CN110489176A - A kind of multiple access edge calculations task discharging method based on bin packing - Google Patents
A kind of multiple access edge calculations task discharging method based on bin packing Download PDFInfo
- Publication number
- CN110489176A CN110489176A CN201910795561.8A CN201910795561A CN110489176A CN 110489176 A CN110489176 A CN 110489176A CN 201910795561 A CN201910795561 A CN 201910795561A CN 110489176 A CN110489176 A CN 110489176A
- Authority
- CN
- China
- Prior art keywords
- task
- edge server
- queue
- edge
- user terminal
- 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
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 proposes a kind of multiple access edge calculations task discharging method based on bin packing, user terminal and Edge Server are regarded as task container by this method, task is regarded as article, to which task unloading decision problem in edge calculations is converted to a bin packing, the Edge Server number enabled in network is minimized by heuristic again, task is solved and unloads decision.Include the following steps: to calculate the input data size for holding loading capability and each terminal task of each Edge Server and the ratio of required computing resource first;Then two queues are formed from big to small according to appearance loading capability and task ratio;Finally the task in task queue is successively taken out on the Edge Server for holding the maximum simultaneously also remaining computing resource of loading capability being configured in container queue, repeats this operation up to task queue is sky.The present invention can be suitably used for the task processes of the multiple access edge calculations network of multiple terminals multitask, can clearly make suitable task unloading scheme, minimizes while meeting task time delay and requiring and calculates energy consumption, save the cost.
Description
Technical field
The present invention relates to multiple access edge calculations fields, and in particular to multiple edge clothes into multiple access edge calculations network
The computational resource allocation problem of business device.
Background technique
Internet of Things (Internet of Things, IoT) technology, it is intended to utilize Radio Frequency Identification Technology, wireless data communication
Technology, global positioning system etc., material object is connected together by the communication protocol according to Internet of Things agreement with internet, and then is realized
Information exchange achievees the purpose that Weigh sensor, positioning, tracking, monitoring and managing internet resource.As number of devices increases
The continuous growth of more, calculated performance enhancing and user demand, edge calculations have become a kind of important expansion of Internet of Things
Spread formula.Multiple access edge calculations (Multi-access Edge Computing, MEC) are that one of network technology is emerging
Concept can allow in the computing resource of network edge closer to user, so that optimized allocation of resources, accelerates access speed, make tradition
Cloud computing service improved.Rough says, user equipment can connect one or more Edge Servers, to unload carrying
The too slow or too high energy consumption computation-intensive task of row.
Nearly 2 years edge calculations technologies develop rapidly in IoT applications, occur more equipment in edge calculations mode
Terminal and more complicated calculating task.In traditional terminal execution pattern, if computation-intensive task fully relies on terminal
It locally executes, may cause that terminal energy consumption is excessive, and Runtime is too long, be unable to satisfy the service requirement of complex application
The problems such as.On the contrary, being handled if all sending cloud center for the data of terminal, although being able to satisfy complex application
Requirement is accurately calculated, but the load of cloud center and network link can be greatly increased, the communication energy consumption for also resulting in terminal is excessive.And
And since the distance between terminal and cloud are partially long, it will cause higher transmission delay, be not able to satisfy some terminals thus and answer
Low latency requirement.
In conclusion appointing for the computing resource of reasonable allocations of edge server when multiple terminal devices processing calculates
When business, it would be desirable to make unloading (offloading) decision.In decision process, it is contemplated that calculating task require time delay and
Two aspects of energy consumption, target are to make the energy consumption of Edge Server minimum while meeting the delay requirement of each task,
It reduces and calculates cost.The problem of we unload task is modeled as a np hard problem, for rapid solving, using heuristic side
The problem is regarded as a bin packing by method, by minimizing the Edge Server quantity for needing to enable, reaches minimum network
The purpose of total energy consumption.This method can solve the unloading decision of terminal task and the Resource Allocation Formula of Edge Server, In
Meet under the delay requirement of terminal task, minimizes task execution energy consumption.
