CN110489176B - Multi-access edge computing task unloading method based on boxing problem - Google Patents

Multi-access edge computing task unloading method based on boxing problem Download PDF

Info

Publication number
CN110489176B
CN110489176B CN201910795561.8A CN201910795561A CN110489176B CN 110489176 B CN110489176 B CN 110489176B CN 201910795561 A CN201910795561 A CN 201910795561A CN 110489176 B CN110489176 B CN 110489176B
Authority
CN
China
Prior art keywords
task
edge server
queue
edge
computing
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
Application number
CN201910795561.8A
Other languages
Chinese (zh)
Other versions
CN110489176A (en
Inventor
刘昊霖
龚夏臻
裴廷睿
李哲涛
朱江
王仕果
崔荣埈
关屋大雄
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xiangtan University
Original Assignee
Xiangtan University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Xiangtan University filed Critical Xiangtan University
Priority to CN201910795561.8A priority Critical patent/CN110489176B/en
Publication of CN110489176A publication Critical patent/CN110489176A/en
Application granted granted Critical
Publication of CN110489176B publication Critical patent/CN110489176B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G06F9/445—Program loading or initiating
    • G06F9/44594—Unloading

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer And Data Communications (AREA)

Abstract

The invention provides a multi-access edge computing task unloading method based on a boxing problem. The method comprises the following steps: firstly, calculating the capacity of each edge server and the ratio of the size of input data of each terminal task to required calculation resources; then two queues are formed from large to small according to the capacity and the task ratio; and finally, sequentially taking out the tasks in the task queue, configuring the tasks on the edge server with the largest capacity and the residual computing resources in the container queue, and repeating the operation until the task queue is empty. The invention can be applied to the task processing process of the multi-terminal multi-task multi-access edge computing network, can definitely make a proper task unloading scheme, meets the task time delay requirement, simultaneously minimizes the computing energy consumption and saves the cost.

Description

Multi-access edge computing task unloading method based on boxing problem
Technical Field
The invention relates to the field of multi-access edge computing, in particular to a computing resource allocation problem of a plurality of edge servers in a multi-access edge computing network.
Background
The Internet of Things (IoT) technology aims to connect a real object with the Internet according to a communication protocol agreed by the Internet of Things by using a radio frequency identification technology, a wireless data communication technology, a global positioning system and the like, so as to realize information exchange and achieve the purposes of intelligently identifying, positioning, tracking, monitoring and managing Internet resources. With the increase of the number of devices, the enhancement of computing performance and the increasing demand of users, edge computing has become an important expansion form of the internet of things. Multi-access Edge Computing (MEC) is an emerging concept in network technology, and can make Computing resources at the Edge of a network closer to users, thereby optimizing resource allocation, increasing access speed, and improving traditional cloud Computing services. Roughly speaking, a user device may connect to one or more edge servers, offloading computationally intensive tasks that run too slowly or are too energy intensive.
In recent two years, the edge computing technology is rapidly developed in the application of the internet of things, and more equipment terminals and more complex computing tasks appear in an edge computing mode. In a traditional terminal execution mode, if a calculation-intensive task is completely executed locally by a terminal, the problems that the terminal energy consumption is too large, the task running time is too long, the running requirements of complex application programs cannot be met and the like can be caused. On the contrary, if the data of the terminal is sent to the cloud center for processing, although the accurate calculation requirement of the complex application program can be met, the load of the cloud center and the network link is greatly increased, and the communication energy consumption of the terminal is also overlarge. Moreover, the longer distance between the terminal and the cloud end causes higher transmission delay, so that the low delay requirement of some terminal applications cannot be met.
In summary, in order to reasonably allocate the computing resources of the edge server, when multiple terminal devices process the computing task, we need to make an offloading (offloading) decision. In the decision making process, two aspects of time delay and energy consumption required by the calculation task are considered, and the aim is to minimize the energy consumption of the edge server and reduce the calculation cost while meeting the time delay requirement of each task. Modeling a task unloading problem into an NP (network performance) difficult problem, adopting a heuristic method for solving quickly, regarding the problem as a boxing problem, and achieving the purpose of minimizing the total energy consumption of a network by minimizing the number of edge servers needing to be started. The method can solve the unloading decision of the terminal task and the resource allocation scheme of the edge server, and minimize the energy consumption of task execution under the condition of meeting the time delay requirement of the terminal task.
Disclosure of Invention
The invention provides a multi-access edge computing task unloading method based on a boxing problem, which is mainly applied to the field of multi-access edge computing and has the main advantages that an unloading decision of each terminal task and an edge server resource allocation scheme can be reasonably arranged, and the energy consumption for task execution is minimized.
1. A Multi-access Edge Computing (MEC) task offloading method based on a binning problem, wherein a user equipment considers a task offloading problem of an MEC as a binning problem and completes a task offloading decision through a heuristic method, the method comprising at least the following steps:
step 1, arranging an MEC network scene, wherein a plurality of user equipment N is {1,2,3 … N }, a plurality of edge computing servers M is {1,2,3 … M }, a plurality of communication channels K are arranged between the user equipment and the servers {1,2,3 … K }, each user equipment has only one computing task in a time slot ts, and the task needs to be executed locally or unloaded to the edge servers for execution;
step 2, calculating the capacity of each edge server, and sorting all the edge servers according to the capacity of the edge servers from large to small to form a queue X ═ X1,X2,X3…Xm}；
Step 3, calculating CP of task of each user terminali(Cost Performance) value, the task of the user terminal is according to CPiThe values are sorted from large to small to form a queue Y ═ Y1,Y2,Y3…Ym}；
Step 4, the task of the user terminal i is expressed as
Figure BDA0002180837630000021
Wherein DiSize of input data for a task, CiIs a resource required to complete a computing task,
Figure BDA0002180837630000022
representing a maximum latency limit for the task;
step 5, calculating the running energy consumption of the tasks at the user terminal and the edge server, taking the total network energy consumption in the minimized time slot ts as a target, taking the task unloading decision of the user terminal as a variable, and constructing an optimization model;
step 6, regarding the task unloading decision problem in the step 5 as a boxing problem, regarding the user terminal and the edge server as task containers, regarding the tasks of the user terminal as articles, minimizing the number of the edge servers started in the network by a heuristic method, and solving the task unloading decision;
