CN111694655B - Multitasking-oriented edge computing resource allocation method - Google Patents

Multitasking-oriented edge computing resource allocation method Download PDF

Info

Publication number
CN111694655B
CN111694655B CN202010172395.9A CN202010172395A CN111694655B CN 111694655 B CN111694655 B CN 111694655B CN 202010172395 A CN202010172395 A CN 202010172395A CN 111694655 B CN111694655 B CN 111694655B
Authority
CN
China
Prior art keywords
task
tasks
calculation
allocation
time
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
CN202010172395.9A
Other languages
Chinese (zh)
Other versions
CN111694655A (en
Inventor
朱卫平
李耀东
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.)
Wuhan University WHU
Original Assignee
Wuhan University WHU
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 Wuhan University WHU filed Critical Wuhan University WHU
Priority to CN202010172395.9A priority Critical patent/CN111694655B/en
Publication of CN111694655A publication Critical patent/CN111694655A/en
Application granted granted Critical
Publication of CN111694655B publication Critical patent/CN111694655B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
    • 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Multi Processors (AREA)

Abstract

The invention discloses a multitasking edge-oriented computing resource allocation method, which comprises the following steps: 1) Constructing an edge computing network scene to obtain data parameters of a mobile terminal task and performance parameters of an edge server; 2) Channel resources are allocated according to the data parameters of the task, the transmission delay of the task is calculated, and the residual maximum delay limit of the task is calculated; 3) Comparing whether the maximum number of tasks which can be executed by the edge server at the same time is smaller than the number of tasks to be unloaded, if not, executing the step S4, and if so, executing the step S5; 4) According to the size relation between the computing resources of the edge server and the minimum computing resources required by completing the tasks and different optimization targets, computing resources are distributed for different unloading tasks; 5) And reasonably scheduling the tasks, determining the starting execution time of each task, and distributing computing resources for each task. The invention realizes the dynamic allocation of the computing resources on the edge server and reduces the total time delay of the task unloading.