Summary of the invention
The invention proposes a kind of multiple access edge calculations task discharging method based on bin packing, is mainly used in more
Edge calculations field is accessed, major advantage is to reasonably arrange the unloading decision of each terminal task and Edge Server to provide
Source allocation plan minimizes task execution energy consumption.
1. a kind of multiple access edge calculations (Multi-access Edge Computing, MEC) based on bin packing are appointed
Business discharging method, which is characterized in that the task unloading problem of MEC is regarded as bin packing by user equipment, by a kind of heuristic
Method completes task and unloads decision, and the method at least includes the following steps:
Step 1 arranges MEC network scenarios, has multiple user equipment N={ 1,2,3 ... n }, multiple edge calculations in network
Server M={ 1,2,3 ... m } has a plurality of communication channel K={ 1,2,3 ... k }, each user between user equipment and server
Equipment has and only one calculating task in time slot ts, needs to select to be performed locally task and task is still offloaded to edge clothes
Business device executes;
Step 2 calculates each Edge Server appearance loading capability, and all Edge Servers are descending by its appearance loading capability
Sequence, forms a queue X={ X1,X2,X3…Xm};
The CP of step 3, the task of each user terminal of calculatingi(Cost Performance) value, by appointing for user terminal
CP is pressed in businessiIt is worth descending sequence, forms a queue Y={ Y1,Y2,Y3…Ym};
Step 4, user terminal i task presentation beWherein DiFor the size of task input data, CiIt is
Resource required for calculating task is completed,The maximum delay of expression task limits;
Step 5, calculating task user terminal and Edge Server operation energy consumption, it is total to minimize network in time slot ts
Energy consumption is target, using user terminal task unloading decision as variable, constructs Optimized model;
The task unloading decision problem of step 5 is regarded as a bin packing by step 6, by user terminal and edge service
Device is regarded as task container, and the task of user terminal is regarded as article, and the edge enabled in network is minimized by heuristic and is taken
Business device quantity solves task and unloads decision;
Task in queue Y is successively taken out the Edge Server being allocated in queue X by step 7, will meet task time delay about
The CP of beamiIt is worth highest task and is unloaded to the appearance maximum Edge Server of loading capability;
Step 8 deletes being allocated successfully for task from queue Y, be up to the maximum Edge Server for holding loading capability from
It is deleted in queue X, repeats step 7 until queue Y is empty.
The present invention has the advantage that
1, the time delay demand that can satisfy each task, does not influence the service experience of user.
2, it quickly calculates to go out on missions by heuristic and unloads decision, reduce task unloading decision and server resource
The waiting time delay of distribution.
3, the Edge Server quantity run in network can be minimized, energy consumption, save the cost are reduced.
Detailed description of the invention
Fig. 1 is abstract flow chart of the invention;
Fig. 2 is algorithm flow chart of the invention.
Specific embodiment
The present invention is described in further detail below in conjunction with the accompanying drawings.