step 7, taking out the tasks in the queue Y in sequence and configuring the tasks to the edge server in the queue X, and meeting the CP (content protection protocol) of the task time delay constraintiUnloading the task with the highest value to the edge server with the largest capacity;
and 8, deleting the successfully distributed tasks from the queue Y, deleting the edge server reaching the maximum capacity from the queue X, and repeating the step 7 until the queue Y is empty.
The invention has the following advantages:
1. the time delay requirement of each task can be met, and the service experience of the user is not influenced.
2. And a task unloading decision is rapidly calculated by a heuristic method, and the waiting time delay of the task unloading decision and the server resource allocation is reduced.
3. The number of edge servers operating in the network can be minimized, energy consumption is reduced, and cost is saved.
Drawings
FIG. 1 is a summary flow chart of the present invention;
fig. 2 is a flow chart of the algorithm of the present invention.
Detailed Description
The present invention is described in further detail below with reference to the attached drawing figures.
Step one, arranging a multi-access edge computing network scene
1) In the current multi-access edge computing network, 5 edge servers are set and are in a dormant state, 10 users in the multi-access edge computing network send task requests in the current time slot, and the total computing resources of the edge servers are enough to meet the computing requirements of all tasks; calculating the capacity of 5 edge servers in the network, and sequencing the edge servers from large to small according to the capacity to form a queueX＝{X1,X2,X3,X4,X5}；
2) The task of user terminal i is represented as
Figure BDA0002180837630000031
Wherein DiSize of input data for a task, CiIs a resource required to complete a computing task,
Figure BDA0002180837630000032
indicating the maximum delay limit of the task, e.g. the task at the client with number 1 is indicated as
Figure BDA0002180837630000033
3) Calculating CP (cost Performance) values of 10 tasks in the current time slot, wherein the calculation mode of the CP values of the tasks is as follows:
CP＝D/C
d is the size of the task input data, and C is the resource needed to complete the computing task
Sorting the tasks according to the size of the CP value from large to small to form a queue Y ═ Y1,Y2,Y3,Y4,Y5,Y6,Y7,Y8,Y9,Y10}；
Step two, establishing an optimization model
Calculating the running energy consumption of the tasks at the user terminal and the edge server, taking the total network energy consumption in the minimized time slot ts as a target, taking the task unloading decision of the user terminal as a variable, and constructing an optimization model: 1) definition ai,j,kIndicates the offloading decision of the user equipment i, ai,j,k＝{0,1}，ai,j,kWhen j is 1, the value of k is meaningless, and the capacity of the default edge server is larger than the amount of computing resources required by each task;
2) if the task is completed at the user terminal, the required time is
Figure BDA0002180837630000034
The energy consumption is
Figure BDA0002180837630000035
Wherein k is a constant and mainly depends on a chip architecture, and f is the computing power of the user terminal;
3) if the task is unloaded to the edge server, the data needs to be transmitted through the mobile network, and the task completion time is
Figure BDA0002180837630000036
Where fx is edge server computing power, B is network bandwidth, and energy consumption includes data transmission energy consumption and edge server operating energy consumption, expressed as
Figure BDA0002180837630000037
Wherein P ismFor power at the edge server run, PiIs the transmission power of the user, wherein the operation energy consumption of the edge server is higher than the operation energy consumption of the task at the user terminal;
4) convert the problem into the question
Figure BDA0002180837630000038
Subject to the following constraints:
Figure BDA0002180837630000039
ai,j,kis the decision variable to be solved.
Step three, algorithm implementation
The optimization model provided in the step two is an NP difficult problem, in order to obtain the solution of the problem within the polynomial time, the problem is regarded as a boxing problem, the task of the user terminal is distributed to the edge server or the user terminal to operate according to the task delay constraint requirement, the purpose of minimizing the network energy consumption is achieved by minimizing the number of the edge servers required to be started after the task is completed, and the heuristic algorithm is specifically realized as follows:
1) will Y1Planning to offload to X1In step (c), calculate completion Y1Required time delay
Figure BDA0002180837630000041
If it is not
Figure BDA0002180837630000042
This unloading is performed, at which point a1,2,k1 is ═ 1; otherwise Y1Is calculated on its user terminal, at this time a1,1,k1 and remove Y from the queue1；
2) Will Y2Planning to offload to X1In step (c), calculate completion Y2Required time delay
Figure BDA0002180837630000043
If it is not
Figure BDA0002180837630000044
This unloading is performed, at which point a2,2,k1 is ═ 1; otherwise Y2Is calculated on its user terminal, at this time a2,1,k1 and remove Y from the queue2；
3) Repeating the above operation until X1The residual capacity can not meet the requirement of the task planned to be unloaded currently, and X is deleted from the queue at the moment1And setting the edge server to be unloaded to X2And processing the remaining tasks in Y according to 1);
4) this is repeated for queue X, Y until queue Y is empty, at which point the offloading of all tasks in the current slot is complete.

