CN113791878A - Distributed task unloading method for deadline perception in edge calculation - Google Patents
Distributed task unloading method for deadline perception in edge calculation Download PDFInfo
- Publication number
- CN113791878A CN113791878A CN202110828570.XA CN202110828570A CN113791878A CN 113791878 A CN113791878 A CN 113791878A CN 202110828570 A CN202110828570 A CN 202110828570A CN 113791878 A CN113791878 A CN 113791878A
- Authority
- CN
- China
- Prior art keywords
- task
- tasks
- deadline
- edge server
- edge
- 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
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000004364 calculation method Methods 0.000 title claims abstract description 17
- 230000008447 perception Effects 0.000 title abstract description 5
- 230000008901 benefit Effects 0.000 claims abstract description 11
- 230000008569 process Effects 0.000 claims description 22
- 230000005540 biological transmission Effects 0.000 claims description 6
- 238000012546 transfer Methods 0.000 claims description 3
- 238000012512 characterization method Methods 0.000 claims description 2
- 238000003780 insertion Methods 0.000 claims description 2
- 230000037431 insertion Effects 0.000 claims description 2
- 230000007246 mechanism Effects 0.000 description 8
- 230000006870 function Effects 0.000 description 7
- 238000005457 optimization Methods 0.000 description 3
- 230000003190 augmentative effect Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- ZXQYGBMAQZUVMI-QQDHXZELSA-N [cyano-(3-phenoxyphenyl)methyl] (1r,3r)-3-[(z)-2-chloro-3,3,3-trifluoroprop-1-enyl]-2,2-dimethylcyclopropane-1-carboxylate Chemical compound CC1(C)[C@@H](\C=C(/Cl)C(F)(F)F)[C@H]1C(=O)OC(C#N)C1=CC=CC(OC=2C=CC=CC=2)=C1 ZXQYGBMAQZUVMI-QQDHXZELSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 230000003116 impacting effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—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
- G06F9/485—Task life-cycle, e.g. stopping, restarting, resuming execution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/44—Arrangements for executing specific programs
- G06F9/445—Program loading or initiating
- G06F9/44594—Unloading
-
- 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
- G06F9/5011—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resources being hardware resources other than CPUs, Servers and Terminals
- G06F9/5022—Mechanisms to release resources
-
- 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
- G06F2209/00—Indexing scheme relating to G06F9/00
- G06F2209/50—Indexing scheme relating to G06F9/50
- G06F2209/509—Offload
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a distributed task unloading method for deadline perception in edge calculation, which comprises the following steps: the mobile user side sends the task unloading request to a plurality of base stations in a signal coverage range; the base station sends the edge server information connected with the base station to the mobile user terminal; the mobile user side distributes the edge servers for the tasks in a distributed manner to calculate the tasks according to the deadline of the tasks, the deadline type of the tasks and the received information of all the edge servers; and the edge server receives the distributed tasks sent by the mobile client, and schedules the tasks in real time according to the types of the deadline dates of the tasks and the states of other tasks running by the edge server so as to maximize the benefit of completing the tasks in the edge computing network. The invention can expand the types of the unloading tasks of the edge computing system; the problems of single-point failure, slow unloading decision and the like caused by a centralized controller are avoided.
Description
Technical Field
The invention relates to the technical field of edge computing, in particular to a distributed task unloading method for deadline perception in edge computing.
Background
With the explosive growth of mobile users, applications sensitive to delay, such as augmented reality, virtual reality, auto-driving, etc., are becoming more popular. Due to limited computing resources on mobile devices, it is often necessary to transfer computing tasks to the cloud for processing. However, unpredictable network delays between the remote cloud and the mobile device often do not meet the requirements of delay-sensitive applications.
To overcome the above problems, edge calculation as a new example has come. Edge computing allows mobile users to offload computing tasks to edge servers at nearby base stations. The tasks that a mobile user offloads to the server typically have a hard expiration date or a soft expiration date. Ideally, all tasks should be completed before the expiration date. However, edge servers have limited capacity compared to cloud servers, so there is no guarantee that all tasks will be completed by the expiration date. When offloading a large number of tasks with different expiration date types to the edge server, an inappropriate task offloading mechanism may significantly reduce the task completion rate, thereby reducing the quality of user experience. Therefore, there is a need to design an efficient task offloading mechanism in an edge computing environment.