Step 1: arrangement multiple access edge calculations network scenarios
1) in current multiple access edge calculations network, 5 Edge Servers, in a dormant state, current time slots have been set
There are 10 users to issue task requests in interior multiple access edge calculations network, total computing resource of Edge Server is sufficient for institute
There is the calculating demand of task;The appearance loading capability for calculating 5 Edge Servers in network, by Edge Server according to its Rong Zaineng
Power sorts from large to small, and forms a queue X={ X1,X2,X3,X4,X5};
2) task presentation of user terminal i isWherein DiFor the size of task input data, CiIt is to complete
Resource required for calculating task,The maximum delay of expression task limits, and the task presentation as numbered the user terminal for being 1 is
3) CP (Cost Performance) value of 10 tasks in current time slots is calculated, task CP value calculation is such as
Under:
CP=D/C
D is the size of task input data, and C is resource required for completing calculating task
Task is sorted from large to small according to the size of CP value, forms a queue Y={ Y1,Y2,Y3,Y4,Y5,Y6,Y7,
Y8,Y9,Y10};
Step 2: establishing Optimized model
Calculating task is to minimize network total energy consumption in time slot ts in the operation energy consumption of user terminal and Edge Server
Target constructs Optimized model: 1) defining a using user terminal task unloading decision as variablei,j,kIndicate the unloading of user equipment i
Decision, ai,j,k={ 0,1 }, ai,j,k=1 expression user i calculates task by mode j, and k represents communication channel, j=
{ 1,2 } respectively indicates task in local computing and is unloaded to Edge Server and calculates, and when j=1, the value of k is meaningless, defaults side
The amount of computational resources of edge server held loading capability and be greater than each required by task;
If 2) task is completed in user terminal, the required time isEnergy consumption isWherein
K is constant, depends on chip architecture, and f is user terminal computing capability;
If 3) task is offloaded to Edge Server, need to transmit data by mobile network, task is completed at this time
Time isWherein fx is Edge Server computing capability, and B is network bandwidth, and energy consumption includes data transmission
Energy consumption and Edge Server operation energy consumption, are expressed asWherein PmPower when being run for Edge Server,
PiIt is the transimission power of user, wherein Edge Server operation energy consumption is higher than task in the operation energy consumption of user terminal;
4) it converts problem to and asksOptimal solution problem, obey it is following about
Beam condition:
ai,j,kIt is decision variable to be solved.
Step 3: algorithm is realized
The Optimized model proposed in step 2 is a np hard problem, in order to acquire the solution of the problem in polynomial time,
The problem is regarded as a bin packing, according to the requirement of task delay constraint, the task of user terminal is distributed to edge service
Device or user terminal self-operating complete the required Edge Server quantity enabled by minimizing task, reach minimum
The purpose of network energy consumption, the heuritic approach are implemented as follows:
1) by Y1Plan is unloaded to X1On, it calculates and completes Y1Required time delayIfThen carry out this
Secondary unloading, at this time a1,2,k=1;Otherwise Y1It is calculated on its user terminal, at this time a1,1,k=1, and Y is deleted from queue1;
2) by Y2Plan is unloaded to X1On, it calculates and completes Y2Required time delayIfThen carry out this
Secondary unloading, at this time a2,2,k=1;Otherwise Y2It is calculated on its user terminal, at this time a2,1,k=1, and Y is deleted from queue2;
3) aforesaid operations are repeated, until X1Residue holds the demand that loading capability is unable to satisfy the task of current planning unloading, this
When X is deleted from queue1, and X is set by Edge Server to be unloaded2, and according to the remaining task 1) handled in Y;
4) aforesaid operations are repeated to queue X, Y, until queue Y is empty, completes all tasks in current time slots at this time
Unloaded operation.
Claims (5)
1. a kind of multiple access edge calculations (Multi-access Edge Computing, MEC) task based on bin packing is unloaded
Support method, which is characterized in that the task unloading problem of MEC is regarded as bin packing by user equipment, passes through a kind of heuristic
Completion task unloads decision, and the method at least includes the following steps:
Step 1 arranges MEC network scenarios, has multiple user equipment N={ 1,2,3 ... n }, multiple edge calculations services in network
Device M={ 1,2,3 ... m } has a plurality of communication channel K={ 1,2,3 ... k }, each user equipment between user equipment and server
Have and only one calculating task in time slot ts, needs to select to be performed locally task task is still offloaded to Edge Server
It executes;
Step 2 calculates each Edge Server appearance loading capability, and all Edge Servers are held the descending row of loading capability by it
Sequence forms a queue X={ X1,X2,X3…Xm};
Step 3, user terminal i task presentation beWherein DiFor the size of task input data, CiIt is to complete meter
The resource that required by task is wanted is calculated,The maximum delay of expression task limits;
The CP of step 4, the task of each user terminal of calculatingi(Cost Performance) value, the task of user terminal is pressed
CPiBe worth it is descending be ranked up, form a queue Y={ Y1,Y2,Y3…Ym};
Step 5, calculating task user terminal and Edge Server operation energy consumption, to minimize network total energy consumption in time slot ts
Optimized model is constructed using user terminal task unloading decision as variable for target;
The task unloading decision problem of step 5 is regarded as a bin packing by step 6, and user terminal and Edge Server are regarded
Make task container, the task of user terminal is regarded as article, minimizes the Edge Server enabled in network by heuristic
Quantity solves task and unloads decision;
Task in queue Y is successively taken out the Edge Server being allocated in queue X by step 7, will meet task delay constraint
CPiIt is worth highest task and is unloaded to the appearance maximum Edge Server of loading capability;
Step 8 deletes being allocated successfully for task from queue Y, is up to the maximum Edge Server for holding loading capability from queue X
Middle deletion repeats step 7 until queue Y is empty.
2. the multiple access edge calculations task discharging method according to claim 1 based on bin packing, feature exist
In carry out calculate unloading when, the communication channel between user terminal and Edge Server be it is orthogonal, will not in communication process
It causes to influence each other, and the network bandwidth of each communication channel is identical.
3. the multiple access edge calculations task discharging method according to claim 1 based on bin packing, feature exist
The task CP value calculation described in step 4 is as follows:
CP=D/C
D is the size of task input data, and C is resource required for completing calculating task.
4. the multiple access edge calculations task discharging method according to claim 1 based on bin packing, feature exist
The Optimized model construction method described in step 5 includes at least following steps:
1) a is definedi,j,kTo indicate the unloading decision of user equipment i, ai,j,k={ 0,1 }, ai,j,k=1 expression user i passes through mode
J calculates task, and k represents communication channel, and j={ 1,2 } respectively indicates task in local computing and is unloaded to edge and takes
Device of being engaged in calculates, and when j=1, the value of k is meaningless, defaults the calculating of Edge Server held loading capability and be greater than each required by task
Stock number;
If 2) task is completed in user terminal, the required time isEnergy consumption isWherein k is
Constant, depends on chip architecture, and f is user terminal computing capability;
If 3) task is offloaded to Edge Server, need to transmit data by mobile network, at this time task completion time
ForWherein fmFor Edge Server computing capability, B is network bandwidth, energy consumption include data transmissions consumption and
Edge Server operation energy consumption, is expressed asWherein PmPower when being run for Edge Server, PiIt is to use
The transimission power at family, wherein Edge Server operation energy consumption is higher than task in the operation energy consumption of user terminal;
4) it converts problem to and asksOptimal solution problem, obey following constraint item
Part:
ai,j,kIt is decision variable to be solved.
5. the multiple access edge calculations task discharging method according to claim 1 based on bin packing, feature exist
It is a np hard problem in the Optimized model proposed for solution procedure 5, in order to acquire the solution of the problem in polynomial time, incites somebody to action
The problem is regarded as a bin packing, and according to the requirement of task delay constraint, the task of user terminal is distributed to Edge Server
Or user terminal self-operating, the required by task Edge Server quantity to be enabled is completed by minimizing, reaches minimum net
The purpose of network energy consumption, the heuristic include at least following steps:
1) first the highest task of CP value in queue Y is handled, the Edge Server of queue X head of the queue is set as destination service
Device first determines whether can be unloaded to the task on object edge server, ifThen unloaded,
Decision variable is set, and otherwise the task is run on the subscriber terminal, decision variable is arranged, and update effective appearance of destination server
It carries, after being assigned, which is deleted from queue Y;
2) step 1 is repeated, until the payload of object edge server can not be further continued for accommodating more multitask, at this time by target
Server is deleted from queue X, is selected the Edge Server of head of the queue as the object edge server in step 1 again, is repeated
Step 1;
3) step 1 and step 2 are repeated, until being sky in queue Y, the completing time slot ts at this time of the task unloads decision.
