CN111857976A - Multi-objective optimization calculation migration method based on decomposition - Google Patents
Multi-objective optimization calculation migration method based on decomposition Download PDFInfo
- Publication number
- CN111857976A CN111857976A CN202010790128.8A CN202010790128A CN111857976A CN 111857976 A CN111857976 A CN 111857976A CN 202010790128 A CN202010790128 A CN 202010790128A CN 111857976 A CN111857976 A CN 111857976A
- Authority
- CN
- China
- Prior art keywords
- individual
- representing
- service provider
- population
- cloud service
- 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.)
- Granted
Links
- 238000013508 migration Methods 0.000 title claims abstract description 42
- 230000005012 migration Effects 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000005457 optimization Methods 0.000 title claims abstract description 30
- 238000004364 calculation method Methods 0.000 title claims abstract description 21
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 13
- 230000002068 genetic effect Effects 0.000 claims abstract description 8
- 239000013598 vector Substances 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 10
- 239000011159 matrix material Substances 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 230000004044 response Effects 0.000 claims description 6
- 210000000349 chromosome Anatomy 0.000 description 7
- 108090000623 proteins and genes Proteins 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 229920003087 methylethyl cellulose Polymers 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000035755 proliferation Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
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
- G06F9/44—Arrangements for executing specific programs
- G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
- G06F9/45533—Hypervisors; Virtual machine monitors
- G06F9/45558—Hypervisor-specific management and integration aspects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/51—Discovery or management thereof, e.g. service location protocol [SLP] or web services
-
- 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
- G06F9/44—Arrangements for executing specific programs
- G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
- G06F9/45533—Hypervisors; Virtual machine monitors
- G06F9/45558—Hypervisor-specific management and integration aspects
- G06F2009/4557—Distribution of virtual machine instances; Migration and load balancing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Networks & Wireless Communication (AREA)
- Evolutionary Biology (AREA)
- Artificial Intelligence (AREA)
- Genetics & Genomics (AREA)
- Physiology (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a multi-objective optimization calculation migration method based on decomposition, which belongs to the field of computers and comprises the following steps: step S10, creating a target model based on the satisfaction degree of the terminal user and the income of the edge cloud service provider; s20, performing iterative evolution on the target model by using a genetic algorithm and a multi-target optimization algorithm; and step S30, performing calculation migration by using a multi-criterion decision, a weighting method and the target model after iterative evolution. The invention has the advantages that: the method greatly improves the speed of computing migration while comprehensively considering the satisfaction degree of the terminal user and the income of the edge cloud service provider.
Description
Technical Field
The invention relates to the field of computers, in particular to a multi-objective optimization computation migration method based on decomposition.
Background
In recent years, the technology of the internet of things is greatly developed, but the computing capability of the internet of things equipment is limited, so that a large number of computing-intensive tasks cannot be computed on local internet of things equipment. In order to extend the battery life of the internet of things devices and meet the Computing requirements, Mobile Cloud Computing (MCC) has come into force, and part of the Computing-intensive task request is migrated to the Cloud for processing. While MCC can effectively reduce the computational overhead and power consumption of internet of things devices, MCC also faces some new challenges at the same time. On one hand, with the proliferation of the internet of things equipment, more and more task requests obviously increase the computing burden of the cloud; on the other hand, the geographic distance between the cloud and the internet of things device is very far, so that the request delay and the transmission delay of part of applications become large, the user experience is further influenced, and even the task execution with some strict time constraints fails.
The MEC uses the Computing and storage functions of the network devices close to the end user and the Edge side to assist the user, and migrates the task request to the Edge server closer to the end user for execution, so as to provide the end user with a Computing Service, a storage Service and a communication Service with higher efficiency and lower time delay, and further improve the Quality of Service (QoS) of the end user.
A typical MEC environment includes a public cloud service provider, an edge cloud service provider, and end users, as shown in fig. 2. The public cloud service provider is used for providing software and hardware resources, completing the construction and maintenance of a server cluster, generating a uniform resource pool through resource pooling, and providing resources such as calculation, storage, bandwidth and the like for a terminal user. The edge cloud service provider leases server resources from a public cloud service provider, constructs a service environment according to the service type of the edge cloud service provider and provides services such as image processing, scientific computing, video coding and decoding and the like for the end user. The end user sends a Service request to an edge cloud Service provider which can provide required Service for the end user, and pays the fee according to a Service-Level Agreement (SLA) and a final return result of an edge server.