Description

Multitasking-oriented edge computing resource allocation method
Technical Field
The present invention relates to mobile communication technology, and in particular, to a method for allocating edge computing resources for multitasking.
Background
With the advent of the universal interconnection age, various intelligent mobile devices are widely applied in the life of people, so that the explosive growth of mobile terminal devices is caused, and meanwhile, massive mobile terminal data is generated. The limited battery capacity and computing storage resources of mobile devices are not capable of handling large amounts of data in a short period of time. One solution is to offload part of the work to a remote cloud server rich in resources to handle, i.e., cloud computing. However, uploading a large amount of mobile terminal data to the cloud center can cause overlarge data volume of a transmission link, network blocking is caused, and a longer transmission distance between the mobile terminal and the cloud computing center is generated, so that obvious transmission delay is generated. To address the latency problem, mobile edge computing applications develop, in a mobile edge computing environment, edge servers are deployed directly at base stations, and mobile devices offload high-energy, high-computation-load tasks to edge servers in the vicinity of the mobile devices, thereby obtaining low-latency, short-range IT environments and computing services. However, since the resources of the edge server are also limited, when multiple tasks contend for the resources of the edge server, the resources required for the multiple tasks would need to be reasonably allocated to meet the requirements of the different tasks. The invention mainly aims at distributing computing resources in the edge server towards multitasking.
The edge computing resource allocation for multitasking will be widely applicable to a variety of applications. The method of the invention can be effectively utilized if the data acquisition problem in food safety is solved. For traditional food safety data such as data in meat processing monitoring scenes, mass data acquired by the terminal equipment are transmitted to a cloud computing center through a transmission link for calculation and storage, and then the result is returned to support decision making and intelligent control. However, on one hand, massive food data can cause channel blockage due to higher network load, and on the other hand, high delay is caused by longer transmission distance, so that a data processing result cannot be timely given out, and related events of food safety cannot be timely monitored. Also, when contaminated meat or non-regular operation of meat occurs, a longer response time delay may lead to a later measure of feedback and response of results, which may cause irreversible damage to food safety. The problems can be solved by applying edge calculation, the edge server is directly deployed on the base station in a mobile edge calculation environment, and the terminal transmits the collected mass food safety data to the edge server which is close to the base station, so that the data transmission and processing time delay is reduced, the analysis and feedback time of the food safety data is shortened, unsafe factors in meat processing can be timely found and changed, and the food safety is ensured from the food source. Meanwhile, when a food safety event occurs, the food safety event can be responded quickly, and timely and effective supervision of the food safety event is ensured. However, the resources of the edge server are limited, and reasonable allocation of the edge server resources is required in order to improve the computational efficiency of task offloading and reduce the total delay.
Disclosure of Invention
The invention aims to provide a multitasking edge computing resource allocation method aiming at the defects in the prior art.
The technical scheme adopted for solving the technical problems is as follows: a multi-task oriented edge computing resource allocation method comprises the following steps:
1) Constructing an edge computing network scene according to the edge server and the mobile terminal with a task to be computed and unloaded to the edge server for processing, and obtaining data parameters of the mobile terminal task and performance parameters of the edge server according to the scene;
the data parameters of the mobile terminal tasks comprise the number n of the mobile terminals, the data quantity of each task and the maximum time delay of each task; the mobile terminals are numbered 1 to n, and the data volume is { D }, respectively 1 ,D 2 ,…,D n Maximum delays of respectivelyThe performance parameters of the edge server comprise computing capacity C of the edge server and the maximum number m of tasks which can be executed simultaneously;
2) Allocating channel resources according to data parameters of task unloading on the mobile terminal, and calculating the transmission delay T of the task i according to the channel resources i trans Determining the remaining maximum delay limit for each task iWherein i=1, 2,..n;
3) Comparing whether the maximum number of tasks which can be executed by the edge server is smaller than the number of unloading tasks n, if m is not smaller than n, executing the step 4), and if m is smaller than n, executing the step 5);
4) Calculating minimum calculation resources required by the unloading tasks under the residual maximum time delay, and distributing calculation resources for different unloading tasks according to the size relation between the edge server calculation resources and the minimum calculation resources required by the tasks and the optimization targets;
5) And reasonably scheduling the tasks, determining the starting execution time of each task, and distributing computing resources for each task according to the set optimization target and combining computing resources according to the edge server and minimum computing resources required by completing the tasks.
According to the scheme, the specific steps in the step 4) are as follows:
in this scenario, i.e. when the maximum number of tasks m that the edge server can simultaneously execute is not less than the number of tasks n, the tasks are offloadedCalculation completion time of i (i=1, 2..n) on edge serverEqual to its calculation time T ci ；
4.1 The minimum computing resource required for completing the unloading task under the residual maximum time delay is calculated as
4.2 If (1)Indicating that all tasks can be processed within the residual maximum time delay, namely that the computing resources of the edge server are sufficient, and distributing the computing resources of each task on the edge server according to the set optimization target;
4.3 If (1)The allocation computing resource duty ratio of each task is obtained according to different allocation algorithms according to the set optimization targets and the actual limitation is carried out, so that the limitation on the computing resources of each task on the edge server is realized.
According to the above scheme, the allocation of the computing resources of each task on the edge server in the step 4.2) adopts the following method:
the computing resource limitation conditions are set as follows:
∑C i ≤C
wherein each task is assigned a computing resource C on the edge server i (i=1, 2, n), each taskThe calculation completion time is
Setting optimization targets including shortest and longest calculation completion timesOr the shortest total computation time min Σt ci The method comprises the steps of carrying out a first treatment on the surface of the The shortest total computation time is used for improving the resource utilization rate of the edge server; the shortest and longest calculation completion time is to ensure timeliness of task completion.
According to the scheme, the shortest and longest calculation completion time processing method comprises the following steps:
(1) Pressing all off-load tasksPerforming ascending order, calculating average calculation time of task with remaining unassigned calculation resources +.>
Wherein domain represents a set of tasks for which computing resources are not allocated, C remain Representing remaining computing resources;
(2) Comparing in turnAnd->If->Computing resource allocation for task i of all unassigned tasks>Ending the distribution;
otherwise, the first residual maximum time delay is taken outLess than->Is to allocate computing resources for task i
(3) Removing the unallocated task from task i, and updating C remain Recalculating
(4) Repeating (2) - (3) until all task assignments are completed.
According to the scheme, the shortest total computation time processing method comprises the following steps: by analyzing the data parameters and the computing resource limitation of the task, and adding the residual maximum time delay limitation condition, the allocation one can be obtained:and (2) distribution II:the method comprises the following specific steps:
(1) Arranging all calculation tasks in ascending order according to the residual maximum time delay;
(2) Pre-distributing all unassigned tasks once according to the distribution, and calculating the calculation time T of each task according to the pre-distribution i pre ；
(3) If it isThe pre-allocation result is used as the calculation resource allocation of the unallocated task, and the allocation is ended;
(4) Otherwise, the first residual maximum time delay is taken outLess than its T i pre According to the allocation II, allocating computing resources for the task i;
(5) Removing the unallocated task from task i, and updating C remain The pre-allocation is carried out again, and the pre-allocation calculation time T is recalculated i pre ；
(6) Repeating (3) - (5) until the allocation of computing resources for all tasks is complete.
According to the above scheme, the allocation of the computing resources of each task on the edge server in the step 4.3) adopts the following method:
the computational resource constraints are:
∑C i ≤C
wherein each task is assigned a computing resource C on the edge server i (i＝1,2,...n)。
The optimization target is min sigma delta T with the task completion calculation time exceeding the sum of the residual maximum time delays ci The method comprises the steps of carrying out a first treatment on the surface of the Wherein DeltaT ci Is the amount by which task i (i=1, 2,., n) completes the calculation time exceeding the remaining maximum delay:
in the method, in the process of the invention,is the calculation completion time of task i. To achieve this goal, this problem may be equivalent to min Σt ci At the same time satisfy->The only difference between this and the shortest calculation time in step S4.2 is +.>Become->
According to the scheme, the sum processing method for the shortest task completion calculation time exceeding the residual maximum time delay comprises the following steps: by analyzing the data parameters and the computing resource limitation of the task, and adding the remaining maximum time delay condition, the allocation one can be obtained:and (2) distribution II: />The method comprises the following specific steps:
(1) Arranging all calculation tasks in ascending order according to the residual maximum time delay;
(2) Pre-distributing all unassigned tasks once according to the distribution, and calculating the calculation time T of each task according to the pre-distribution i pre ；
(3) If it isThe pre-allocation result is used as the calculation resource allocation of the unallocated task, and the allocation is ended;
(4) Otherwise, the first residual maximum time delay is taken outGreater than T thereof i pre According to the allocation II, allocating computing resources for the task i;
(5) Removing the unallocated task from task i, and updating C remain The pre-allocation is carried out again, and the pre-allocation calculation time T is recalculated i pre ；
(6) Repeating (3) - (5) until the allocation of computing resources for all tasks is complete.
According to the above scheme, when the maximum number of tasks m that the edge server can simultaneously execute in the step 5) is smaller than the number of tasks n to be offloaded, the computing completion time of the tasks i (i=1, 2..n) on the edge server is offloadedNot equal to its calculation time T ci ，
The computing resource limitation conditions are set as follows:
any time Sigma C i ≤C(i∈S)
Wherein S represents a task set processed simultaneously;
the optimization objective comprises the sum of the maximum calculation completion time and the calculation time of the shortest all tasks and the remaining maximum time delayAnd the sum of the shortest all task calculation time and the calculation completion time exceeding the remaining maximum delay +.>Wherein->Representing task i (i=1, 2.,.. n) start calculating time.
According to the scheme, the processing method for optimizing the sum of the maximum calculation completion time and the calculation time exceeding the residual maximum time delay of the target for shortest all tasks is as follows:
(1) Placing all tasks in a priority queue according to priority, wherein the shorter the maximum time delay of the tasks is, the higher the priority is;
(2) Taking out the first m tasks in the priority queue;
(3) Calculate the m tasksIf->Then computing resources are allocated to the m tasks using a shortest longest computation completion time processing method;
the shortest and longest calculation completion time processing method comprises the following steps:
3.1.1 Any of all unloadedPressing the businessPerforming ascending order, calculating average calculation time of task with remaining unassigned calculation resources +.>
Wherein domain represents a set of tasks for which computing resources are not allocated, C remain Representing remaining computing resources;
3.1.2 In turn)And->If->Computing resource allocation for task i of all unassigned tasks>Ending the distribution;
otherwise, the first residual maximum time delay is taken outLess than->Is to allocate computing resources for task i
3.1.3 Removing task i from unassigned task, updating C remain Recalculating
3.1.4 Repeating 3.1.2) -3.1.3) until all task assignments are completed;
otherwise, using the processing method that the shortest task completion calculation time exceeds the sum of the remaining maximum delays to allocate calculation resources for the m tasks, wherein the maximum calculation completion time of the tasks is T max ；
The processing method for the sum of the shortest task completion calculation time exceeding the residual maximum time delay is as follows:
by analyzing the data parameters and the computing resource limitation of the task, and adding the remaining maximum time delay condition, the allocation one can be obtained:and (2) distribution II: />The method comprises the following specific steps:
3.2.1 Arranging all calculation tasks according to the ascending order of the residual maximum time delay;
3.2.2 Pre-distributing all unassigned tasks once according to distribution, and calculating the calculation time T of each task according to the pre-distribution i pre ；
3.2.3 If (1)The pre-allocation result is used as the calculation resource allocation of the unallocated task, and the allocation is ended;
3.2.4 Otherwise, take out the first remaining maximum time delayGreater than T thereof i pre According to the allocation II, allocating computing resources for the task i;
3.2.5 Removing task i from unassigned task, updating C remain The pre-allocation is carried out again, and the pre-allocation calculation time T is recalculated i pre ；
3.2.6 Repeating 3.2.3) -3.2.5) until the allocation of computing resources for all tasks is complete;
(4) Not reaching T max When the calculation of one task is completed, another task is taken out of the queue and then the calculation resource of the task is processed;
(5) At the arrival T max When the method is used, for a task being calculated and a task to be calculated, calculating the residual data quantity and the residual maximum time delay of the task to be calculated;
(6) Then based on the remaining amount of tasks and the remaining maximum time delay of these tasks, using step (3) to re-allocate and re-calculate the maximum end time T max ；
(7) Repeating (4) - (6) until the queue is empty.
According to the scheme, the sum processing method for calculating the total time and the calculation completion time of the shortest all tasks and exceeding the residual maximum time delay comprises the following steps:
(1) Placing all tasks in a priority queue according to priority, wherein the shorter the maximum time delay of the tasks is, the higher the priority is;
(2) Taking out the first m tasks in the priority queue;
(3) Calculate the m tasks fetchedIf->Then computing resources are allocated to the m tasks using a shortest total computation time processing method;
the shortest total computation time processing method comprises the following steps: by analyzing the data parameters and the computing resource limitation of the task, and adding the residual maximum time delay limitation condition, the allocation one can be obtained:and (2) distribution II: />The method comprises the following specific steps:
3.1.1 Arranging all calculation tasks according to the ascending order of the residual maximum time delay;
3.1.2 Pre-distributing all unassigned tasks once according to distribution, and calculating the calculation time T of each task according to the pre-distribution i pre ；
3.1.3 If (1)The pre-allocation result is used as the calculation resource allocation of the unallocated task, and the allocation is ended;
3.1.4 Otherwise, take out the first remaining maximum time delayLess than its T i pre According to the allocation II, allocating computing resources for the task i;
3.1.5 Removing task i from unassigned task, updating C remain The pre-allocation is carried out again, and the pre-allocation calculation time T is recalculated i pre ；
3.1.6 Repeating 3.1.3) -3.1.5) until the allocation of computing resources for all tasks is complete.
Otherwise, using the processing method that the shortest task completion calculation time exceeds the sum of the remaining maximum delays to allocate calculation resources for the m tasks, wherein the maximum calculation completion time of the tasks is T max ；
The processing method for the sum of the shortest task completion calculation time exceeding the residual maximum time delay is as follows:
by analyzing the data parameters and the computing resource limitation of the task, and adding the remaining maximum time delay condition, the allocation one can be obtained:and (2) distribution II: />The method comprises the following specific steps:
3.2.1 Arranging all calculation tasks according to the ascending order of the residual maximum time delay;
3.2.2 Pre-distributing all unassigned tasks once according to distribution, and calculating the calculation time T of each task according to the pre-distribution i pre ；
3.2.3 If (1)The pre-allocation result is used as the calculation resource allocation of the unallocated task, and the allocation is ended;
3.2.4 Otherwise, take out the first remaining maximum time delayGreater than T thereof i pre According to the allocation II, allocating computing resources for the task i;
3.2.5 Removing task i from unassigned task, updating C remain The pre-allocation is carried out again, and the pre-allocation calculation time T is recalculated i pre ；
3.2.6 Repeating 3.2.3) -3.2.5) until the allocation of computing resources for all tasks is complete;
(4) Not reaching T max When the calculation of one task is completed, another task is taken out of the queue and then the calculation resource of the task is processed;
(5) At the arrival T max When the method is used, for a task being calculated and a task to be calculated, calculating the residual data quantity and the residual maximum time delay of the task to be calculated;
(6) Then based on the remaining amount of tasks and the remaining maximum time delay of these tasks, using step (3) to re-allocate and re-calculate the maximum end time T max ；
(7) Repeating (4) - (6) until the queue is empty.
The invention has the beneficial effects that:
1. the multi-task computing resource allocation under different edge computing scenes is realized, the timely processing of the task unloading is ensured, and the total time delay or the sum exceeding the maximum time delay in the task unloading process is reduced;
2. the edge computing scene is divided into two cases that the number of the unloading tasks is smaller than the maximum execution number of the edge servers and the number of the unloading tasks is larger than the maximum execution number of the edge servers. For the former, aiming at the condition of sufficient and insufficient system computing resources, the shortest total computing time and the shortest longest finishing computing time and the shortest sum and the optimal three distribution algorithms exceeding the maximum time delay are respectively realized. For the latter, reasonable scheduling of tasks and computing resource allocation are achieved using a multi-round based reassignment strategy.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a flow chart of a method of an embodiment of the present invention.
Fig. 2 is a system configuration block diagram of an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the following examples in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
As shown in fig. 1 and 2, a method for allocating edge computing resources facing multitasking includes the following steps:
s1) constructing an edge computing network scene according to an edge server and a mobile terminal with a task to be processed by unloading computing to the edge server, and obtaining data parameters of the mobile terminal task and performance parameters of the edge server according to the scene;
the data parameters of the mobile terminal tasks comprise the number n of the mobile terminals, the data quantity of each task and the maximum time delay of each task; the mobile terminals are numbered 1 to n, and the data volume is { D }, respectively 1 ,D 2 ,…,D n Maximum delays of respectively
The performance parameters of the edge server comprise computing capacity C of the edge server and the maximum number m of tasks which can be executed simultaneously;
s2) distributing channel resources according to the data parameters of the task unloading on the mobile terminal, and calculating the transmission time delay T of the task i according to the channel resources i trans Determining the remaining maximum delay limit for each task iWherein i=1, 2,..n;
the specific steps of step S2 are as follows:
step S2.1: let the bandwidth of the wireless communication link for transmitting data be B, B i In order to obtain the communication bandwidth of the task i in the transmission process, the communication bandwidth of the task i in the transmission process is in direct proportion to the task data quantity, so that the task i reaches the edge server end at the same time to start the distribution of computing resources. I.e.
The transmission rate of task i (i=1, 2..n) in the communication channel transmitted by the mobile terminal to the edge calculation server can be expressed by the following formula:
wherein, beta is the channel gain between the ith mobile terminal where the task i is located and the edge server, and beta is a random independent co-distributed variable; p (P) i Transmitting power N provided when transmitting task i to edge server for mobile terminal where ith task is located 0 Is the power of the noise in the channel.
Because the transmitting power of the edge server is larger, the data transmission rate in the process that the edge server transmits the calculation result back to the mobile terminal is larger, and the consumed time is negligible for the sake of simplification. The transmission delay of the communication consumption of task i during the entire computation offload is therefore:
step S2.2: for each task, the remaining maximum latency of each task i (i=1, 2..n), i.e. the maximum latency of completing the processing on the edge server, is calculated
S3) comparing whether the maximum number m of tasks which can be executed by the edge server simultaneously is smaller than the number n of unloading tasks, if m is not smaller than n, executing the step S4), and if m is smaller than n, executing the step S5);
s4) calculating minimum calculation resources required by the unloading task under the residual maximum time delay, and distributing the calculation resources for different unloading tasks according to the size relation between the edge server calculation resources and the minimum calculation resources required by the task completion and different optimization targets;
in this scenario, i.e. when the maximum number of tasks m that the edge server can simultaneously execute is not less than the number of offloaded tasks n, the computation completion time of the offloaded tasks i (i=1, 2..n) on the edge serverEqual to its calculation time T ci 。
Step S4.1: the minimum computing resource required for completing the computing unloading task under the residual maximum time delay is as follows
Step S4.2: if it isMeaning that all tasks can be processed within the remaining maximum latency, i.e., the edge server computing resources are sufficient.
According to the duty ratio of the data volume of each task in the data volume of all tasks, according to different optimization targets and different allocation algorithms, the duty ratio of the allocation computing resources of each task is obtained, and the actual limitation is carried out, so that the limitation on the computing resources of each task on an edge server is realized.
The computational resource constraints are:
∑C i ≤C
wherein each task is assigned a computing resource C on the edge server i (i=1, 2, n), the calculation completion time of each task is
The optimization objective includes the shortest and longest calculation completion timesMinimum total computation time min Σt ci . The shortest total computation time is used for improving the resource utilization rate of the edge server; the shortest and longest execution time is to ensure timeliness of task completion. We select one of the optimization objectives as needed.
The shortest and longest calculation completion time processing method comprises the following steps:
(1) Pressing all off-load tasksPerforming ascending order, calculating average calculation time of task with remaining unassigned calculation resources +.>
Wherein domain represents a set of tasks for which computing resources are not allocated, C remain Representing the remaining computing resources.
Comparing in turnAnd->
(2) If it isComputing resource allocation for task i of all unassigned tasks>Ending the distribution;
(3) Otherwise, the first residual maximum time delay is taken outLess than->Is to allocate computing resources for task i
(4) Removing the unallocated task from task i, and updating C remain Recalculating
(5) Repeating (2) - (4) until all task assignments are completed;
preferably, the shortest total computation time processing method includes: by analyzing the data parameters and computing resource limitation of the task and adding the residual maximum time delay limitation conditionTo obtain an allocation of one:and (2) distribution II:the method comprises the following specific steps:
(1) Arranging all calculation tasks in ascending order according to the residual maximum time delay;
(2) Pre-distributing all unassigned tasks once according to the distribution, and calculating the calculation time T of each task according to the pre-distribution i pre ；
(3) If it isThe pre-allocation result is used as the calculation resource allocation of the unallocated task, and the allocation is ended;
(4) Otherwise, the first residual maximum time delay is taken outLess than its T i pre According to the allocation II, allocating computing resources for the task i;
(5) Removing the unallocated task from task i, and updating C remain The pre-allocation is carried out again, and the pre-allocation calculation time T is recalculated i pre ；
(6) Repeating (3) - (5) until the allocation of computing resources for all tasks is complete.
Step S4.3: if it isIndicating that the processing latency of one or more tasks exceeds the remaining maximum completion latency, i.e., the computational resources are deficient.
At this time, the resource allocation policy adopted is as follows:
the computational resource constraints are:
∑C i ≤C
wherein each task is assigned a computing resource C on the edge server i (i＝1,2,...n)。
Optimization targets are min ΣΔt, the shortest task completion calculation time exceeding the sum of the remaining maximum delays ci Wherein DeltaT ci Is the amount by which task i (i=1, 2,., n) completes the calculation time exceeding the remaining maximum delay:
in the method, in the process of the invention,is the calculation completion time of task i. To achieve this goal, this problem may be equivalent to min Σt ci At the same time satisfy->
The only difference between this and the shortest calculation time in step S4.2 is in the calculation resource constraintBecome->
The processing method for the sum of the shortest task completion calculation time exceeding the remaining maximum time delay comprises the following steps: by analyzing the data parameters and the computing resource limitation of the task, and adding the remaining maximum time delay condition, the allocation one can be obtained:and (2) distribution II: />The method comprises the following specific steps:
(1) Arranging all calculation tasks in ascending order according to the residual maximum time delay;
(2) Pre-distributing all unassigned tasks once according to distribution, and calculating calculation of each task according to pre-distributionTime T i pre ；
(3) If it isThe pre-allocation result is used as the calculation resource allocation of the unallocated task, and the allocation is ended;
(4) Otherwise, the first residual maximum time delay is taken outGreater than T thereof i pre According to the allocation II, allocating computing resources for the task i;
(5) Removing the unallocated task from task i, and updating C iemain The pre-allocation is carried out again, and the pre-allocation calculation time T is recalculated i pre ；
(6) Repeating (3) - (5) until the allocation of computing resources for all tasks is complete.
S5) sequencing the tasks according to the set optimization target, determining the starting execution time of each task, and allocating computing resources for each task by combining the resource allocation strategy in the last step;
when the maximum number of tasks m that the edge server can simultaneously execute in step S5) is smaller than the number of tasks to be offloaded n, the calculation completion time of the tasks to be offloaded i (i=1, 2..n) on the edge serverNot equal to its calculation time T ci
The computational resource constraints are:
any time Sigma C i ≤C(i∈S)
Where S represents a set of tasks processed simultaneously.
The optimization objective includes the sum of the maximum computation completion time of the shortest all tasks and the computation time exceeding the remaining maximum time delayThe shortest total time of all task calculation and the calculation completion time exceed the residual maximum time delaySum->Wherein->The start calculation time of task i (i=1, 2, …, n) is represented.
The sum processing method for the maximum calculation completion time and the calculation time of the shortest all tasks exceeding the residual maximum time delay comprises the following steps:
(1) Placing all tasks in a priority queue according to priority, wherein the shorter the maximum time delay of the tasks is, the higher the priority is;
(2) Taking out the first m tasks;
(3) Calculate the m tasks fetchedIf->Then computing resources are allocated to the m tasks using the shortest longest computation completion time processing method of step S4.2; otherwise, using the processing method that the shortest task completion calculation time in the step S4.3 exceeds the sum of the remaining maximum time delays to allocate calculation resources for the m tasks, wherein the maximum calculation completion time of the tasks is T max ；
(4) Not reaching T max When the calculation of one task is completed, another task is taken out of the queue and then the calculation resource of the task is processed;
(5) At the arrival T max When the method is used, for a task being calculated and a task to be calculated, calculating the residual data quantity and the residual maximum time delay of the task to be calculated;
(6) Then based on the remaining amount of tasks and the remaining maximum time delay of these tasks, using step (3) to re-allocate and re-calculate the maximum end time T max ；
(7) Repeating (4) - (6) until the queue is empty;
the sum processing method for calculating the total time and the calculation completion time of the shortest all tasks to exceed the residual maximum time delay comprises the following steps:
(1) Placing all tasks in a priority queue according to priority, wherein the shorter the maximum time delay of the tasks is, the higher the priority is;
(2) Taking out the first m tasks;
(3) Calculate the m tasks fetchedIf->Then computing resources are allocated to the m tasks using the shortest total computation time processing method of step S4.2; otherwise, using the processing method that the shortest task completion calculation time in the step S4.3 exceeds the sum of the remaining maximum time delays to allocate calculation resources for the m tasks, wherein the maximum calculation completion time of the tasks is T max ；
(4) Not reaching T max When the calculation of one task is completed, another task is taken out of the queue and then the calculation resource of the task is processed;
(5) At the arrival T max When the method is used, for a task being calculated and a task to be calculated, calculating the residual data quantity and the residual maximum time delay of the task to be calculated;
(6) Then based on the remaining amount of tasks and the remaining maximum time delay of these tasks, using step (3) to re-allocate and re-calculate the maximum end time T max ；
(7) Repeating (4) - (6) until the queue is empty;
the sum processing method for calculating the total time and the calculation completion time of the shortest all tasks to exceed the residual maximum time delay comprises the following steps:
the method is similar to the processing method that the maximum calculation completion time and the calculation time of the shortest all tasks exceed the sum of the residual maximum time delay, and only the following steps (3) are needed: if it isThen at the shortest longest calculation time using step S4.2The reason is that the allocation of computing resources for these m tasks is changed to if +.>The computing resources are allocated to the m tasks using the shortest total computation time processing method of step S4.2, while the other steps are unchanged.
It will be understood that modifications and variations will be apparent to those skilled in the art from the foregoing description, and it is intended that all such modifications and variations be included within the scope of the following claims.

Claims (7)

1. A multitasking oriented edge computing resource allocation method, comprising the steps of:
1) Constructing an edge computing network scene according to the edge server and the mobile terminal with a task to be computed and unloaded to the edge server for processing, and obtaining data parameters of the mobile terminal task and performance parameters of the edge server according to the scene;
the data parameters of the mobile terminal tasks comprise the number n of the mobile terminals, the data quantity of each task and the maximum time delay of each task; the mobile terminals are numbered 1 to n, and the data volume is { D }, respectively 1 ,D 2 ,…,D n Maximum delays of respectively
The performance parameters of the edge server comprise computing capacity C of the edge server and the maximum number m of tasks which can be executed simultaneously;
2) Allocating channel resources according to data parameters of task unloading on the mobile terminal, and calculating the transmission delay of task i according to the channel resourcesDetermining the remaining maximum delay limit for each task i>Wherein i=1, 2,..n;
3) Comparing whether the maximum number of tasks which can be executed by the edge server is smaller than the number of unloading tasks n, if m is not smaller than n, executing the step 4), and if m is smaller than n, executing the step 5);
4) Calculating minimum calculation resources required by the unloading tasks under the residual maximum time delay, and distributing calculation resources for different unloading tasks according to the size relation between the edge server calculation resources and the minimum calculation resources required by the task completion;
the specific steps in the step 4) are as follows:
in this scenario, the maximum number of tasks that the edge server can simultaneously execute m is not less than the number of offload tasks n, offload tasks i, i=1, 2,..n, the computation completion time on the edge serverEqual to its calculation time T ci ；
4.1 The minimum computing resource required for completing the unloading task under the residual maximum time delay is calculated as
4.2 If (1)Indicating that all tasks are processed within the residual maximum time delay, namely that the computing resources of the edge server are sufficient, and distributing the computing resources of each task on the edge server according to the set optimization target;
the computing resources of each task on the edge server are allocated in the step 4.2) by adopting the following method:
the computing resource limitation conditions are set as follows:
∑C i ≤C
wherein each task is assigned a computing resource C on the edge server i I=1, 2,..n, the calculated completion time for each task is
Setting optimization targets including shortest and longest calculation completion timesOr the shortest total computation time min Σt ci The method comprises the steps of carrying out a first treatment on the surface of the The shortest total computation time is used for improving the resource utilization rate of the edge server; the shortest and longest calculation completion time is used for guaranteeing timeliness of task completion;
4.3 If (1)The allocation computing resource duty ratio of each task is obtained according to different allocation algorithms according to the set optimization targets and the actual limitation is carried out, so that the limitation on the computing resources of each task on the edge server is realized;
5) Reasonably scheduling the tasks, determining the starting execution time of each task, and distributing computing resources for each task according to the set optimization target and combining computing resources according to the edge server and minimum computing resources required by completing the tasks;
when the maximum number m of tasks which can be executed by the edge server in the step 5) is smaller than the number n of tasks to be offloaded, the calculation completion time of the task i to be offloaded on the edge server is calculatedNot equal to its calculation time T ci Setting a computational resource limitThe preparation conditions are as follows:
any time Sigma C i ≤C；i∈S；
Wherein S represents a task set processed simultaneously;
the optimization objective comprises the sum of the maximum calculation completion time and the calculation time of the shortest all tasks and the remaining maximum time delayAnd the sum of the shortest all task calculation time and the calculation completion time exceeding the remaining maximum delay +.>Wherein->Indicating the time at which task i begins to calculate.
2. The method for allocating edge computing resources facing multitasking according to claim 1, wherein when the optimization objective is the shortest longest computing completion time, the task allocation processing method is as follows:
(1) Pressing all off-load tasksPerforming ascending order, calculating average calculation time of task with remaining unassigned calculation resources +.>
Wherein domain represents a set of tasks for which computing resources are not allocated, C remain Representing remaining computing resources;
(2) Comparing in turnAnd->If->Allocating computing resources for task i of all unallocated tasksEnding the distribution;
otherwise, the first residual maximum time delay is taken outLess than->Is to allocate computing resources for task i
(3) Removing the unallocated task from task i, and updating C remain Recalculating
(4) Repeating (2) - (3) until all task assignments are completed.
3. The method for assigning multi-tasking oriented edge computing resources according to claim 1, wherein when the optimization objective is the shortest total computation time, the task assignment processing method is as follows:
the data parameters and the computing resource limit of the task are analyzed, and the remaining maximum time delay limit condition is added, so that an allocation one is obtained:and (2) distribution II: />The method comprises the following specific steps:
(1) Arranging all calculation tasks in ascending order according to the residual maximum time delay;
(2) Pre-distributing all unassigned tasks once according to distribution, and calculating calculation time of each task according to pre-distribution
(3) If it isThe pre-allocation result is used as the calculation resource allocation of the unallocated task, and the allocation is ended;
(4) Otherwise, the first residual maximum time delay is taken outLess than->According to the allocation II, allocating computing resources for the task i;
(5) Removing the unallocated task from task i, and updating C remain The pre-allocation is carried out again, and the pre-allocation calculation time is recalculated
(6) Repeating (3) - (5) until the allocation of computing resources for all tasks is complete.
4. The method for assigning computing resources on an edge server according to claim 1, wherein the assigning computing resources on an edge server for each task in step 4.3) adopts the following method:
the computational resource constraints are:
∑C i ≤C
wherein each task is assigned a computing resource C on the edge server i ,i＝1,2,...n。
5. The method for computing resource allocation for edges in multiple tasks as claimed in claim 4 wherein the optimization objective is to minimize min ΣΔt by which the task completion computation time exceeds the sum of the remaining maximum delays ci The method comprises the steps of carrying out a first treatment on the surface of the Wherein DeltaT c Is the amount by which task i completes the computation time exceeding the remaining maximum delay, i=1, 2,..n;
in the method, in the process of the invention,is the calculation completion time of task i, to achieve this goal, this problem is equivalent to min Σt ci At the same time satisfy->
The processing method for the sum of the shortest task completion calculation time exceeding the remaining maximum time delay comprises the following steps: the data parameters and the computing resource limit of the task are analyzed, and the remaining maximum time delay condition is added, so that an allocation one is obtained:and (2) distribution II:the method comprises the following specific steps:
(1) Arranging all calculation tasks in ascending order according to the residual maximum time delay;
(2) Pre-distributing all unassigned tasks once according to distribution, and calculating calculation time of each task according to pre-distribution
(3) If it isThe pre-allocation result is used as the calculation resource allocation of the unallocated task, and the allocation is ended;
(4) Otherwise, the first residual maximum time delay is taken outGreater than->According to the allocation II, allocating computing resources for the task i;
(5) Removing the unallocated task from task i, and updating C remain The pre-allocation is carried out again, and the pre-allocation calculation time is recalculated
(6) Repeating (3) - (5) until the allocation of computing resources for all tasks is complete.
6. The method for assigning multi-task oriented edge computing resources according to claim 1, wherein the optimization objective in step 5) is that the maximum computing completion time and computing time of the shortest all tasks exceed the sum of the remaining maximum delays, and the processing method is as follows:
(1) Placing all tasks in a priority queue according to priority, wherein the shorter the maximum time delay of the tasks is, the higher the priority is;
(2) Taking out the first m tasks in the priority queue;
(3) Calculate the m tasks
The minimum computational resource required to complete the task off-load with the maximum latency remaining is +.>
If it isThen computing resources are allocated to the m tasks using a shortest longest computation completion time processing method;
the shortest and longest calculation completion time processing method comprises the following steps:
3.1.1 All off-load tasks as perPerforming ascending order, calculating average calculation time of task with remaining unassigned calculation resources +.>
Wherein domain represents a set of tasks for which computing resources are not allocated, C remain Representing remaining computing resources;
3.1.2 In turn)And->If->Allocating computing resources for task i of all unallocated tasksEnding the distribution;
otherwise, the first residual maximum time delay is taken outLess than->Is to allocate computing resources for task i
3.1.3 Removing task i from unassigned task, updating C remain Recalculating
3.1.4 Repeating 3.1.2) -3.1.3) until all task assignments are completed;
otherwise, using the processing method that the shortest task completion calculation time exceeds the sum of the remaining maximum delays to allocate calculation resources for the m tasks, wherein the maximum calculation completion time of the tasks is T max ；
The processing method for the sum of the shortest task completion calculation time exceeding the residual maximum time delay is as follows:
the data parameters and the computing resource limit of the task are analyzed, and the remaining maximum time delay condition is added, so that an allocation one is obtained:and (2) distribution II: />The method comprises the following specific steps:
3.2.1 Arranging all calculation tasks according to the ascending order of the residual maximum time delay;
3.2.2 Pre-distributing all unassigned tasks once according to distribution, and calculating the calculation time of each task according to the pre-distribution
3.2.3 If (1)The pre-allocation result is used as the calculation resource allocation of the unallocated task, and the allocation is ended;
3.2.4 Otherwise, take out the first remaining maximum time delayGreater than->According to the allocation II, allocating computing resources for the task i;
3.2.5 Removing task i from unassigned task, updating C aremain The pre-allocation is carried out again, and the pre-allocation calculation time is recalculated
3.2.6 Repeating 3.2.3) -3.2.5) until the allocation of computing resources for all tasks is complete;
(4) Not reaching T max When the calculation of one task is completed, another task is taken out of the queue and then the calculation resource of the task is processed;
(5) At the arrival T max When the method is used, for a task being calculated and a task to be calculated, calculating the residual data quantity and the residual maximum time delay of the task to be calculated;
(6) Based on the remaining amount of tasks and the remaining maximum time delay for these tasks,reassigning and recalculating the maximum end time T using step (3) max ；
(7) Repeating (4) - (6) until the queue is empty.
7. The method for assigning multi-task oriented edge computing resources according to claim 1, wherein when the optimization objective in the step 5) is that the sum of the total computing time and the computing completion time of the shortest all tasks exceeds the sum of the remaining maximum delays, the processing method comprises:
(1) Placing all tasks in a priority queue according to priority, wherein the shorter the maximum time delay of the tasks is, the higher the priority is;
(2) Taking out the first m tasks in the priority queue;
(3) Calculate the m tasks fetchedIf->Then computing resources are allocated to the m tasks using a shortest total computation time processing method;
the shortest total computation time processing method comprises the following steps: the data parameters and the computing resource limit of the task are analyzed, and the remaining maximum time delay limit condition is added, so that an allocation one is obtained:and (2) distribution II: />The method comprises the following specific steps:
3.1.1 Arranging all calculation tasks according to the ascending order of the residual maximum time delay;
3.1.2 Pre-distributing all unassigned tasks once according to distribution, and calculating the calculation time of each task according to the pre-distribution
3.1.3 If (1)The pre-allocation result is used as the calculation resource allocation of the unallocated task, and the allocation is ended;
3.1.4 Otherwise, take out the first remaining maximum time delayLess than->According to the allocation II, allocating computing resources for the task i;
3.1.5 Removing task i from unassigned task, updating C remain The pre-allocation is carried out again, and the pre-allocation calculation time is recalculated
3.1.6 Repeating 3.1.3) -3.1.5) until the allocation of computing resources for all tasks is complete;
otherwise, using the processing method that the shortest task completion calculation time exceeds the sum of the remaining maximum delays to allocate calculation resources for the m tasks, wherein the maximum calculation completion time of the tasks is T max ；
The processing method for the sum of the shortest task completion calculation time exceeding the residual maximum time delay is as follows:
the data parameters and the computing resource limit of the task are analyzed, and the remaining maximum time delay condition is added, so that an allocation one is obtained:and (2) distribution II: />The method comprises the following specific steps:
3.2.1 Arranging all calculation tasks according to the ascending order of the residual maximum time delay;
3.2.2 Pre-distributing all unassigned tasks once according to distribution, and calculating the calculation time of each task according to the pre-distribution
3.2.3 If (1)The pre-allocation result is used as the calculation resource allocation of the unallocated task, and the allocation is ended;
3.2.4 Otherwise, take out the first remaining maximum time delayGreater than->According to the allocation II, allocating computing resources for the task i;
3.2.5 Removing task i from unassigned task, updating C remain The pre-allocation is carried out again, and the pre-allocation calculation time is recalculated
3.2.6 Repeating 3.2.3) -3.2.5) until the allocation of computing resources for all tasks is complete;
(4) Not reaching T max When the calculation of one task is completed, another task is taken out of the queue and then the calculation resource of the task is processed;
(5) At the arrival T max When the method is used, for a task being calculated and a task to be calculated, calculating the residual data quantity and the residual maximum time delay of the task to be calculated;
(6) Then based on the remaining amount of tasks and the remaining maximum time delay of these tasks, using step (3) to re-allocate and re-calculate the maximum end time T max ；
(7) Repeating (4) - (6) until the queue is empty.
CN202010172395.9A 2020-03-12 2020-03-12 Multitasking-oriented edge computing resource allocation method Active CN111694655B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010172395.9A CN111694655B (en) 2020-03-12 2020-03-12 Multitasking-oriented edge computing resource allocation method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010172395.9A CN111694655B (en) 2020-03-12 2020-03-12 Multitasking-oriented edge computing resource allocation method