In recent years, in order to achieve different optimization goals, various task offloading mechanisms have been proposed to achieve different optimization goals, such as maximizing task completion rate, maximizing task computation efficiency, minimizing task response time, minimizing energy consumption, and the like. Previous studies were primarily directed to tasks with hard deadlines, tasks that missed a hard deadline being simply dropped or offloaded onto the cloud. However, in the real world, not all delay sensitive tasks have to be completed before a certain deadline. Some tasks, such as multimedia tasks, video analytics tasks, have soft expiration dates. These tasks may miss the main expiration date, but still be completed in a timely manner. Furthermore, another key issue with existing task offloading mechanisms is the delay caused by centralized optimization. These task offloading mechanisms typically rely on a centralized controller to determine the computational offloading policy in an offline manner, that is, existing offloading mechanisms assume that the complete information (e.g., task location, task release time, task deadline) for all tasks is known in advance. In fact, since the computational offload problem is np-hard, the centralized controller takes too much time to determine the offload strategy before offloading. Furthermore, in edge computing systems, user arrival and departure tend to be random. Therefore, the tasks they want to offload are also random. However, off-line computational offload methods often fail to handle a series of randomly arriving tasks.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides the distributed task unloading method for the deadline perception in the edge computing, so that the edge computing system can not only process the tasks with hard deadline, but also effectively process the tasks with soft deadline, and the types of the unloading tasks of the edge computing system are expanded; the distributed task scheduling of the edge server is supported, and the problems of single-point failure, slow unloading decision and the like caused by a centralized controller are avoided.
In order to achieve the purpose, the invention adopts the following technical scheme:
an expiration date aware distributed task offloading method in edge computing, the offloading method comprising the steps of:
s1, the mobile client sends the unloading request of the task generated by the delay sensitive application to a plurality of base stations in the signal coverage range, the edge server corresponding to the base station calculates the unloading task, and each base station is connected with at least one edge server;
s2, the base station receiving the task unloading request sends the edge server information connected with the base station to the mobile client, wherein the edge server information comprises the server computing capacity, the data transmission bandwidth and the running state of each task on the server;
s3, the mobile client distributes the edge servers for the task in a distributed manner to calculate the task by combining the task deadline, the task deadline type and all the received edge server information; the task deadline type comprises a hard deadline and a soft deadline, the income of the hard deadline completed by date is greater than a preset income threshold, the loss of the soft deadline incomplete is less than a preset loss threshold, and the income and the loss are calculated and obtained based on the application performance of the delay sensitive application; the constraints of the allocation process are: tasks with hard expiration dates can always be completed;
s4, the edge server receives the distributed tasks sent by the mobile client, and schedules the tasks in real time according to the types of the deadline of the tasks and the states of other tasks running by the edge server so as to maximize the benefit of completing the tasks in the edge computing network;
s5, the edge server completes the calculation process of the task according to the scheduling result of the step S4.
In order to optimize the technical scheme, the specific measures adopted further comprise:
further, in step S1, the ith task T generated by the delay-sensitive applicationiIs a quadruple
Wherein d isiNumber of bits to be transmitted for a task, wiIs the workload of the task or tasks,
indicating the expiration date of the execution of the task,
is the release time of the task or tasks,
as the deadline of the task, IiThe expiration date type of the characterization task.
Further, in step S2, the edge server information includes the ith task TiChannel bandwidth b between corresponding mobile ue and edge server ji，jTransmission power pi，jChannel gain gi，jNoise power woCPU frequency fjAnd the running state of each task in the edge server; the tasks are divided into two types according to the running state of the tasks in the edge server: one is that the server has been off-loaded to the edge server, is performing calculations or is waiting to be counted in a queueAnother type of task is a task that has been decided to be offloaded to the server, which is currently in transit.
Further, in step S3, the process that the mobile client distributively allocates the edge servers for the task to perform task calculation in combination with the expiration date of the task, the expiration date type of the task, and all the received edge server information includes the following steps:
s31, calculating the task T according to the following formulaiUpload time of transfer to edge server j
Wherein r isi，jFor the task transmission rate, it is expressed as:
s32, estimating the task T if the task is to be processed according to the running state of each task in the edge serveriUploading to the edge server j, receiving the task from the edge server j and waiting time of the task in the process of executing the task
S33, calculating the task T according to the following formulaiCompute time on edge server j
S34, calculating the task T according to the following formulaiTotal completion time t on edge server ji，j：
S35, according to t calculated in step S34i，jDetecting task TiWhether it can be completed on the edge server j before the corresponding expiration date, and if so, selecting the task TiUnloading the server to an edge server j, if the server can not be unloaded, traversing all edge servers covering the mobile user in sequence according to the same calculation formula until the edge server allowing unloading is found, and ending the process; if all the edge servers cannot complete the task before the task' S expiration date, go to step S36;
s36, only considering the tasks with hard expiration date in the edge server, recalculating the task waiting time
And total completion time ti，j(ii) a Counting the number of tasks violating the deadline caused by the insertion operation, and assigning the task TiAnd unloading the data to the edge server with the minimum number of tasks violating the deadline, and ending the flow.
Further, in step S4, the process of scheduling the task in real time according to the expiration date type of the task and the status of other tasks that are running by the task includes the following steps:
s41, establishing a task income model according to the task deadline;
s42, dividing the tasks in the edge server into two types, wherein one type is the task which can be before the deadline currently and is marked as { i }cOne category, tasks that cannot currently be completed by the expiration date, is denoted as { i }u}; definition icThe loss function c of the task ini，jThe system is used for representing the loss brought to the system if the execution time of the task i is pushed back, so that the task cannot be completed before the deadline; definition iuGet function r of tasks in }i′，jIs used to indicate if you will want to doThe execution time of the task i' is advanced, so that the task can be completed before the deadline, and benefits are brought to the system;
s43, from { icSelect one with minci，jTask i of (1), let a be minci，jFrom { i ] simultaneouslyuChoose to have maxr ini′，jTask i' of (a), let b be maxri′，j(ii) a If b ≧ a, the execution order of the above tasks is interchanged, i.e., task i' is changed from { i ≧ auRemove it and advance its execution time, get task i from { i }cRemoving the task and pushing the task execution time;
s44, judging { icAnd { i }uWhether all tasks in the sequence are traversed or not, if not, repeating the step S43, otherwise, ending the process.
The invention has the beneficial effects that:
(1) the invention provides the task unloading method for sensing the deadline in the edge calculation for the first time, so that the edge calculation system can process the tasks with the hard deadline and can effectively process the tasks with the soft deadline, and the types of the tasks unloaded by the edge calculation system are expanded.
(2) The task allocation algorithm designed by the invention supports the client to allocate the proper edge server for the task in a distributed manner to carry out calculation. The task scheduling algorithm designed by the invention supports the distributed task scheduling of the edge server. The adopted distributed algorithm avoids the problems of single-point failure, slow unloading decision and the like caused by a centralized controller.
(3) The invention supports real-time task unloading decision, namely, when the invention carries out task unloading decision, the invention does not need to know all information (such as task position, task release time, task deadline and the like) of all tasks in advance, and the real-time unloading decision conforms to the characteristic that a user dynamically sends a task unloading request in an edge computing system.
Drawings
FIG. 1 is a flow diagram of an expiration date-aware distributed task offloading method in edge computing of the present invention.
FIG. 2 is a schematic diagram of an edge computing task offloading scenario of the present invention.
Fig. 3 is a flowchart of the client task allocation algorithm of the present invention.
Fig. 4 is a flowchart of an edge server-side task scheduling algorithm of the present invention.
Fig. 5 is a flowchart of the interaction between the user side and the edge server side according to the present invention.
Detailed Description
The present invention will now be described in further detail with reference to the accompanying drawings.
It should be noted that the terms "upper", "lower", "left", "right", "front", "back", etc. used in the present invention are for clarity of description only, and are not intended to limit the scope of the present invention, and the relative relationship between the terms and the terms is not limited by the technical contents of the essential changes.
With reference to fig. 1, a distributed task offloading method for deadline awareness in edge computing, the offloading method comprising the steps of:
and S1, the mobile user side sends the unloading requests of the tasks generated by the delay sensitive application to a plurality of base stations in a signal coverage range, the edge servers corresponding to the base stations calculate the unloading tasks, and each base station is connected with at least one edge server.
And S2, the base station receiving the task unloading request sends the edge server information connected with the base station to the mobile client, wherein the edge server information comprises the computing capacity of the server, the data transmission bandwidth and the running state of each task on the server.
S3, the mobile client distributes the edge servers for the task in a distributed manner to calculate the task by combining the task deadline, the task deadline type and all the received edge server information; the task deadline type comprises a hard deadline and a soft deadline, the income of the hard deadline completed by date is greater than a preset income threshold, the loss of the soft deadline incomplete is less than a preset loss threshold, and the income and the loss are calculated and obtained based on the application performance of the delay sensitive application; the constraints of the allocation process are: tasks with hard expiration dates can always be completed.
And S4, the edge server receives the distributed tasks sent by the mobile user terminal, and schedules the tasks in real time according to the types of the deadline dates of the tasks and the states of other tasks running by the edge server so as to maximize the benefit of completing the tasks in the edge computing network.
S5, the edge server completes the calculation process of the task according to the scheduling result of the step S4.
The invention provides a method for carrying out task unloading decision in real time aiming at the defects of the existing task unloading mechanism, when a plurality of users carry out task unloading, different users can carry out task allocation decision in a distributed mode according to the type of an unloading task deadline and the information of a peripheral edge server. After receiving the tasks, the edge server also schedules the tasks in real time according to the types of the task deadline dates and the states of other tasks running by the edge server so as to maximize the benefit of completing the tasks.
FIG. 2 is an edge computing task offload scenario of the present invention. In the edge computing system, an edge server with certain storage and computing capabilities is arranged beside each base station, and each base station has a certain signal coverage area. Mobile users within the signal range of the base station can offload the computation tasks to an edge server at the base station for computation. In edge computing systems, where base stations are densely deployed, each mobile user may be within the signal coverage of multiple base stations, and a mobile user may offload tasks to one of the base stations. Furthermore, due to the large number of mobile users, each edge server needs to handle multiple computing tasks. Due to the limited processing power of the edge server, the state of these tasks can be of two types: one is the task the server is executing and the other is the task waiting to be executed in the queuing sequence.
FIG. 1 is a flowchart of an expiration date-aware task offloading mechanism in edge computing according to the present invention, which includes the following steps:
step one, a task which is generated by a mobile user and has the following information: task TiIs a quadruple
Wherein d isiNumber of bits to be transmitted for a task, wiIs the workload of the task or tasks,
indicating the expiration date of the execution of the task,
is the release time of the task or tasks,
as the deadline of the task, IiThe type of task deadline is characterized. In the edge computing environment, tasks can be divided into two types according to the expiration date of the tasks, one type is tasks with hard expiration dates, such as virtual reality and augmented reality tasks, the tasks must be completed strictly before the expiration date, and otherwise, the performance of the application is greatly reduced. Another type of task is a task with a soft deadline, such as a video analytics task, which violates its deadline properly without significantly impacting the performance of the application. Different users generate different tasks in real time and broadcast the unloading requests of the tasks to nearby base stations.
Step two, the base station receives the task unloading request of the user and sends the following information in the edge server connected with the base station to the mobile user: channel bandwidth b between user and server ji，jTransmission power pi，jChannel gain gi，jPower w of noiseoFrequency of CPU fjAnd the running state of each task in the edge server. The tasks can be divided into two types according to the running state of the tasks in the server: one is a task that has been offloaded to the edge server, is performing a computation or is waiting in a queue for a computation, and the other is a task that has been decided to be offloaded to the server, is currently in transit.
And step three, the mobile user receives information of each edge server transmitted by the nearby base station, and runs a task allocation algorithm to perform task unloading for task allocation edge servers with different expiration dates. The specific allocation algorithm flow is shown in fig. 3. The specific flow of the allocation algorithm is as follows: firstly, the mobile user calculates the total completion time of task unloading to the server according to the running state of the existing task on the edge server and the information of the unloading task, wherein the completion time comprises the task uploading time, the task calculating time and the task waiting time. The user then checks whether the task can be completed on the edge server before the expiration date. If it can be done, the user distributes the task to the edge server. If not, the user determines whether there are other candidate edge servers. If yes, the completion time of the task on the server is calculated according to the same formula, and whether the task can be completed before the expiration date is judged. If not, the user only considers the tasks with hard expiration dates on the server and recalculates the total completion time of the tasks. This operation is intended to insert the offloaded task before the task with the soft deadline. An insert operation may cause the completion time of a task with a soft expiration date to violate its expiration date. And counting the number of tasks violating the deadline caused by the inserting operation, and unloading the tasks to the edge server with the minimum number of the tasks violating the deadline. The above-described allocation ensures that tasks with hard expiration dates are always completed and that tasks with soft expiration dates violate their expiration dates appropriately.
And step four, the edge server receives the tasks unloaded by the user, and a task scheduling algorithm is operated to determine the execution time and sequence of each task on the server so as to maximize the benefit of completing the tasks in the edge computing network. The specific scheduling algorithm flow is shown in fig. 4.
In the invention, an edge server receives tasks unloaded by a user, and a task scheduling algorithm is operated to determine the execution time of each task on the server so as to maximize the benefit of completing the tasks in an edge computing network, wherein the scheduling process comprises the following steps:
the specific flow of the scheduling algorithm is as follows: first, the edge server builds a task revenue model based on the task deadline. Specifically, tasks with hard deadlines, if availableThe system will receive a greater gain if done before the expiration date, and a greater penalty if not done before the expiration date. Compared to tasks with hard expiration dates, tasks with soft expiration dates will receive less revenue if completed before the expiration date, and less penalty if not completed before the expiration date. Secondly, the edge server divides the tasks into two categories, one is the task which can be before the deadline at present and is marked as { icOne category, tasks that cannot currently be completed by the expiration date, is denoted as { i }u}. Definition icThe loss function c of the task ini，jThe loss function represents the loss caused to the system if the execution time of the task is pushed back so that the task cannot be completed before the expiration date. In addition, { i } is defineduGet function r of tasks in }i′，iThe revenue function represents the revenue that the system would bring if the execution time of the task was advanced so that the task could be completed before the expiration date. Subsequently, the edge server starts from { i }cIs selected to have min ci，jLet a be min ci，jFrom { i ] simultaneouslyuChoose to have maxr ini′，jLet b be maxri′，j. If b ≧ a, the execution order of the above tasks is interchanged, i.e., task i' is changed from { i ≧ auRemove it and advance its execution time, get task i from { i }cRemove it and push back its task execution time. If b < a, the routine ends. Subsequently, the edge server determines { i }cAnd { i }uAnd (4) whether all the tasks in the queue are traversed or not, and if not, repeating the scheduling steps. Otherwise, the routine ends.
The scheduling algorithm ensures that the obtained system benefit is a monotone increasing function when task replacement is carried out each time. The system gains maximum benefit when replacement is stopped.
And step five, each task on the edge server obtains the execution time and sequence of the task through a scheduling algorithm. The tasks are executed in sequence according to the sequence determined by the scheduling algorithm so as to obtain greater system benefit.
Fig. 5 is a flowchart of the interaction between the user side and the edge server side according to the present invention. First, a task offload request is sent by a mobile user to a nearby edge server. After receiving the request of the user, the edge server sends the information of the computing capacity of the server, the data transmission bandwidth, the running state of each task on the server and the like to the mobile user. And the mobile user receives the information sent by the edge server, runs a task allocation algorithm and allocates a proper edge server for the task to unload the task. And the edge server receives the tasks uploaded by the user, runs a task scheduling algorithm and sets execution time and sequence for the tasks. And after the task completes calculation according to a scheduling algorithm, the server sends a calculation result to the mobile user. At this point, the task offload process ends.
The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above-mentioned embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and embellishments within the scope of the invention may be made by those skilled in the art without departing from the principle of the invention.
Claims (5)
1. An expiration date-aware distributed task offloading method in edge computing, the offloading method comprising the steps of:
s1, the mobile client sends the unloading request of the task generated by the delay sensitive application to a plurality of base stations in the signal coverage range, the edge server corresponding to the base station calculates the unloading task, and each base station is connected with at least one edge server;
s2, the base station receiving the task unloading request sends the edge server information connected with the base station to the mobile client, wherein the edge server information comprises the server computing capacity, the data transmission bandwidth and the running state of each task on the server;
s3, the mobile client distributes the edge servers for the task in a distributed manner to calculate the task by combining the task deadline, the task deadline type and all the received edge server information; the task deadline type comprises a hard deadline and a soft deadline, the income of the hard deadline completed by date is greater than a preset income threshold, the loss of the soft deadline incomplete is less than a preset loss threshold, and the income and the loss are calculated and obtained based on the application performance of the delay sensitive application; the constraints of the allocation process are: tasks with hard expiration dates can always be completed;
s4, the edge server receives the distributed tasks sent by the mobile client, and schedules the tasks in real time according to the types of the deadline of the tasks and the states of other tasks running by the edge server so as to maximize the benefit of completing the tasks in the edge computing network;
s5, the edge server completes the calculation process of the task according to the scheduling result of the step S4.
2. The method for distributed task offloading as defined in claim 1 wherein in step S1, the ith task T generated by the delay-sensitive application is generatediIs a quadruple
Wherein d isiNumber of bits to be transmitted for a task, wiIs the workload of the task or tasks,
indicating the expiration date of the execution of the task,
is the release time of the task or tasks,
as the deadline of the task, IiThe expiration date type of the characterization task.
3. The distributed task offload method for deadline awareness in edge computing according to claim 2, wherein in step S2, the edge server sends a message to the edge serverInformation includes the ith task TiChannel bandwidth b between corresponding mobile ue and edge server ji，jTransmission power pi，jChannel gain gi，jNoise power woCPU frequency fjAnd the running state of each task in the edge server; the tasks are divided into two types according to the running state of the tasks in the edge server: one is a task that has been offloaded to the edge server, is performing a computation or is waiting in a queue for a computation, and the other is a task that has been decided to be offloaded to the server, is currently in transit.
4. The method for unloading tasks in an edge computing with distributed task awareness of expiration date as claimed in claim 3, wherein in step S3, the mobile client combines the expiration date of the task, the type of the expiration date of the task, and all the received edge server information, and the process of distributively allocating the edge servers for the task to perform the task computing includes the following steps:
s31, calculating the task T according to the following formulaiUpload time of transfer to edge server j
Wherein r isi，jFor the task transmission rate, it is expressed as:
s32, estimating the task T if the task is to be processed according to the running state of each task in the edge serveriUploading to the edge server j, receiving the task from the edge server j and waiting time of the task in the process of executing the task
S33, calculating the task T according to the following formulaiCompute time on edge server j
S34, calculating the task T according to the following formulaiTotal completion time t on edge server ji，j：
S35, according to t calculated in step S34i，jDetecting task TiWhether it can be completed on the edge server j before the corresponding expiration date, and if so, selecting the task TiUnloading the server to an edge server j, if the server can not be unloaded, traversing all edge servers covering the mobile user in sequence according to the same calculation formula until the edge server allowing unloading is found, and ending the process; if all the edge servers cannot complete the task before the task' S expiration date, go to step S36;
s36, only considering the tasks with hard expiration date in the edge server, recalculating the task waiting time
And total completion time ti，j(ii) a Counting the number of tasks violating the deadline caused by the insertion operation, and assigning the task TiAnd unloading the data to the edge server with the minimum number of tasks violating the deadline, and ending the flow.