Ghamkhar et al, in the document "Energy and Performance Management of Green data centers: A Profit knowledge approach.2013", consider the factors of Service Level Agreements (SLAs) currently existing between a data center and its customers, and randomness of workload of the data center, and propose a new optimization-based data center Profit Maximization strategy. Li Meng in the document "research on a cloud computing resource oriented profit optimization model and a task allocation algorithm.2015", the cloud computing profit optimization model provided considers the profit of a service provider and also considers an SLA protocol, estimates the response time of a service request by using a queuing theory, calculates the processing time of the request, and provides a cloud resource profit model oriented PAO-ACO algorithm to solve the model, so that the algorithm can dynamically find an optimal solution, and when the population scale is large, the parallel solution can save the operation time and improve the execution efficiency. 2017, in the cloud computing environment, the optimal pricing of the cloud service and the optimal configuration problem of the multi-server are researched by taking the maximum profit as a target, and a monetariy rewarded Reward model can be provided according to the change of the service price on the premise of conforming to the user demand law in the market environment in a pricing mode, so that a low service quality cost compensation algorithm based on service request response time is designed. Zhankuaihui et al established a Markov game model in ' Markov game-based cloud agent and cloudlet profit optimization ' 2018 ' to analyze a cloud server and cloudlet, and obtain a Nash equilibrium strategy through a reverse iteration algorithm, and finally proved that the system profit can be obviously improved by adopting the Markov game.
The above documents are directed to the optimization of revenue in a cloud computing environment, however, as MECs develop, some researchers have begun to study the maximization of revenue for MEC service providers. For example, in the document "a profit optimization strategy for an MEC server with limited computing resources" by Huangdonyan et al, 2020 ", aiming at the profit optimization problem for an MEC server with limited computing resources, with the maximization of the profit of the MEC server as an optimization target, an algorithm based on a branch-and-bound method is provided to obtain an optimal access strategy and a task execution sequence, and the algorithm can effectively improve the average profit of the MEC server in a heavy load network.
The methods described in the above documents all have problems that the migration speed of calculation is slow and the satisfaction of the end user is neglected. Therefore, how to provide a computing migration method based on multi-objective optimization (MOEA/D) is an urgent problem to be solved, so that the satisfaction of the terminal user and the income of the edge cloud service provider are comprehensively considered, and meanwhile, the computing migration speed is improved.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a multi-objective optimization calculation migration method based on decomposition, so that the satisfaction of a terminal user and the income of an edge cloud service provider are comprehensively considered, and the calculation migration speed is improved.
The invention is realized by the following steps: a computational migration method based on decomposition multi-objective optimization comprises the following steps:
step S10, creating a target model based on the satisfaction degree of the terminal user and the income of the edge cloud service provider;
s20, performing iterative evolution on the target model by using a genetic algorithm and a multi-target optimization algorithm;
and step S30, performing calculation migration by using a multi-criterion decision, a weighting method and the target model after iterative evolution.
Further, the step S10 is specifically:
creating an end user satisfaction model:
wherein ,represents the satisfaction of the end user; smaxRepresents the maximum user satisfaction; t isuIndicating a time that the user desires to complete; t isDDLRepresents the service request's expiration time; t is ti,j(τp,q) Representing the average response time of the jth virtual machine in the ith edge server; tau isp,qRepresents the completion time of the qth service request of the end user p at the edge server; m represents the total number of virtual machines; u. ofi,jRepresenting the task processing rate of the jth virtual machine in the ith edge server; lambda [ alpha ]i,jRepresenting the task arrival rate of the jth virtual machine in the ith edge server; u represents the total number of end users; vpRepresenting the total number of service requests; w is ap,qDenotes τp,qThe number of instructions of (1); b is a boolean function, where B ═ 0 indicates that the qth service request of the end user p is not migrated to the jth virtual machine of the ith edge server, and B ═ 1 indicates that the qth service request of the end user p is migrated to the jth virtual machine of the ith edge server; i. j, M, p, q, U, VpAre all positive integers;
the calculation formula of the total profit of the edge cloud service provider is as follows:
wherein R represents the total revenue of the edge cloud service provider; r (tau)p,q,ti,j(τp,q) Represents the charging of the edge server to handle the qth service request of end user p; p is a radical ofmRepresents a price for each service request;
the cost of the edge cloud service provider is calculated as follows:
wherein C represents the cost of the edge cloud service provider; c. CmRepresents the cost of each service request;
two goals for end user satisfaction and edge cloud service provider revenue are defined as:
s.t.op,q∈{0,1,...,N+1};
wherein op,qThe q-th service request representing end user p assigns a migration policy.