Publications (2)

Publication Number Publication Date
CN111694655A CN111694655A (en) 2020-09-22
CN111694655B true CN111694655B (en) 2023-08-29

Family

ID=72476269

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010172395.9A Active CN111694655B (en) 2020-03-12 2020-03-12 Multitasking-oriented edge computing resource allocation method

Country Status (1)

Country Link
CN (1) CN111694655B (en)

Families Citing this family (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112130999B (en) * 2020-09-23 2024-02-13 南方电网科学研究院有限责任公司 Electric power heterogeneous data processing method based on edge calculation
CN112181658B (en) * 2020-09-30 2024-04-05 南京工程学院 Calculation task allocation method for maximizing network benefits in heterogeneous network
CN112312325B (en) * 2020-10-29 2022-08-16 陕西师范大学 Mobile edge task unloading method based on three decision models
CN112256420B (en) * 2020-10-30 2022-12-02 重庆紫光华山智安科技有限公司 Task allocation method and device and electronic equipment
CN112612549B (en) * 2020-12-30 2022-06-24 润联软件***(深圳)有限公司 Multi-edge service task selection unloading method and device and related equipment
CN113032120B (en) * 2021-03-26 2023-06-30 重庆大学 Industrial field big data task cooperative scheduling method based on edge calculation
CN113344152A (en) * 2021-04-30 2021-09-03 华中农业大学 System and method for intelligently detecting and uploading full-chain production information of dairy products
CN113406929B (en) * 2021-05-02 2023-03-10 华中农业大学 Food safety big data intelligent acquisition and application system
CN113472842B (en) * 2021-05-24 2023-01-10 北京邮电大学 User state perception method in mobile edge computing network and related equipment
CN113326126B (en) * 2021-05-28 2024-04-05 湘潭大学 Task processing method, task scheduling method, device and computer equipment
CN114020447B (en) * 2021-09-29 2024-04-12 中通服咨询设计研究院有限公司 Method for distributing server resources for processing traffic flow big data
CN114844555B (en) * 2022-04-15 2023-06-20 中国电子科技集团公司第五十四研究所 Cooperative processing method for unmanned aerial vehicle multitasking execution resource limitation
CN115665805A (en) * 2022-12-06 2023-01-31 北京交通大学 Point cloud analysis task-oriented edge computing resource scheduling method and system
CN116483573B (en) * 2023-04-25 2023-11-03 格莱杰(深圳)科技有限公司 Computing resource scheduling method and device in response to task to be processed
CN117591302B (en) * 2024-01-18 2024-06-18 中国电子科技集团公司第十五研究所 Project resource optimization allocation method based on unconstrained optimization algorithm

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109918201A (en) * 2019-03-05 2019-06-21 中国联合网络通信集团有限公司 The control method and system of task unloading
CN110113195A (en) * 2019-04-26 2019-08-09 山西大学 A kind of method of joint unloading judgement and resource allocation in mobile edge calculations system
CN110351760A (en) * 2019-07-19 2019-10-18 重庆邮电大学 A kind of mobile edge calculations system dynamic task unloading and resource allocation methods

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10440096B2 (en) * 2016-12-28 2019-10-08 Intel IP Corporation Application computation offloading for mobile edge computing
US10659526B2 (en) * 2018-01-03 2020-05-19 Verizon Patent And Licensing Inc. Edge compute systems and methods

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109918201A (en) * 2019-03-05 2019-06-21 中国联合网络通信集团有限公司 The control method and system of task unloading
CN110113195A (en) * 2019-04-26 2019-08-09 山西大学 A kind of method of joint unloading judgement and resource allocation in mobile edge calculations system
CN110351760A (en) * 2019-07-19 2019-10-18 重庆邮电大学 A kind of mobile edge calculations system dynamic task unloading and resource allocation methods

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
代美玲等.基于终端能耗和***时延最小化的边缘计算卸载及资源分配机制.《电子与信息学报》.2019,全文. *

Also Published As

Publication number Publication date
CN111694655A (en) 2020-09-22

Similar Documents

Publication Publication Date Title
CN111694655B (en) Multitasking-oriented edge computing resource allocation method
CN110099384B (en) Multi-user multi-MEC task unloading resource scheduling method based on edge-end cooperation
US11032212B2 (en) Systems and methods for provision of a guaranteed batch
CN105718317A (en) Task scheduling method and task scheduling device
CN111176840B (en) Distribution optimization method and device for distributed tasks, storage medium and electronic device
US9178763B2 (en) Weight-based collocation management
CN110543336A (en) Edge calculation task unloading method and device based on non-orthogonal multiple access technology
CN105373429A (en) Task scheduling method, device and system
CN111246586A (en) Method and system for distributing smart grid resources based on genetic algorithm
CN111104211A (en) Task dependency based computation offload method, system, device and medium
CN103455375B (en) Load-monitoring-based hybrid scheduling method under Hadoop cloud platform
CN113992945B (en) Multi-server multi-user video analysis task unloading method based on game theory
CN110996390B (en) Wireless access network computing resource allocation method and network system
CN112771502A (en) Elastic resource pooling for dynamic throughput rebalancing
CN111776896B (en) Elevator dispatching method and device
CN102333375A (en) Method used for guaranteeing CCE (Communication Control Equipment) distribution fairness in LTE (Long Term Evolution) system
CN116708451A (en) Edge cloud cooperative scheduling method and system
CN117032937B (en) Task scheduling method based on GPU, electronic device and storage medium
CN107204930B (en) Token adding method and device
CN111542091B (en) Wireless and computing resource joint allocation method for network slice
CN116302404B (en) Resource decoupling data center-oriented server non-perception calculation scheduling method
CN111353712B (en) Distribution task scheduling method and device, server and storage medium
CN114884957B (en) Method and device for unloading computing tasks in air-space-ground network and electronic equipment
CN114610503A (en) Multi-application online computing unloading resource scheduling method and system based on overall resource requirements of task graph in edge computing
CN115567599A (en) Task migration method, device and equipment for cloud-edge cooperative computing

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