
updating the location and speed of the individual is accomplished using the following,
V k (t+1)＝ωV k (t)+c 1 r 1 (pbesy k -X k (t))+c 2 r 2 (gbest-X k (t)) (17)
X k (t+1)＝X k (t)+V k (t+1) (18)
wherein t represents the number of iterations required by the population, c 1 And c 2 R is the learning factor of the population 1 And r 2 Is a random number between (0, 1), and ω is an inertial weight; v (V) k (t) is recorded as a memory term representing the velocity V of the last iteration k (t) for the current iteration speed V k The effect of the magnitude and direction of (t+1); c 1 r 1 (pbest k -X k (t)) is recorded as a self-cognition item, is a vector from the current position to the self-optimal position of the individual, and represents that the movement of the individual is somewhat from own experience;
c 2 r 2 (gbest-X k (t)) is noted as a population cognitive term, which is a vector of individuals from a current location to an optimal location in the population;
and checking the individual according to the following formula;
equation (13) indicates that any one of the set of overloaded clouds must have its task arrival rate λ less than or equal to the task flow that redirects any one of the set of overloaded clouds i to the other underloaded clouds i The method comprises the steps of carrying out a first treatment on the surface of the Equation (14) represents that the total task flow received by any one of the clouds j in the underrun cloudlet set must be less than the total service rate of its server minus its own task arrival rate lambda j 。
4. The method for scheduling micro cloud load balancing tasks based on PSO-GA as recited in claim 3, wherein the particle position and velocity update is performed by adding an inertia weight linear decreasing strategy to the formula (17) as follows
ω＝ω max -t·(ω max -ω min )/N，ω∈[ω min ，ω max ] (19)。
5. The method for scheduling the micro cloud load balancing task based on the PSO-GA according to claim 1, wherein updating the individual extremum of each particle and the global extremum of the population is specifically as follows:
wherein equation (20) represents obtaining the self-optimal position of the kth individual since the t-th iteration, the arg min function being used to return the self-position X with the smallest fitness value from the 1 st to the t-th iteration; equation (21) represents the optimal position in the population in the t-th iteration, and the arg min function is used to obtain the position X corresponding to the individual with the smallest fitness value in the population.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011118823.6A CN112256415B (en) | 2020-10-19 | 2020-10-19 | Micro cloud load balancing task scheduling method based on PSO-GA |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011118823.6A CN112256415B (en) | 2020-10-19 | 2020-10-19 | Micro cloud load balancing task scheduling method based on PSO-GA |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112256415A CN112256415A (en) | 2021-01-22 |
| CN112256415B true CN112256415B (en) | 2023-08-04 |
Family
ID=74244959
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011118823.6A Active CN112256415B (en) | 2020-10-19 | 2020-10-19 | Micro cloud load balancing task scheduling method based on PSO-GA |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112256415B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115118728B (en) * | 2022-06-21 | 2024-01-19 | 福州大学 | Edge load balancing task scheduling method based on ant colony algorithm |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107609650A (en) * | 2017-09-20 | 2018-01-19 | 河南理工大学 | Based on the particle genetic algorithm optimization method, apparatus and system that time load is balanced |
| CN109491761A (en) * | 2018-11-07 | 2019-03-19 | 中国石油大学(华东) | Cloud computing multiple target method for scheduling task based on EDA-GA hybrid algorithm |
| CN110809275A (en) * | 2019-11-08 | 2020-02-18 | 福州大学 | Micro cloud node placement method based on wireless metropolitan area network |
| CN110851272A (en) * | 2019-10-30 | 2020-02-28 | 内蒙古农业大学 | Cloud task scheduling method based on phagocytic particle swarm genetic hybrid algorithm |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109643247B (en) * | 2016-07-20 | 2023-07-04 | 世界线公司 | Multi-standard adaptive scheduling for market-oriented hybrid cloud infrastructure |
-
2020
- 2020-10-19 CN CN202011118823.6A patent/CN112256415B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107609650A (en) * | 2017-09-20 | 2018-01-19 | 河南理工大学 | Based on the particle genetic algorithm optimization method, apparatus and system that time load is balanced |
| CN109491761A (en) * | 2018-11-07 | 2019-03-19 | 中国石油大学(华东) | Cloud computing multiple target method for scheduling task based on EDA-GA hybrid algorithm |
| CN110851272A (en) * | 2019-10-30 | 2020-02-28 | 内蒙古农业大学 | Cloud task scheduling method based on phagocytic particle swarm genetic hybrid algorithm |
| CN110809275A (en) * | 2019-11-08 | 2020-02-18 | 福州大学 | Micro cloud node placement method based on wireless metropolitan area network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112256415A (en) | 2021-01-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109857546B (en) | Multi-server mobile edge computing unloading method and device based on Lyapunov optimization | |
| Zeng et al. | Joint optimization of task scheduling and image placement in fog computing supported software-defined embedded system | |
| Wang et al. | Towards a load balancing in a three-level cloud computing network | |
| WO2016045515A1 (en) | Cloud task scheduling algorithm based on user satisfaction | |
| CN102662764B (en) | A kind of dynamic cloud computational resource optimizing distribution method based on SMDP | |
| Wang et al. | Adaptive application offloading decision and transmission scheduling for mobile cloud computing | |
| CN111163178B (en) | Game theory-based service deployment and task unloading method in edge computing | |
| Hasan et al. | A krill herd behaviour inspired load balancing of tasks in cloud computing | |
| CN112003660B (en) | Dimension measurement method of resources in network, calculation force scheduling method and storage medium | |
| CN114024970A (en) | Power internet of things work load distribution method based on edge calculation | |
| CN114567895A (en) | Method for realizing intelligent cooperation strategy of MEC server cluster | |
| CN109005211B (en) | Micro-cloud deployment and user task scheduling method in wireless metropolitan area network environment | |
| Hsu et al. | Virtual network mapping algorithm in the cloud infrastructure | |
| Wu et al. | Request dispatching for minimizing service response time in edge cloud systems | |
| CN112256415B (en) | Micro cloud load balancing task scheduling method based on PSO-GA | |
| Wang et al. | Dependency-aware dynamic task scheduling in mobile-edge computing | |
| Xu et al. | Online learning algorithms for offloading augmented reality requests with uncertain demands in MECs | |
| CN114785692A (en) | Virtual power plant aggregation regulation and control communication network flow balancing method and device | |
| Shi et al. | A framework for mobile cloud computing selective service system | |
| CN117407160A (en) | Mixed deployment method for online task and offline task in edge computing scene | |
| Moreira et al. | Task allocation framework for software-defined fog v-RAN | |
| Addya et al. | A hybrid queuing model for virtual machine placement in cloud data center | |
| Jabour et al. | An optimized approach for efficient-power and low-latency fog environment based on the PSO algorithm | |
| Shukla et al. | Energy Efficient Workflow Scheduling Algorithm for Latency-Sensitive Applications using Cloud-Fog Collaboration | |
| Jayasinghe et al. | Task assignment in multiple server farms using preemptive migration and flow control |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |