CN109561148A - Distributed task dispatching method in edge calculations network based on directed acyclic graph - Google Patents
Distributed task dispatching method in edge calculations network based on directed acyclic graph Download PDFInfo
- Publication number
- CN109561148A CN109561148A CN201811462177.8A CN201811462177A CN109561148A CN 109561148 A CN109561148 A CN 109561148A CN 201811462177 A CN201811462177 A CN 201811462177A CN 109561148 A CN109561148 A CN 109561148A
- Authority
- CN
- China
- Prior art keywords
- task
- processor
- directed acyclic
- acyclic graph
- node
- 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
-
- 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
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1004—Server selection for load balancing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/48—Program initiating; Program switching, e.g. by interrupt
- G06F9/4806—Task transfer initiation or dispatching
- G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
- G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5072—Grid computing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5083—Techniques for rebalancing the load in a distributed system
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/142—Network analysis or design using statistical or mathematical methods
Abstract
The invention proposes the distributed task dispatching methods in edge calculations network based on directed acyclic graph.The all data parameter of the required scheduler task based on directed acyclic graph and the performance parameter of fringe node processor are respectively obtained from mobile terminal and edge network center control nodes first, in accordance with center type node scheduling mode;Then the task that in-degree in directed acyclic graph is 0 is formed into a task sequence, and deletes these tasks in figure, then update directed acyclic graph, repeat the step until tasks all in original image are contained in variant layer task sequence;Finally to optimize the task schedule time, task communication time, queuing time are assigned to task in each layer task sequence on modal processor as target and execute.The present invention can be suitably used for the edge calculations network data processing of different scales size, in linearization process directed acyclic graph with the task of precedence constraints relationship, can optimize overall scheduling time.
Description
Technical field
The invention belongs to mobile edge calculations fields, especially relate in edge calculations network based on directed acyclic graph
Distributed task dispatching method.
Background technique
Internet of Things (Internet Of Things) is the important component of generation information technology, and when informationization
The important development stage in generation.However, as mobile device number increases, the enhancing and user demand of edge device calculated performance
Continuous growth, edge calculations have become a kind of important extension form of Internet of Things.Edge calculations refer to close to object or
The side of data source header, the open platform being integrated using network, calculating, storage, application core ability, provides service nearby.
Edge calculations different from the maximum of cloud computing are exactly that edge device is used to provide service for user, these edge devices can
It can be intelligent terminal, automobile, electric appliance, factory, wireless base station etc..Edge calculations are intended to alleviate cloud meter using these edge devices
The pressure of calculation, the fluency and agility for reducing energy consumption, improve data transfer that data calculate meet client and server-side
Requirement of real-time.
With the fast development of edge calculations in IoT applications, more and more terminal devices have been added to edge meter
In the mode of calculation.Traditional end application tupe is appointed if complex application leans on local computing completely
Business execution overlong time, energy consumption is excessive, computational accuracy is low, is unable to satisfy the service requirement of complex application.For centralization
Cloud computing mode is handled if being transferred to all terminal complex application data apart from farther away cloud computing center,
Although it can support accurately calculating for complex application, great amount of terminals is transferred data at cloud data center
Reason, this will increase the load of cloud data center and network link, also increases the communication energy consumption of terminal;And terminal and cloud service
Long range between device, unstable backbone network also lead to higher transmission delay, it is difficult to meet the low latency of terminal applies
It is required that.
In conclusion being utilized to meet the service quality of user's complicated applications (Quality Of Service) requirement
Edge calculations platform pushes cloud service to network edge, when terminal processes complicated applications, makes unloading (offloading) certainly
A part of plan, application is executed in terminal, and computationally intensive another part will be discharged on nearest edge device and execute, this
The execution efficiency that sample had both improved application has ensured that the real-time of data processing also saves network bandwidth resources and reduces terminal
Operation energy consumption.In order to realize the edge calculations processing of complicated applications, by broken down into program pending in terminal at task sequence, task
Sequence is born by dependence before and after directed acyclic graph (Directed Acyclic Graph, DAG) reflection task according to task
Equilibrium model is carried, task sequence is assigned on different edge device processors and is run, held to reduce terminal complicated process
Row energy consumption and program execution time.
Summary of the invention
The invention proposes a kind of distributed task dispatching methods based on directed acyclic graph in edge calculations network, mainly
Advantage is in linearization process directed acyclic graph with the task of precedence constraints relationship, can to optimize overall scheduling time.
1. the distributed task dispatching method in edge calculations network based on directed acyclic graph, it is characterised in that the method
It at least includes the following steps:
Step 1 arranges edge calculations network scenarios, and each edge service modal processor is by P={ p in network1,p2,...,
pt,...,pmIndicate, P1Indicate local terminal processor, processor ptData processing speed CtIt indicates, at terminal and node
Managing the channel width between device is Wvp, to CtQueue Y is formed by sorting from large to small;
The directed acyclic graph (DAG) of step 2, input terminal waiting task, G={ V, E }, wherein node set V=
{v1,v2,...,vi,...,vnIndicate task to edge calculations network processes, the dependence between line set E expression task,
Sequence task can just start to process after completing before sequence task has to wait for afterwards;
Node in step 3, traversing graph G finds out the node i.e. not task v of previous task that in-degree is 0i, by these
Task is according to its data scale SiIt is ranked up from big to small, forms a task sequence Qj, QjIn task between no longer have first
Dependence afterwards, wherein j indicates the task sequence level number, and j is smaller to indicate that the task priority in the sequence is higher, and j's is initial
Value is 1, KjIndicate the task number in the task sequence;
Step 4 deletes Q in figure GjNode corresponding to middle task updates remaining node in-degree, generates new figure G, j
=j+1;
Step 5 repeats step 3, until there is no until node in figure G;
Step 6, to optimize the task schedule time as target, using the method for scheduling task based on load balancing by above-mentioned step
Suddenly the task in task sequence set obtained is assigned on edge calculations nodes processor, is dispatched and is successfully then used ai,t
=1 indicates, otherwise ai,t=0.
2. the distributed task dispatching side based on directed acyclic graph in edge calculations network according to claim 1
Method, it is characterised in that divided the task in DAG in stratification without successive dependence by finding the node that node in-degree is 0
Task sequence set, if x < y, QyInterior task has to wait for QxAfter interior task is all disposed at ability
Reason.
3. the distributed task dispatching side based on directed acyclic graph in edge calculations network according to claim 1
Method, it is characterised in that in order to reduce the scheduling time of single layer task sequence, task is assigned to difference using load-balancing method
Parallel processing on modal processor, to optimize the latest finishing time of every layer of task.
4. the distributed task dispatching side based on directed acyclic graph in edge calculations network according to claim 1
Method, it is characterised in that the method for scheduling task of load balancing includes at least following steps:
1) level number j initial value j=1 is set;
2) compare QjIn task number and processor sum, if QjIn waiting task number KjLess than or equal to place
Device sum m is managed, then chooses preceding K in queue Yj4) a modal processor executes, otherwise execute 3);
3) Q is taken according to processor sum mjPreceding m task processing;
4) transmission time that current layer waiting task is transferred to different modal processors is calculatedri,tIt indicates
Channel transmission rate, ri,t=Wvplog2(1+Hi,tgi,t/σ2);Hi,tIndicate transimission power, gi,tIndicate channel gain, σ2Expression is made an uproar
Acoustical power, WvpIndicate channel width;
5) processing time of the current layer waiting task on different modal processors is calculated
6) average time waited when current layer waiting task is lined up, X are calculatedi=Si/ λ, λ indicate the average speed of subchannel
Rate;
7) current layer waiting task and current available node processor are obtained about scheduling total timeMatrix B,
8) since the 1st row of matrix, choose the row scheduling total time the smallest column, using processor corresponding to the column as
The execution processor of the row task, if the processor that certain row task is chosen is chosen and is not chosen selected by uplink task
Execution processor of the processor taken as the row task after selection finishes, jumps next line and repeats the step, until
Task corresponding to all rows is completed the selection for executing processor in matrix;
9) in task processes, if QjIn there is also not processed tasks, then when available free processor occur
When, successively by QjIn untreated task be assigned in idle processor and run;
If 10) QjNo longer there is not processed task, then j=j+1, jump procedure 2 carries out next layer of task point
With process, until tasks all in Q all it is processed finish until.