Further, the step S20 specifically includes:
step S21, based on the target model, randomly generating a scale Q in the feasible region omegapGroup G of0:
wherein Represents a population G0Middle (Q)p(ii) individuals of individuals; qpIs a positive integer;
step S22, create QpIndividual weight vector sigmaj:
Wherein j is a positive integer, and j is 1,2p(ii) a k is a positive integer;
step S23 of calculating each of the weight vectors σjEuclidean distance d between each twoi,jBased on the aboveEuclidean distance di,jGenerating a distance matrix d;
selecting Q based on the distance matrix dneiA nearest individual Xi(i=1,2,...,Qp) And forming a neighbor set:
for every nearest individual, letThe weight vector σjMost recent QneiThe individual weight vectors are:
step S24, calculating each individual Xi(i=1,2,...,Qp) The objective function value of (1):
f1(Xi),f2(Xi),...,fk(Xi),;
setting the ideal points of the objective function values as follows:
step S25, setting external population O*And (2) carrying out iterative evolution on each individual, wherein phi is the number of population iterations t and t is a positive integer:
randomly from the neighbor set CiTwo individuals are selected to generate a new individualThe new individual is treatedAdding to population GtIn (1),namely, it is
Updating neighbor set C of individualsi:
Let sigmai,lRepresenting an individual XiNeighbor set C ofiThe weight vector of each element, l 1,2p,
wherein Xi,lRepresents a neighbor set CiEach element of (1);represents the Chebyshev value; f (X)i) Represents XiFitness function values corresponding to the individuals;
updating external population O*:
Judging the external population O*Whether or not a new individual exists inDominant solution, if any, culling the outer population O*Chinese quilt new individualA solution of dominance; if not, the new individual is addedAdding an external population O*Performing the following steps;
step S26, for population G0Selecting, crossing and mutating to generate a new population, judging whether the iteration number t of the population is less than the preset maximum iteration number, if so, entering the step S24; if not, the process proceeds to step S30.
Further, the step S30 is specifically:
the practical value of the satisfaction of the terminal user is set as follows:
the utility value of the revenue of the edge cloud service provider is:
population G0The practical value of each individual is as follows:
the individuals with the greatest practical value are:
wherein SminA minimum value representing end user satisfaction; smaxA maximum value representing end user satisfaction; s (X)i) Representing an individual XiEnd user satisfaction of; rminRepresenting a minimum value of edge cloud service provider revenue; rmaxRepresenting a maximum value of edge cloud service provider revenue; r (X)i) Representing an individual XiEdge cloud service provider revenue of (a); w is a1Weight, w, representing end user satisfaction2Weight, w, representing edge cloud service provider revenue1+w2=1。
The invention has the advantages that:
the method comprises the steps of establishing a target model through the satisfaction of a terminal user and the profit of an edge cloud service provider, then carrying out iterative evolution on the target model by utilizing a genetic algorithm and a multi-objective optimization algorithm, and finally carrying out calculation migration by utilizing a multi-criterion decision method, a weighting method and the target model after iterative evolution, namely finding out an individual with the maximum practical value, so that the satisfaction of the terminal user and the profit of the edge cloud service provider are both maximized, namely, the comprehensive consideration of the satisfaction of the terminal user and the profit of the edge cloud service provider is realized.
Drawings
The invention will be further described with reference to the following examples with reference to the accompanying drawings.
FIG. 1 is a flow chart of a computational migration method based on decomposition multi-objective optimization according to the present invention.
Fig. 2 is a mobile edge computing network architecture diagram of the present invention.
Fig. 3 is a schematic diagram of the migration policy encoding of the present invention.
Figure 4 is a schematic illustration of the present invention minimizing problems.
Fig. 5 is a schematic diagram of a policy coding crossover of the present invention.
Detailed Description
Referring to fig. 1 to 5, an embodiment of a computation migration method based on decomposition multi-objective optimization according to the present invention includes the following steps:
step S10, creating a target model based on the satisfaction degree of the terminal user and the income of the edge cloud service provider;
s20, performing iterative evolution on the target model by using a genetic algorithm and a multi-target optimization algorithm;
and step S30, performing calculation migration by using a multi-criterion decision, a weighting method and the target model after iterative evolution.
Before step S10, it is necessary to build an MEC network architecture, and set the processor performance, the number of virtual machines, and the working power of the mobile device, the edge server, and the cloud server in the MEC network architecture.
The invention adopts integersThe encoding mode is that each gene in each chromosome is an integer, the satisfaction of an end user and the income of an edge cloud service provider are used as optimization targets, the migration strategy of each group is regarded as a chromosome consisting of a plurality of genes, and when N edge servers exist on a platform, the migration strategy is encoded into 0, 1. Migration policy encoding as shown in fig. 3, the number 0 represents that the task does not migrate and the calculation is performed at the local device; numbers 1 to N indicate that the task is to be migrated to a designated edge server for execution; the number N +1 indicates that the task is to be migrated to the cloud server for execution. Before the genetic algorithm starts, the population size Q needs to be determinedpWeight vector σjSize of neighbor set QneiMaximum number of iterations GmaxCross probability pcAnd the like. The ith chromosome in the population can be represented as Xi={o p,q1, | p ═ 1,2, ·, U; q ═ 1,2, ·, V }. And (3) selecting a part of individuals in the population for later crossover and mutation operations, and forming a new population with better fitness based on the fitness evaluation of the individuals.
The step S10 specifically includes:
creating an end user satisfaction model:
wherein ,represents the satisfaction of the end user; smaxRepresents the maximum user satisfaction; t isuIndicating a time that the user desires to complete; t isDDLRepresents the service request's expiration time; t is ti,j(τp,q) Representing the average response time of the jth virtual machine in the ith edge server; tau isp,qRepresents the completion time of the qth service request of the end user p at the edge server; m represents the total number of virtual machines; u. ofi,jRepresenting the task processing rate of the jth virtual machine in the ith edge server; lambda [ alpha ]i,jThe task arrival rate of the jth virtual machine in the ith edge server is shown, namely how many MIPS instructions are reached in the j layer sub-request every second; u represents the total number of end users; vpRepresenting the total number of service requests; w is ap,qDenotes τp,qThe number of instructions of (1); b is a boolean function, where B ═ 0 indicates that the qth service request of the end user p is not migrated to the jth virtual machine of the ith edge server, and B ═ 1 indicates that the qth service request of the end user p is migrated to the jth virtual machine of the ith edge server; i. j, M, p, q, U, VpAre all positive integers;
the calculation formula of the total profit of the edge cloud service provider is as follows:
wherein R represents the total revenue of the edge cloud service provider; r (tau)p,q,ti,j(τp,q) Represents the charging of the edge server to handle the qth service request of end user p; p is a radical ofmRepresents a price for each service request;
the cost of the edge cloud service provider is calculated as follows:
where C represents the cost of the edge cloud service providerThen, the process is carried out; c. CmRepresents the cost of each service request; the revenue of the edge cloud service provider is equal to the total input minus the cost;
two goals for end user satisfaction and edge cloud service provider revenue are defined as:
s.t.op,q∈{0,1,...,N+1};
wherein op,qThe q-th service request representing end user p assigns a migration policy.
The step S20 specifically includes:
step S21, based on the target model, randomly generating a scale Q in the feasible region omegapGroup G of0:
wherein Represents a population G0Middle (Q)p(ii) individuals of individuals; qpIs a positive integer;
step S22, create QpIndividual weight vector sigmaj:
Wherein j is a positive integer, and j is 1,2p(ii) a k is a positive integer;
step S23 of calculating each of the weight vectors σjEuclidean distance d between each twoi,jBased on the Euclidean distance di,jGenerating a distance matrix d;
based onSelecting Q from the distance matrix dneiA nearest individual Xi(i=1,2,...,Qp) And forming a neighbor set:
for every nearest individual, letThe weight vector σjMost recent QneiThe individual weight vectors are:
step S24, calculating each individual Xi(i=1,2,...,Qp) The objective function value of (1):
f1(Xi),f2(Xi),...,fk(Xi),;
setting the ideal points of the objective function values as follows:
step S25, setting external population O*And (2) carrying out iterative evolution on each individual, wherein phi is the number of population iterations t and t is a positive integer:
randomly from the neighbor set CiTwo individuals are selected to generate a new individualThe new individual is treatedAdding to population GtIn, i.e.
For example, randomly selecting two sequence numbers a and b from B (i), using genetic operator Xa and XbGeneration of novel individuals XcThen to XcApplying test problem based repair and improvement heuristic generation
Updating neighbor set C of individualsi:
Let sigmai,lRepresenting an individual XiNeighbor set C ofiThe weight vector of each element, l 1,2p,
wherein Xi,lRepresents a neighbor set CiEach element of (1);represents the Chebyshev value; f (X)i) Represents XiFitness function values corresponding to the individuals;
updating external population O*:
Judging the external population O*Whether or not a new individual exists inDominant solution, if any, culling the outer population O*Chinese quilt new individualA solution of dominance; if not, the new individual is addedAdding an external population O*Performing the following steps;
step S26, for population G0Selecting, crossing and mutating to generate a new population, as shown in fig. 3 and 5, judging whether the population iteration number t is less than a preset maximum iteration number, if so, entering step S24; if not, the process proceeds to step S30.
The step S30 specifically includes:
the practical value of the satisfaction of the terminal user is set as follows:
the utility value of the revenue of the edge cloud service provider is:
population G0The practical value of each individual is as follows:
the individuals with the greatest practical value are:
wherein SminA minimum value representing end user satisfaction; smaxA maximum value representing end user satisfaction; s (X)i) Representing an individual XiEnd user satisfaction of; rminRepresenting a minimum value of edge cloud service provider revenue; rmaxRepresenting a maximum value of edge cloud service provider revenue; r (X)i) Representing an individual XiEdge cloud service provider revenue of (a); w is a1Weight, w, representing end user satisfaction2Weight, w, representing edge cloud service provider revenue1+w2=1。
In the second embodiment of the computational migration method based on the multi-objective optimization of the decomposition, 4 edge servers and 1 cloud server are assumed to exist in the MEC network architecture, and the condition that part of task requests of end users are executed locally is considered.
Assume a set of mobile users U ═ U1,1,u1,2,…,u1,q,u2,1,u2,2,…,u2,q,…,up,1,up,2,…,up,q}, wherein up,qA qth application representing a pth user; op,qDenoted as the migration policy assigned to the qth request of the p-th user, op,q1,2, …, N denotes that application migration is proceeding to the edge server, op,qN +1 represents the migration of the application to the cloud for execution; tau isp,qRepresents the qth service request of user p; t (τ)p,q) The completion time of the qth service request denoted as user p; r (tau)p,q,t(τp,q) Charge expressed as the qth service request for user p; n represents a set of the number of edge servers in the system; m represents the number of virtual machines of the edge server in the system; w is ap,qIs denoted by τp,qThe number of instructions of (1); p is a radical ofmA price expressed as a number of pieces per instruction; c. CmCost expressed as number of instructions per instruction; t isuExpressed as a user-desired completion time; t isDDLExpressed as the deadline for the service request; sτp,qIs denoted by τp,qUser satisfaction of (2); u is expressed as a task processing rate; λ is expressed as the task arrival rate; smaxExpressed as maximum user satisfaction.
The case where task requests are assumed not to migrate is labeled up,qThe case of migration to the 1 st to 4 th edge server is marked u 0p,q=1,up,q=2,up,q=3 and u p,q4, the case of migration to the cloud is marked u p,q5. As shown in the migration policy encoding of FIG. 3, each migration case flagA large number of genes in a chromosome are converted into a chromosome, and the operations of crossing and compiling are carried out in the subsequent evolution process.
From the queuing theory, a processing time model, τ, is calculatedp,qThe average response time in the jth virtual machine in the ith server is:
wherein ,ui,jThe task processing rate of the jth virtual machine in the ith edge server is obtained; lambda [ alpha ]i,jThe task arrival rate of the jth virtual machine in the ith edge server, i.e. how many MIPS instructions arrive per second on the j-tier sub-request, is calculated by the following formula:
b is a boolean function, where B ═ 0 indicates that the qth service sub-request of the user p is not migrated to the jth virtual machine of the ith server; b ═ 1 indicates that the qth service sub-request of the user p is migrated to the jth virtual machine of the ith server.
Therefore, the completion time of the qth service request of user p in the ith server is:
from the processing time calculation model of the above formulaDescribing the q-th service request of the user p, and obtaining the corresponding satisfaction degree in different service request completion time. The calculation formula is as follows:
with R (tau)p,q,t(τp,q) Indicating that the edge server handles the qth service request of user pFor a fee, then the total revenue R for the edge server provider is:
and R (tau)p,q,t(τp,q) The formula for calculation) is:
the cost of the edge server is denoted by C, and the revenue of the edge cloud service provider is derived from the revenue of the edge server minus the cost, which is simply calculated as:
in summary, the overall objective of the computational migration studied by the present invention is to maximize user satisfaction and profit for edge cloud service providers while meeting the task requirements of mobile users, wherein the multi-objective optimization problem can be defined as:
s.t.op,q∈{0,1,2,3,4,5};
each chromosome contains the migration condition of the task request of each user, the migration coding strategy is substituted into the relational expression of the user satisfaction and the edge cloud service provider profit of the above expression to obtain an adaptive value, and evolution operation is performed.
When the evolution algebra reaches the maximum, the population PtmaxPresence population size QpEach individual represents the calculation migration strategy of the optimal solution, the gene value of the chromosome is brought into the established model, the target value of the optimal solution can be obtained, but the situation that part of individuals are different in the optimal solution can occur, and at the moment, more individuals are neededSelecting optimal solution from criterion decision and simple weighting method, XiAnd (3) representing the ith individual in the optimal solution population, wherein the practical value of the ith individual is defined as:
LBmax and LBminExpressed as the target maximum and minimum values of user satisfaction in the optimal solution population.
Tmax and TminExpressed as the maximum and minimum of edge server revenue in the optimal solution population.