Claims (4)

1. A multi-access edge computing task offloading method based on a binning problem is characterized in that a user equipment considers the task offloading problem of an MEC as the binning problem and completes task offloading decision through a heuristic method, wherein the method at least comprises the following steps:
step 1, arranging an MEC network scene, wherein a plurality of user equipment N is {1,2,3 … N }, a plurality of edge computing servers M is {1,2,3 … M }, a plurality of communication channels K are arranged between the user equipment and the servers {1,2,3 … K }, each user equipment has only one computing task in a time slot ts, and the task needs to be executed locally or unloaded to the edge servers for execution;
step 2, calculating the capacity of each edge server, and sorting all the edge servers according to the capacity of the edge servers from large to small to form a queue X ═ X1，X2，X3…Xm}；
Step 3, the task of the user terminal i is expressed as (D)i，Ci，
Figure FDA0003553650950000011
) Wherein D isiSize of input data for a task, CiIs a resource required to complete a computing task,
Figure FDA0003553650950000012
representing a maximum latency limit for the task;
step 4, calculating CP of task of each user terminaliValue, CPiThe values are calculated as follows:
CPi＝Di/Ci，
pressing the task of the user terminal to CPiSorting the values from large to small to form a queue Y ═ Y1，Y2，Y3…Ym}；
Step 5, calculating the running energy consumption of the tasks at the user terminal and the edge server, taking the total network energy consumption in the minimized time slot ts as a target, taking the task unloading decision of the user terminal as a variable, and constructing an optimization model;
step 6, regarding the task unloading decision problem in the step 5 as a boxing problem, regarding the user terminal and the edge server as task containers, regarding the tasks of the user terminal as articles, minimizing the number of the edge servers started in the network by a heuristic method, and solving the task unloading decision;
step 7, in sequenceTaking out the task in the queue Y and configuring the task to the edge server in the queue X, and meeting the CP of the task time delay constraintiUnloading the task with the highest value to the edge server with the largest capacity;
and 8, deleting the successfully distributed tasks from the queue Y, deleting the edge server reaching the maximum capacity from the queue X, and repeating the step 7 until the queue Y is empty.
2. The bin packing problem based multiple access edge computing task offload method of claim 1, wherein the communication channels between the user terminal and the edge server are orthogonal when performing the computing offload, and do not affect each other during the communication process, and the network bandwidth of each communication channel is the same.
3. The bin packing problem-based multi-access edge computing task offloading method according to claim 1, wherein the optimization model building method of step 5 comprises at least the following steps:
1) definition ai，j，kTo indicate the offloading decision of the user equipment i, ai，j，k＝{0，1}，ai，j，kWhen j is 1, the value of k is meaningless, and the capacity of the default edge server is larger than the amount of computing resources required by each task;
2) if the task is completed at the user terminal, the required time is
Figure FDA0003553650950000021
The energy consumption is
Figure FDA0003553650950000022
Wherein k is a constant, dependent on the chip architecture, and f is the user terminal computing power;
3) if the task is offloaded to an edge server, data needs to be transmitted over the mobile network, whichThe time of task completion is
Figure FDA0003553650950000023
Figure FDA0003553650950000024
Wherein f ismComputing power for the edge server, B network bandwidth, energy consumption including data transmission energy consumption and edge server operation energy consumption, expressed as
Figure FDA0003553650950000025
Wherein P ismFor power at the edge server run, PiIs the transmission power of the user, wherein the operation energy consumption of the edge server is higher than the operation energy consumption of the task at the user terminal;
4) convert the problem into the question
Figure FDA0003553650950000026
Subject to the following constraints:
Figure FDA0003553650950000027
i∈N，ai，j，k＝{0，1}，
ai，j，kis the decision variable to be solved.
4. The method for offloading task of multiple access edge computing based on binning problem as claimed in claim 1, wherein the optimization model proposed in the step 5 is an NP-hard problem, the problem is regarded as a binning problem in order to solve the problem within polynomial time, the task of the ue is allocated to the edge server or the ue itself is run according to the task delay constraint requirement, and the purpose of minimizing network energy consumption is achieved by minimizing the number of edge servers required to be started for completing the task, the heuristic method at least comprises the following steps:
1) firstly, the task with the highest CP value in the queue Y is processed, and the edge server at the head of the queue X is set to doFor the target server, first determine if the task can be offloaded to the target edge server, if so
Figure FDA0003553650950000028
Unloading is carried out, decision variables are set, otherwise, the task runs on the user terminal, the decision variables are set, the effective capacity of the target server is updated, and the task is deleted from the queue Y after the distribution is finished;
2) repeating the step 1 until the effective load of the target edge server can not continuously accommodate more tasks, deleting the target server from the queue X at the moment, selecting the edge server at the head of the queue as the target edge server in the step 1 again, and repeating the step 1;
3) and repeating the step 1 and the step 2 until the queue Y is empty, and finishing the task unloading decision of the time slot ts at the moment.
CN201910795561.8A 2019-08-27 2019-08-27 Multi-access edge computing task unloading method based on boxing problem Active CN110489176B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910795561.8A CN110489176B (en) 2019-08-27 2019-08-27 Multi-access edge computing task unloading method based on boxing problem

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910795561.8A CN110489176B (en) 2019-08-27 2019-08-27 Multi-access edge computing task unloading method based on boxing problem

Publications (2)

Publication Number Publication Date
CN110489176A CN110489176A (en) 2019-11-22
CN110489176B true CN110489176B (en) 2022-05-10

Family

ID=68554529

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910795561.8A Active CN110489176B (en) 2019-08-27 2019-08-27 Multi-access edge computing task unloading method based on boxing problem