Compared with prior art, advantage of the process is that
The distributed task dispatching method in edge calculations network based on directed acyclic graph is proposed, is made in directed acyclic graph
With precedence constraints relationship task linearisation, then consider call duration time, processing time, queuing time global optimization, times
Business is dispatched on most suitable modal processor.
Detailed description of the invention
Fig. 1 is abstract flow chart of the invention;
Fig. 2 is the task layering flow chart in the present invention;
Fig. 3 is the method for allocating tasks figure in the present invention;
Fig. 4 is edge calculations Task Scheduling Model.
Specific embodiment
The present invention is described in further detail below in conjunction with the accompanying drawings.
Parallel task: the complicated applications run on intelligent terminal can be indicated with directed acyclic graph DAG, be defined as G
{V,E}。
Parallel Task Scheduling is generally divided into three steps, it is first determined then the priority of task node determines that task schedule is total
The task of priority from high to low is finally sequentially allocated suitable processor according to task schedule overall goal by body target
On.
Assuming that the DAG figure in Fig. 4 is the Task-decomposing of game application by taking the game application run on intelligent terminal as an example
Figure, three receiving points are nearest edge calculations equipment.
Step 1: determining task priority, the present invention is layered the priority of the task of determination, building method using DAG task
It is as follows:
A) node in traversing graph G finds out the node i.e. not task v of previous task that in-degree is 0i, by these tasks
According to its data scale SiIt is ranked up from big to small, forms a task sequence Qj, QjIn task between no longer have successively according to
The relationship of relying, wherein j indicates the task sequence level number, and j is smaller to indicate that the task priority in the sequence is higher, and the initial value of j is
1, KjIndicate the task number in the task sequence;Find in-degree be 0 task specific method: in DAG figure, task with appoint
There is the matrix about degree between business, task 1 and task 2 have in-degree relationship to be then labeled as 1, and task 1 and task 2 have out-degree relationship
It is then labeled as 0, not related label is;The relationship spent between logger task and task according to matrix;In-degree is 0 at this time
Node is n1, it is put into Q1, first layer task is { n1};
B) Q is deleted in figure GjNode corresponding to middle task updates remaining node in-degree, generates new figure G, j=j+
1;
C) step a), b) is repeated, until until node is not present in figure G.
It is { n by above-mentioned steps first layer task1};Second layer task is { n6,n2,n5,n4,n3};Third layer task is
{n7,n8,n9};4th layer of task is { n10}。
Step 2: determining task totality regulation goal:
The target of entire task schedule process minimizes the task schedule time, and the task schedule time includes the transmission of task
Time, processing time and queuing time.
Step 3: task is assigned on appropriate node processor according to layering result:
A) modal processor is sorted from fast to slow according to calculating speed, then determines every layer of task number for needing to dispatch
With the relationship between processor number, when processor number be greater than current task number when: such as first layer, third layer and the 4th
Layer task, more every layer of total task number and processor sum, if the task number to be allocated of this layer is less than or equal to processor number
Mesh is then chosen with the same number of modal processor of current task as dispatch processor, and selection rule is preferential selection processing
The strong modal processor of ability;
B) when processor number is less than current task number: such as second layer, according to processor number pmExecute the layer
Preceding pmA task;
C) transmission time for calculating each task of current layer and current each available processors, calculates the time, queuing time:
The each task of current layer and the transmission time formula of current each available processors are
The each task of current layer and the processing time formula of current each available processors are
The average time formula that current layer waiting task waits when being lined up is Xi=Si/λ;
D) current layer waiting task and current available node processor are obtained about scheduling total timeMatrix B,
E) since the 1st row of matrix, choose the row scheduling total time the smallest column, using processor corresponding to the column as
The execution processor of the row task, if the processor that certain row task is chosen is chosen and is not chosen selected by uplink task
Execution processor of the processor taken as the row task after selection finishes, jumps next line and repeats the step, until
Task corresponding to all rows is completed the selection for executing processor in matrix;
F) in task processes, if QjIn there is also not processed tasks, then when available free processor occur
When, successively by QjIn untreated task be assigned in idle processor and run;
If g) QjNo longer there is not processed task, then j=j+1, carries out next layer of task assignment procedure, until
All tasks are all disposed in Q.
Claims (4)
1. the distributed task dispatching method in edge calculations network based on directed acyclic graph, it is characterised in that the method is at least
The following steps are included:
Step 1 arranges edge calculations network scenarios, and each edge service modal processor is by P={ p in network1,p2,...,
pt,...,pmIndicate, P1Indicate local terminal processor, processor ptData processing speed CtIt indicates, at terminal and node
Managing the channel width between device is Wvp, to CtQueue Y is formed by sorting from large to small;
The directed acyclic graph (DAG) of step 2, input terminal waiting task, G={ V, E }, wherein node set V={ v1,
v2,...,vi,...,vnIndicate task to edge calculations network processes, the dependence between line set E expression task, postorder
Sequence task can just start to process after completing before task has to wait for;
Node in step 3, traversing graph G finds out the node i.e. not task v of previous task that in-degree is 0i, by these tasks according to
According to its data scale SiIt is ranked up from big to small, forms a task sequence Qj, QjIn task between no longer have successively rely on
Relationship, wherein j indicates the task sequence level number, and j is smaller to indicate that the task priority in the sequence is higher, and the initial value of j is 1,
KjIndicate the task number in the task sequence;
Step 4 deletes Q in figure GjNode corresponding to middle task updates remaining node in-degree, generates new figure G, j=j+1;
Step 5 repeats step 3, until there is no until node in figure G;
Step 6, to optimize the task schedule time as target, above-mentioned steps are obtained using the method for scheduling task based on load balancing
To task sequence set in task be assigned on edge calculations nodes processor, dispatch and successfully then use ai,t=1 table
Show, otherwise ai,t=0.
2. the distributed task dispatching method in edge calculations network according to claim 1 based on directed acyclic graph,
It is characterized in that the task in DAG is divided in stratification without successive dependence by finding the node that node in-degree is 0
Task sequence set, if x < y, QyInterior task has to wait for QxInterior task could be handled after being all disposed.
3. the distributed task dispatching method in edge calculations network according to claim 1 based on directed acyclic graph,
It is characterized in that task is assigned to different nodes using load-balancing method in order to reduce the scheduling time of single layer task sequence
Parallel processing on processor, to optimize the latest finishing time of every layer of task.
4. the distributed task dispatching method in edge calculations network according to claim 1 based on directed acyclic graph,
It is characterized in that the method for scheduling task of load balancing includes at least following steps:
1) level number j initial value j=1 is set;
2) compare QjIn task number and processor sum, if QjIn waiting task number KjLess than or equal to processor
Total m then chooses preceding K in queue Yj4) a modal processor executes, otherwise execute 3);
3) Q is taken according to processor sum mjPreceding m task processing;
4) transmission time that current layer waiting task is transferred to different modal processors is calculatedri,tIndicate channel
Transmission rate, ri,t=Wvplog2(1+Hi,tgi,t/σ2);Hi,tIndicate transimission power, gi,tIndicate channel gain, σ2Indicate noise function
Rate, WvpIndicate channel width;
5) processing time of the current layer waiting task on different modal processors is calculated
6) average time waited when current layer waiting task is lined up, X are calculatedi=Si/ λ, λ indicate subchannel Mean Speed;
7) current layer waiting task and current available node processor are obtained about scheduling total timeMatrix B,
8) since the 1st row of matrix, row scheduling total time the smallest column are chosen, using processor corresponding to the column as the row
The execution processor of task, if the processor that certain row task is chosen is chosen unselected selected by uplink task
Execution processor of the processor as the row task after selection finishes, jumps next line and repeats the step, until matrix
Task corresponding to interior all rows is completed the selection for executing processor;
9) in task processes, if QjIn there is also not processed tasks, then when available free processor occurs, according to
It is secondary by QjIn untreated task be assigned in idle processor and run;
If 10) QjNo longer there is not processed task, then j=j+1, the task that jump procedure 2 carries out next layer was distributed
Journey, until tasks all in Q all it is processed finish until.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811462177.8A CN109561148B (en) | 2018-11-30 | 2018-11-30 | Distributed task scheduling method based on directed acyclic graph in edge computing network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811462177.8A CN109561148B (en) | 2018-11-30 | 2018-11-30 | Distributed task scheduling method based on directed acyclic graph in edge computing network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109561148A true CN109561148A (en) | 2019-04-02 |
CN109561148B CN109561148B (en) | 2021-03-23 |
Family
ID=65868556
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811462177.8A Active CN109561148B (en) | 2018-11-30 | 2018-11-30 | Distributed task scheduling method based on directed acyclic graph in edge computing network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109561148B (en) |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110069341A (en) * | 2019-04-10 | 2019-07-30 | 中国科学技术大学 | What binding function configured on demand has the dispatching method of dependence task in edge calculations |
CN110134505A (en) * | 2019-05-15 | 2019-08-16 | 湖南麒麟信安科技有限公司 | A kind of distributed computing method of group system, system and medium |
CN110231984A (en) * | 2019-06-06 | 2019-09-13 | 湖南大学 | Multi-workflow method for allocating tasks, device, computer equipment and storage medium |
CN111367644A (en) * | 2020-03-17 | 2020-07-03 | 中国科学技术大学 | Task scheduling method and device for heterogeneous fusion system |
CN111427679A (en) * | 2020-03-25 | 2020-07-17 | 中国科学院自动化研究所 | Computing task scheduling method, system and device facing edge computing |
WO2020207457A1 (en) * | 2019-04-12 | 2020-10-15 | 阿里巴巴集团控股有限公司 | Method and system for identifying directed graph, and server |
CN111782389A (en) * | 2020-06-22 | 2020-10-16 | 中科边缘智慧信息科技(苏州)有限公司 | Task scheduling system and method under mobile edge information service network |
CN111930448A (en) * | 2020-09-23 | 2020-11-13 | 南京梦饷网络科技有限公司 | Method, electronic device, and storage medium for service distribution |
CN112130927A (en) * | 2020-09-21 | 2020-12-25 | 安阳师范学院 | Reliability-enhanced mobile edge computing task unloading method |
CN112181655A (en) * | 2020-09-30 | 2021-01-05 | 杭州电子科技大学 | Hybrid genetic algorithm-based calculation unloading method in mobile edge calculation |
CN112463397A (en) * | 2020-12-10 | 2021-03-09 | 中国科学院深圳先进技术研究院 | Lock-free distributed deadlock avoidance method and device, computer equipment and readable storage medium |
CN112884247A (en) * | 2021-03-23 | 2021-06-01 | 中国人民解放军国防科技大学 | Command control center and control method |
CN113535367A (en) * | 2021-09-07 | 2021-10-22 | 北京达佳互联信息技术有限公司 | Task scheduling method and related device |
CN115145711A (en) * | 2022-09-02 | 2022-10-04 | 北京睿企信息科技有限公司 | Data processing system for acquiring directed acyclic graph task result |
CN115840631A (en) * | 2023-01-04 | 2023-03-24 | 中科金瑞(北京)大数据科技有限公司 | RAFT-based high-availability distributed task scheduling method and equipment |
CN116170846A (en) * | 2023-04-25 | 2023-05-26 | 天津市工业和信息化研究院(天津市节能中心、天津市工业和信息化局教育中心) | Distributed edge computing system and edge computing method for data communication |
CN116302396A (en) * | 2023-02-13 | 2023-06-23 | 上海浦东发展银行股份有限公司 | Distributed task scheduling method based on directed acyclic |
US11887353B1 (en) | 2022-09-21 | 2024-01-30 | Zhejiang Lab | Deep learning image classification oriented to heterogeneous computing device |
WO2024060810A1 (en) * | 2022-09-21 | 2024-03-28 | 之江实验室 | Heterogeneous computing device-oriented deep learning image classification |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102831012A (en) * | 2011-06-16 | 2012-12-19 | 日立(中国)研究开发有限公司 | Task scheduling device and task scheduling method in multimode distributive system |
CN103235640A (en) * | 2013-01-08 | 2013-08-07 | 北京邮电大学 | DVFS-based energy-saving dispatching method for large-scale parallel tasks |
CN103631657A (en) * | 2013-11-19 | 2014-03-12 | 浪潮电子信息产业股份有限公司 | Task scheduling algorithm based on MapReduce |
CN104052811A (en) * | 2014-06-17 | 2014-09-17 | 华为技术有限公司 | Service scheduling method and device and system |
CN104965754A (en) * | 2015-03-31 | 2015-10-07 | 腾讯科技(深圳)有限公司 | Task scheduling method and task scheduling apparatus |
US20170223097A1 (en) * | 2016-02-01 | 2017-08-03 | Sas Institute Inc. | Node device function and cache aware task assignment |
US20180152361A1 (en) * | 2016-11-29 | 2018-05-31 | Hong-Min Chu | Distributed assignment of video analytics tasks in cloud computing environments to reduce bandwidth utilization |
CN108737462A (en) * | 2017-04-17 | 2018-11-02 | 华东师范大学 | A kind of cloud computation data center method for scheduling task based on graph theory |
CN108897625A (en) * | 2018-07-06 | 2018-11-27 | 陈霖 | Method of Scheduling Parallel based on DAG model |
-
2018
- 2018-11-30 CN CN201811462177.8A patent/CN109561148B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102831012A (en) * | 2011-06-16 | 2012-12-19 | 日立(中国)研究开发有限公司 | Task scheduling device and task scheduling method in multimode distributive system |
CN103235640A (en) * | 2013-01-08 | 2013-08-07 | 北京邮电大学 | DVFS-based energy-saving dispatching method for large-scale parallel tasks |
CN103631657A (en) * | 2013-11-19 | 2014-03-12 | 浪潮电子信息产业股份有限公司 | Task scheduling algorithm based on MapReduce |
CN104052811A (en) * | 2014-06-17 | 2014-09-17 | 华为技术有限公司 | Service scheduling method and device and system |
CN104965754A (en) * | 2015-03-31 | 2015-10-07 | 腾讯科技(深圳)有限公司 | Task scheduling method and task scheduling apparatus |