The satisfaction of the practical value of each individual in the optimal solution population is calculated by a weight w, wherein w1+w2=1:
After the practical value of each individual is obtained, the individual with the maximum practical value needs to be selected from the population, and the individual with the maximum practical value is the obtained optimal solution individual:
in summary, the invention has the advantages that:
the method comprises the steps of establishing a target model through the satisfaction of a terminal user and the profit of an edge cloud service provider, then carrying out iterative evolution on the target model by utilizing a genetic algorithm and a multi-objective optimization algorithm, and finally carrying out calculation migration by utilizing a multi-criterion decision method, a weighting method and the target model after iterative evolution, namely finding out an individual with the maximum practical value, so that the satisfaction of the terminal user and the profit of the edge cloud service provider are both maximized, namely, the comprehensive consideration of the satisfaction of the terminal user and the profit of the edge cloud service provider is realized.
Although specific embodiments of the invention have been described above, it will be understood by those skilled in the art that the specific embodiments described are illustrative only and are not limiting upon the scope of the invention, and that equivalent modifications and variations can be made by those skilled in the art without departing from the spirit of the invention, which is to be limited only by the appended claims.
Claims (4)
1. A computational migration method based on decomposition multi-objective optimization is characterized in that: the method comprises the following steps:
step S10, creating a target model based on the satisfaction degree of the terminal user and the income of the edge cloud service provider;
s20, performing iterative evolution on the target model by using a genetic algorithm and a multi-target optimization algorithm;
and step S30, performing calculation migration by using a multi-criterion decision, a weighting method and the target model after iterative evolution.
2. The method of computational migration based on multi-objective optimization of decomposition as claimed in claim 1, wherein: the step S10 specifically includes:
creating an end user satisfaction model:
wherein ,represents the satisfaction of the end user; smaxRepresents the maximum user satisfaction; t isuIndicating a time that the user desires to complete; t isDDLRepresents the service request's expiration time; t is ti,j(τp,q) Representing the average response time of the jth virtual machine in the ith edge server; tau isp,qRepresents the completion time of the qth service request of the end user p at the edge server; m represents the total number of virtual machines; u. ofi,jRepresenting the task processing rate of the jth virtual machine in the ith edge server; lambda [ alpha ]i,jRepresenting the task arrival rate of the jth virtual machine in the ith edge server; u represents the total number of end users; vpRepresenting the total number of service requests; w is ap,qDenotes τp,qThe number of instructions of (1); b is a boolean function, where B ═ 0 indicates that the qth service request of the end user p is not migrated to the jth virtual machine of the ith edge server, and B ═ 1 indicates that the qth service request of the end user p is migrated to the jth virtual machine of the ith edge server; i. j, M, p, q, U, VpAre all positive integers;
the calculation formula of the total profit of the edge cloud service provider is as follows:
wherein R represents the total revenue of the edge cloud service provider; r (tau)p,q,ti,j(τp,q) Represents the charging of the edge server to handle the qth service request of end user p; p is a radical ofmRepresents a price for each service request;
the cost of the edge cloud service provider is calculated as follows:
wherein C represents the cost of the edge cloud service provider; c. CmRepresents the cost of each service request;
two goals for end user satisfaction and edge cloud service provider revenue are defined as:
s.t.op,q∈{0,1,...,N+1};
wherein op,qThe q-th service request representing end user p assigns a migration policy.
3. The method of computational migration based on multi-objective optimization of decomposition as claimed in claim 1, wherein: the step S20 specifically includes:
step S21, based on the target model, randomly generating a scale Q in the feasible region omegapGroup G of0:
wherein Represents a population G0Middle (Q)p(ii) individuals of individuals; qpIs a positive integer;
step S22, create QpIndividual weight vector sigmaj:
Wherein j is a positive integer, and j is 1,2p(ii) a k is a positive integer;
step S23 of calculating each of the weight vectors σjEuclidean distance d between each twoi,jBased on the Euclidean distance di,jGenerating a distance matrix d;
selecting Q based on the distance matrix dneiA nearest individual Xi(i=1,2,...,Qp) And forming a neighbor set:
for every nearest individual, letThe weight vector σjMost recent QneiThe individual weight vectors are:
step S24, calculating each individual Xi(i=1,2,...,Qp) The objective function value of (1):
f1(Xi),f2(Xi),...,fk(Xi),;
setting the ideal points of the objective function values as follows:
step S25, setting external population O*And (2) carrying out iterative evolution on each individual, wherein phi is the number of population iterations t and t is a positive integer:
randomly from the neighborLiving collection CiTwo individuals are selected to generate a new individualThe new individual is treatedAdding to population GtIn, i.e.
Updating neighbor set C of individualsi:
Let sigmai,lRepresenting an individual XiNeighbor set C ofiThe weight vector of each element, l 1,2p,
If g iste(Xi|σi,l,y*)≤gte(Xi,l|σi,l,y*) Then Xi,l=Xi;
wherein Xi,lRepresents a neighbor set CiEach element of (1);represents the Chebyshev value; f (X)i) Represents XiFitness function values corresponding to the individuals;
updating external population O*:
Judging the external population O*Whether or not a new individual exists inDominant solutions, if any, culling the outer populationO*Chinese quilt new individualA solution of dominance; if not, the new individual is addedAdding an external population O*Performing the following steps;
step S26, for population G0Selecting, crossing and mutating to generate a new population, judging whether the iteration number t of the population is less than the preset maximum iteration number, if so, entering the step S24; if not, the process proceeds to step S30.
4. The method of computational migration based on multi-objective optimization of decomposition as claimed in claim 3, wherein: the step S30 specifically includes:
the practical value of the satisfaction of the terminal user is set as follows:
the utility value of the revenue of the edge cloud service provider is:
population G0The practical value of each individual is as follows:
the individuals with the greatest practical value are:
wherein SminA minimum value representing end user satisfaction; smaxMaximum degree of satisfaction of end userA large value; s (X)i) Representing an individual XiEnd user satisfaction of; rminRepresenting a minimum value of edge cloud service provider revenue; rmaxRepresenting a maximum value of edge cloud service provider revenue; r (X)i) Representing an individual XiEdge cloud service provider revenue of (a); w is a1Weight, w, representing end user satisfaction2Weight, w, representing edge cloud service provider revenue1+w2=1。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010790128.8A CN111857976B (en) | 2020-08-07 | 2020-08-07 | Multi-objective optimization computing migration method based on decomposition |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010790128.8A CN111857976B (en) | 2020-08-07 | 2020-08-07 | Multi-objective optimization computing migration method based on decomposition |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111857976A true CN111857976A (en) | 2020-10-30 |
CN111857976B CN111857976B (en) | 2023-06-02 |
Family
ID=72971189
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010790128.8A Active CN111857976B (en) | 2020-08-07 | 2020-08-07 | Multi-objective optimization computing migration method based on decomposition |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111857976B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113094973A (en) * | 2021-03-18 | 2021-07-09 | 西北工业大学 | Civil aircraft demand optimization method based on multi-objective optimization algorithm |
CN113112143A (en) * | 2021-04-08 | 2021-07-13 | 杭州电子科技大学 | Multi-level manufacturer collaborative production task allocation method and system under cloud manufacturing environment |
CN113139639A (en) * | 2021-05-17 | 2021-07-20 | 华侨大学 | MOMBI-based smart city application-oriented multi-target calculation migration method and device |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100153299A1 (en) * | 2008-12-16 | 2010-06-17 | Sean Coleman Keenan | Methods and systems for generating transition probability matrices through an optimization framework |
CN102984137A (en) * | 2012-11-14 | 2013-03-20 | 江苏南开之星软件技术有限公司 | Multi-target server scheduling method based on multi-target genetic algorithm |
CN107734558A (en) * | 2017-10-26 | 2018-02-23 | 北京邮电大学 | A kind of control of mobile edge calculations and resource regulating method based on multiserver |
CN108984307A (en) * | 2018-07-19 | 2018-12-11 | 中国联合网络通信集团有限公司 | Calculating task moving method and calculating task transference apparatus |
CN109413676A (en) * | 2018-12-11 | 2019-03-01 | 西北大学 | Combine the edge calculations moving method of lower uplink in a kind of ultra dense heterogeneous network |
CN109710374A (en) * | 2018-12-05 | 2019-05-03 | 重庆邮电大学 | The VM migration strategy of task unloading expense is minimized under mobile edge calculations environment |
CN110069319A (en) * | 2019-04-15 | 2019-07-30 | 南京信息工程大学 | A kind of multiple target dispatching method of virtual machine and system towards cloudlet resource management |
CN110839075A (en) * | 2019-11-08 | 2020-02-25 | 重庆大学 | Service migration method based on particle swarm in edge computing environment |
-
2020
- 2020-08-07 CN CN202010790128.8A patent/CN111857976B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100153299A1 (en) * | 2008-12-16 | 2010-06-17 | Sean Coleman Keenan | Methods and systems for generating transition probability matrices through an optimization framework |
CN102984137A (en) * | 2012-11-14 | 2013-03-20 | 江苏南开之星软件技术有限公司 | Multi-target server scheduling method based on multi-target genetic algorithm |