Country Status (1)

Country Link
CN (1) CN110489176B (en)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111049917B (en) * 2019-12-19 2020-12-29 电子科技大学 Mobile-aware multi-user unloading optimization method based on mobile edge calculation
CN111641891B (en) * 2020-04-16 2021-11-23 北京邮电大学 Task peer-to-peer unloading method and device in multi-access edge computing system
CN111970323A (en) * 2020-07-10 2020-11-20 北京大学 Time delay optimization method and device for cloud-edge multi-layer cooperation in edge computing network
CN112486674A (en) * 2020-11-18 2021-03-12 北京邮电大学 Artificial intelligence service-oriented resource allocation method and device and electronic equipment
CN112612553B (en) * 2021-01-06 2023-09-26 重庆邮电大学 Edge computing task unloading method based on container technology
CN113015217B (en) * 2021-02-07 2022-05-20 重庆邮电大学 Edge cloud cooperation low-cost online multifunctional business computing unloading method
CN114567587B (en) * 2022-03-14 2024-02-20 广州穗华能源科技有限公司 Routing decision method for multi-user access edge computing terminal of power distribution Internet of things

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109710336A (en) * 2019-01-11 2019-05-03 中南林业科技大学 The mobile edge calculations method for scheduling task of joint energy and delay optimization
CN110099384A (en) * 2019-04-25 2019-08-06 南京邮电大学 Resource regulating method is unloaded based on side-end collaboration more MEC tasks of multi-user

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11423254B2 (en) * 2019-03-28 2022-08-23 Intel Corporation Technologies for distributing iterative computations in heterogeneous computing environments

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109710336A (en) * 2019-01-11 2019-05-03 中南林业科技大学 The mobile edge calculations method for scheduling task of joint energy and delay optimization
CN110099384A (en) * 2019-04-25 2019-08-06 南京邮电大学 Resource regulating method is unloaded based on side-end collaboration more MEC tasks of multi-user

Also Published As

Publication number Publication date
CN110489176A (en) 2019-11-22

Similar Documents

Publication Publication Date Title
CN110489176B (en) Multi-access edge computing task unloading method based on boxing problem
CN109814951B (en) Joint optimization method for task unloading and resource allocation in mobile edge computing network
CN109857546B (en) Multi-server mobile edge computing unloading method and device based on Lyapunov optimization
CN111278132B (en) Resource allocation method for low-delay high-reliability service in mobile edge calculation
CN107995660B (en) Joint task scheduling and resource allocation method supporting D2D-edge server unloading
CN110096362B (en) Multitask unloading method based on edge server cooperation
CN110928654B (en) Distributed online task unloading scheduling method in edge computing system
CN110519370B (en) Edge computing resource allocation method based on facility site selection problem
CN111240701A (en) Task unloading optimization method for end-edge-cloud collaborative computing
CN111711962B (en) Cooperative scheduling method for subtasks of mobile edge computing system
US11172490B2 (en) Apparatus for constructing energy-efficient communication and computation resource allocation framework for multiple communication service and method thereof
CN112000481B (en) Task unloading method for maximizing computing capacity of D2D-MEC system
CN112996056A (en) Method and device for unloading time delay optimized computing task under cloud edge cooperation
CN110647403B (en) Cloud computing resource allocation method in multi-user MEC system
CN109756912B (en) Multi-user multi-base station joint task unloading and resource allocation method
US11968122B2 (en) Joint optimization method and system for delay and spectrum occupation in cloud-edge collaborative network
CN111757361B (en) Task unloading method based on unmanned aerial vehicle assistance in fog network
CN110177055B (en) Pre-allocation method of edge domain resources in edge computing scene
CN110740473A (en) management method for mobile edge calculation and edge server
CN111935677B (en) Internet of vehicles V2I mode task unloading method and system
CN110399210B (en) Task scheduling method and device based on edge cloud
CN114938381B (en) D2D-MEC unloading method based on deep reinforcement learning
CN112988347A (en) Edge computing unloading method and system for reducing system energy consumption and cost sum
CN116233927A (en) Load-aware computing unloading energy-saving optimization method in mobile edge computing
CN110545302A (en) calculation migration method, equipment and storage medium

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