US20170223097A1 (en) * | 2016-02-01 | 2017-08-03 | Sas Institute Inc. | Node device function and cache aware task assignment |
US20180152361A1 (en) * | 2016-11-29 | 2018-05-31 | Hong-Min Chu | Distributed assignment of video analytics tasks in cloud computing environments to reduce bandwidth utilization |
CN108737462A (en) * | 2017-04-17 | 2018-11-02 | 华东师范大学 | A kind of cloud computation data center method for scheduling task based on graph theory |
CN108897625A (en) * | 2018-07-06 | 2018-11-27 | 陈霖 | Method of Scheduling Parallel based on DAG model |
Non-Patent Citations (3)
Title |
---|
ANUM MASOOD等: ""HETS: Heterogeneous Edge and Task Scheduling Algorithm for Heterogeneous Computing Systems"", 《2015 IEEE 17TH INTERNATIONAL CONFERENCE ON HIGH PERFORMANCE COMPUTING AND COMMUNICATIONS, 2015 IEEE 7TH INTERNATIONAL SYMPOSIUM ON CYBERSPACE SAFETY AND SECURITY, AND 2015 IEEE 12TH INTERNATIONAL CONFERENCE ON EMBEDDED SOFTWARE AND SYSTEMS》 * |
UMA BOREGOWDA;VENUGOPAL R. CHAKRAVARTHY: ""A Hybrid Task Scheduler for DAG Applications on a Cluster of Processors"", 《2014 FOURTH INTERNATIONAL CONFERENCE ON ADVANCES IN COMPUTING AND COMMUNICATIONS》 * |
尹杨美: ""一种改进的异构多处理器实时任务调度算法研究"", 《中国硕士学位论文全文数据库信息科技辑》 * |
Cited By (26)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110069341A (en) * | 2019-04-10 | 2019-07-30 | 中国科学技术大学 | What binding function configured on demand has the dispatching method of dependence task in edge calculations |
CN110069341B (en) * | 2019-04-10 | 2022-09-06 | 中国科学技术大学 | Method for scheduling tasks with dependency relationship configured according to needs by combining functions in edge computing |
WO2020207457A1 (en) * | 2019-04-12 | 2020-10-15 | 阿里巴巴集团控股有限公司 | Method and system for identifying directed graph, and server |
CN110134505A (en) * | 2019-05-15 | 2019-08-16 | 湖南麒麟信安科技有限公司 | A kind of distributed computing method of group system, system and medium |
CN110231984A (en) * | 2019-06-06 | 2019-09-13 | 湖南大学 | Multi-workflow method for allocating tasks, device, computer equipment and storage medium |
CN111367644A (en) * | 2020-03-17 | 2020-07-03 | 中国科学技术大学 | Task scheduling method and device for heterogeneous fusion system |
CN111427679A (en) * | 2020-03-25 | 2020-07-17 | 中国科学院自动化研究所 | Computing task scheduling method, system and device facing edge computing |
CN111427679B (en) * | 2020-03-25 | 2023-12-22 | 中国科学院自动化研究所 | Computing task scheduling method, system and device for edge computing |
CN111782389A (en) * | 2020-06-22 | 2020-10-16 | 中科边缘智慧信息科技(苏州)有限公司 | Task scheduling system and method under mobile edge information service network |
CN112130927A (en) * | 2020-09-21 | 2020-12-25 | 安阳师范学院 | Reliability-enhanced mobile edge computing task unloading method |
CN112130927B (en) * | 2020-09-21 | 2022-09-13 | 安阳师范学院 | Reliability-enhanced mobile edge computing task unloading method |
CN111930448A (en) * | 2020-09-23 | 2020-11-13 | 南京梦饷网络科技有限公司 | Method, electronic device, and storage medium for service distribution |
CN112181655A (en) * | 2020-09-30 | 2021-01-05 | 杭州电子科技大学 | Hybrid genetic algorithm-based calculation unloading method in mobile edge calculation |
CN112463397A (en) * | 2020-12-10 | 2021-03-09 | 中国科学院深圳先进技术研究院 | Lock-free distributed deadlock avoidance method and device, computer equipment and readable storage medium |
CN112463397B (en) * | 2020-12-10 | 2023-02-10 | 中国科学院深圳先进技术研究院 | Lock-free distributed deadlock avoidance method and device, computer equipment and readable storage medium |
CN112884247A (en) * | 2021-03-23 | 2021-06-01 | 中国人民解放军国防科技大学 | Command control center and control method |
CN112884247B (en) * | 2021-03-23 | 2023-07-14 | 中国人民解放军国防科技大学 | Command management and control center and control method |
CN113535367A (en) * | 2021-09-07 | 2021-10-22 | 北京达佳互联信息技术有限公司 | Task scheduling method and related device |
CN113535367B (en) * | 2021-09-07 | 2022-01-25 | 北京达佳互联信息技术有限公司 | Task scheduling method and related device |