CN107734558A (en) * | 2017-10-26 | 2018-02-23 | 北京邮电大学 | A kind of control of mobile edge calculations and resource regulating method based on multiserver |
CN108984307A (en) * | 2018-07-19 | 2018-12-11 | 中国联合网络通信集团有限公司 | Calculating task moving method and calculating task transference apparatus |
CN109710374A (en) * | 2018-12-05 | 2019-05-03 | 重庆邮电大学 | The VM migration strategy of task unloading expense is minimized under mobile edge calculations environment |
CN109413676A (en) * | 2018-12-11 | 2019-03-01 | 西北大学 | Combine the edge calculations moving method of lower uplink in a kind of ultra dense heterogeneous network |
CN110069319A (en) * | 2019-04-15 | 2019-07-30 | 南京信息工程大学 | A kind of multiple target dispatching method of virtual machine and system towards cloudlet resource management |
CN110839075A (en) * | 2019-11-08 | 2020-02-25 | 重庆大学 | Service migration method based on particle swarm in edge computing environment |
Non-Patent Citations (6)
Title |
---|
伏舒存;付章杰;邢国稳;刘庆祥;许小龙;: "移动边缘环境下面向工作流管理的计算迁移方法", 计算机应用 * |
廖大强;: "面向多目标的云计算资源调度算法", 计算机系统应用 * |
朱友康;乐光学;杨晓慧;刘建生;: "边缘计算迁移研究综述", 电信科学 * |
郭棉;李绮琦;: "面向延迟敏感型物联网应用的计算迁移策略", 计算机应用 * |
陈臣;马晓亭;: "云计算环境下数字图书馆云服务平台与云服务模式研究", 情报资料工作 * |
颉斌;杨扬;旷毅;: "基于多目标遗传算法的云服务部署优化方法", 华中科技大学学报(自然科学版) * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113094973A (en) * | 2021-03-18 | 2021-07-09 | 西北工业大学 | Civil aircraft demand optimization method based on multi-objective optimization algorithm |
CN113112143A (en) * | 2021-04-08 | 2021-07-13 | 杭州电子科技大学 | Multi-level manufacturer collaborative production task allocation method and system under cloud manufacturing environment |
CN113112143B (en) * | 2021-04-08 | 2024-04-23 | 杭州电子科技大学 | Multi-level manufacturer collaborative production task allocation method and system in cloud manufacturing environment |
CN113139639A (en) * | 2021-05-17 | 2021-07-20 | 华侨大学 | MOMBI-based smart city application-oriented multi-target calculation migration method and device |
CN113139639B (en) * | 2021-05-17 | 2023-05-30 | 华侨大学 | MOMBI-oriented smart city application multi-target computing migration method and device |
Also Published As
Publication number | Publication date |
---|---|
CN111857976B (en) | 2023-06-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Tong et al. | Adaptive computation offloading and resource allocation strategy in a mobile edge computing environment | |
Manikandan et al. | Bee optimization based random double adaptive whale optimization model for task scheduling in cloud computing environment | |
CN111857976B (en) | Multi-objective optimization computing migration method based on decomposition | |
Mapetu et al. | A dynamic VM consolidation approach based on load balancing using Pearson correlation in cloud computing | |
Shu et al. | A novel energy-efficient resource allocation algorithm based on immune clonal optimization for green cloud computing | |
Xu et al. | Multiobjective computation offloading for workflow management in cloudlet‐based mobile cloud using NSGA‐II | |
CN108416465B (en) | Workflow optimization method in mobile cloud environment | |
Supreeth et al. | Hybrid genetic algorithm and modified-particle swarm optimization algorithm (GA-MPSO) for predicting scheduling virtual machines in educational cloud platforms | |
CN113485826B (en) | Load balancing method and system for edge server | |
CN111813506A (en) | Resource sensing calculation migration method, device and medium based on particle swarm algorithm | |
CN111836284B (en) | Energy consumption optimization calculation and unloading method and system based on mobile edge calculation | |
Tong et al. | DDQN-TS: A novel bi-objective intelligent scheduling algorithm in the cloud environment | |
Wu et al. | Request dispatching for minimizing service response time in edge cloud systems | |
He et al. | Intent-based resource matching strategy in cloud | |
Li et al. | Computation offloading and service allocation in mobile edge computing | |
CN113992677A (en) | MEC calculation unloading method for delay and energy consumption joint optimization | |
CN117032902A (en) | Cloud task scheduling method for improving discrete particle swarm algorithm based on load | |
CN112835684A (en) | Virtual machine deployment method facing mobile edge calculation | |
Li et al. | DQN-enabled content caching and quantum ant colony-based computation offloading in MEC | |
CN113139639B (en) | MOMBI-oriented smart city application multi-target computing migration method and device | |
Huang et al. | Computation offloading for multimedia workflows with deadline constraints in cloudlet-based mobile cloud | |
Yang et al. | Multi-objective deep reinforcement learning for mobile edge computing | |
Talha et al. | A chaos opposition‐based dwarf mongoose approach for workflow scheduling in cloud | |
CN111708620A (en) | Task unloading method with charging mechanism | |
Yuan et al. | A DRL-Based Container Placement Scheme with Auxiliary Tasks. |
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 |