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 true CN110489176A (en) | 2019-11-22 |
CN110489176B 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) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111049917A (en) * | 2019-12-19 | 2020-04-21 | 电子科技大学 | Mobile-aware multi-user unloading optimization method based on mobile edge calculation |
CN111641891A (en) * | 2020-04-16 | 2020-09-08 | 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 |
CN112612553A (en) * | 2021-01-06 | 2021-04-06 | 重庆邮电大学 | Container technology-based edge computing task unloading method |
CN113015217A (en) * | 2021-02-07 | 2021-06-22 | 重庆邮电大学 | Edge cloud cooperation low-cost online multifunctional business computing unloading method |
CN114567587A (en) * | 2022-03-14 | 2022-05-31 | 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 (3)
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 |
US20190220703A1 (en) * | 2019-03-28 | 2019-07-18 | Intel Corporation | Technologies for distributing iterative computations in heterogeneous computing environments |
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 |
-
2019
- 2019-08-27 CN CN201910795561.8A patent/CN110489176B/en active Active
Patent Citations (3)
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 |
US20190220703A1 (en) * | 2019-03-28 | 2019-07-18 | Intel Corporation | Technologies for distributing iterative computations in heterogeneous computing environments |
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 |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111049917A (en) * | 2019-12-19 | 2020-04-21 | 电子科技大学 | Mobile-aware multi-user unloading optimization method based on mobile edge calculation |
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 |
CN111641891A (en) * | 2020-04-16 | 2020-09-08 | 北京邮电大学 | Task peer-to-peer unloading method and device in multi-access edge computing system |
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 |
CN112612553A (en) * | 2021-01-06 | 2021-04-06 | 重庆邮电大学 | Container technology-based edge computing task unloading method |
CN112612553B (en) * | 2021-01-06 | 2023-09-26 | 重庆邮电大学 | Edge computing task unloading method based on container technology |
CN113015217A (en) * | 2021-02-07 | 2021-06-22 | 重庆邮电大学 | Edge cloud cooperation low-cost online multifunctional business computing unloading method |
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 |
CN114567587A (en) * | 2022-03-14 | 2022-05-31 | 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 |
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 |
Also Published As
Publication number | Publication date |
---|---|
CN110489176B (en) | 2022-05-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110489176A (en) | A kind of multiple access edge calculations task discharging method based on bin packing | |
Meng et al. | Dedas: Online task dispatching and scheduling with bandwidth constraint in edge computing | |
Liu et al. | Delay-optimal computation task scheduling for mobile-edge computing systems | |
CN110096362B (en) | Multitask unloading method based on edge server cooperation | |
US8695009B2 (en) | Allocating tasks to machines in computing clusters | |
CN110099384A (en) | Resource regulating method is unloaded based on side-end collaboration more MEC tasks of multi-user | |
Huang et al. | DeePar: A hybrid device-edge-cloud execution framework for mobile deep learning applications | |
CN107122249A (en) | A kind of task unloading decision-making technique based on edge cloud pricing mechanism | |
CN110928654A (en) | Distributed online task unloading scheduling method in edge computing system | |
CN108600299B (en) | Distributed multi-user computing task unloading method and system | |
CN109951869A (en) | A kind of car networking resource allocation methods calculated based on cloud and mist mixing | |
CN110519370A (en) | A kind of edge calculations resource allocation methods based on Facility Location Problem | |
CN104023408A (en) | Dispatcher and data dispatching method based on network multi-path parallel transmission | |
CN110489233A (en) | Equipment task unloading and cpu frequency modulation method and system based on mobile edge calculations | |
El Haber et al. | Computational cost and energy efficient task offloading in hierarchical edge-clouds | |
Din et al. | Mobility-aware resource allocation in multi-access edge computing using deep reinforcement learning | |
CN114938381A (en) | D2D-MEC unloading method based on deep reinforcement learning and computer program product | |
CN111158893B (en) | Task unloading method, system, equipment and medium applied to fog computing network | |
CN114710571A (en) | Data packet processing system | |
CN117579701A (en) | Mobile edge network computing and unloading method and system | |
Chen et al. | Joint optimization of task caching, computation offloading and resource allocation for mobile edge computing | |
Li | Optimization of task offloading problem based on simulated annealing algorithm in MEC | |
CN110177383A (en) | The efficiency optimization method of task based access control scheduling and power distribution in mobile edge calculations | |
CN115550983A (en) | Hierarchical control-based mobile environment communication transmission method | |
Omer et al. | Optimization model for time sensitive IoT requests |
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 |