CN115145711A (en) * | 2022-09-02 | 2022-10-04 | 北京睿企信息科技有限公司 | Data processing system for acquiring directed acyclic graph task result |
US11887353B1 (en) | 2022-09-21 | 2024-01-30 | Zhejiang Lab | Deep learning image classification oriented to heterogeneous computing device |
WO2024060810A1 (en) * | 2022-09-21 | 2024-03-28 | 之江实验室 | Heterogeneous computing device-oriented deep learning image classification |
CN115840631A (en) * | 2023-01-04 | 2023-03-24 | 中科金瑞(北京)大数据科技有限公司 | RAFT-based high-availability distributed task scheduling method and equipment |
CN116302396A (en) * | 2023-02-13 | 2023-06-23 | 上海浦东发展银行股份有限公司 | Distributed task scheduling method based on directed acyclic |
CN116302396B (en) * | 2023-02-13 | 2023-09-01 | 上海浦东发展银行股份有限公司 | Distributed task scheduling method based on directed acyclic |
CN116170846A (en) * | 2023-04-25 | 2023-05-26 | 天津市工业和信息化研究院(天津市节能中心、天津市工业和信息化局教育中心) | Distributed edge computing system and edge computing method for data communication |
Also Published As
Publication number | Publication date |
---|---|
CN109561148B (en) | 2021-03-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109561148A (en) | Distributed task dispatching method in edge calculations network based on directed acyclic graph | |
CN111756812B (en) | Energy consumption perception edge cloud cooperation dynamic unloading scheduling method | |
US11514309B2 (en) | Method and apparatus for accelerating distributed training of a deep neural network | |
CN109669768B (en) | Resource allocation and task scheduling method for edge cloud combined architecture | |
CN113950066B (en) | Single server part calculation unloading method, system and equipment under mobile edge environment | |
CN111240701B (en) | Task unloading optimization method for end-side-cloud collaborative computing | |
CN112000388B (en) | Concurrent task scheduling method and device based on multi-edge cluster cooperation | |
CN109656713B (en) | Container scheduling method based on edge computing framework | |
CN108055292B (en) | Optimization method for mapping from virtual machine to physical machine | |
CN112799823B (en) | Online dispatching and scheduling method and system for edge computing tasks | |
CN112231085B (en) | Mobile terminal task migration method based on time perception in collaborative environment | |
CN108769182A (en) | A kind of prediction executes the Combinatorial Optimization dispatching method of task execution time | |
WO2023124947A1 (en) | Task processing method and apparatus, and related device | |
CN114443249A (en) | Container cluster resource scheduling method and system based on deep reinforcement learning | |
CN109005211B (en) | Micro-cloud deployment and user task scheduling method in wireless metropolitan area network environment | |
CN114564312A (en) | Cloud edge-side cooperative computing method based on adaptive deep neural network | |
CN113138860A (en) | Message queue management method and device | |
CN111988787A (en) | Method and system for selecting network access and service placement positions of tasks | |
CN114647515A (en) | GPU cluster-oriented dynamic resource scheduling method | |
CN110048966B (en) | Coflow scheduling method for minimizing system overhead based on deadline | |
CN113741999B (en) | Dependency-oriented task unloading method and device based on mobile edge calculation | |
CN115033359A (en) | Internet of things agent multi-task scheduling method and system based on time delay control | |
CN113139639B (en) | MOMBI-oriented smart city application multi-target computing migration method and device | |
CN102184124B (en) | Task scheduling method and system | |
CN112231117A (en) | Cloud robot service selection method and system based on dynamic vector hybrid genetic algorithm |
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 |