5. The method for distributed task offloading with deadline awareness in edge computing according to claim 1, wherein in step S4, the process of scheduling the task in real time according to the deadline type of the task and the status of other tasks running by the task comprises the following steps:
s41, establishing a task income model according to the task deadline;
s42, dividing the tasks in the edge server into two types, wherein one type is the task which can be before the deadline currently and is marked as { i }cOne category, tasks that cannot currently be completed by the expiration date, is denoted as { i }u}; definition icThe loss function c of the task ini，jThe system is used for representing the loss brought to the system if the execution time of the task i is pushed back, so that the task cannot be completed before the deadline; definition iuGet function r of tasks in }i′，jRepresenting the revenue for the system if the execution time of task i' is advanced so that the task can be completed before the expiration date;
s43, from { icIs selected to have min ci，jTask i of (1), let a be min ci，jFrom { i ] simultaneouslyuIs selected with max ri′，jTask i' of (1), let b be max ri′，j(ii) a If b ≧ a, the execution order of the above tasks is interchanged, i.e., task i' is changed from { i ≧ auRemove it and advance its execution time, get task i from { i }cRemoving the task and pushing the task execution time;
s44, judging { icAnd { i }uWhether all tasks in the sequence are traversed or not, if not, repeating the step S43, otherwise, ending the process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110828570.XA CN113791878B (en) | 2021-07-21 | 2021-07-21 | Distributed task unloading method for perceiving expiration date in edge calculation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110828570.XA CN113791878B (en) | 2021-07-21 | 2021-07-21 | Distributed task unloading method for perceiving expiration date in edge calculation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113791878A true CN113791878A (en) | 2021-12-14 |
| CN113791878B CN113791878B (en) | 2023-11-17 |
Family
ID=79181232
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110828570.XA Active CN113791878B (en) | 2021-07-21 | 2021-07-21 | Distributed task unloading method for perceiving expiration date in edge calculation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113791878B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170288957A1 (en) * | 2016-03-30 | 2017-10-05 | International Business Machines Corporation | Input method engine management for edge services |
| CN108920279A (en) * | 2018-07-13 | 2018-11-30 | 哈尔滨工业大学 | A kind of mobile edge calculations task discharging method under multi-user scene |
| CN110647403A (en) * | 2019-10-31 | 2020-01-03 | 桂林电子科技大学 | Cloud computing resource allocation method in multi-user MEC system |
| CN112099932A (en) * | 2020-09-16 | 2020-12-18 | 广东石油化工学院 | Optimal pricing method and system for soft-hard deadline task offloading in edge computing |
| CN112231009A (en) * | 2020-09-17 | 2021-01-15 | 浙江工业大学 | Energy capture network model task calculation unloading decision and scheduling method |
| CN112351503A (en) * | 2020-11-05 | 2021-02-09 | 大连理工大学 | Task prediction-based multi-unmanned-aerial-vehicle-assisted edge computing resource allocation method |
| CN112995023A (en) * | 2021-03-02 | 2021-06-18 | 北京邮电大学 | Multi-access edge computing network computing unloading system and computing unloading method thereof |
| US20210191827A1 (en) * | 2018-08-30 | 2021-06-24 | Teiefonaktiebolaget LM Ericsson (pub!) | System and method for collaborative task offloading automation in smart containers |
-
2021
- 2021-07-21 CN CN202110828570.XA patent/CN113791878B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170288957A1 (en) * | 2016-03-30 | 2017-10-05 | International Business Machines Corporation | Input method engine management for edge services |
| CN108920279A (en) * | 2018-07-13 | 2018-11-30 | 哈尔滨工业大学 | A kind of mobile edge calculations task discharging method under multi-user scene |
| US20210191827A1 (en) * | 2018-08-30 | 2021-06-24 | Teiefonaktiebolaget LM Ericsson (pub!) | System and method for collaborative task offloading automation in smart containers |
| CN110647403A (en) * | 2019-10-31 | 2020-01-03 | 桂林电子科技大学 | Cloud computing resource allocation method in multi-user MEC system |
| CN112099932A (en) * | 2020-09-16 | 2020-12-18 | 广东石油化工学院 | Optimal pricing method and system for soft-hard deadline task offloading in edge computing |
| CN112231009A (en) * | 2020-09-17 | 2021-01-15 | 浙江工业大学 | Energy capture network model task calculation unloading decision and scheduling method |
| CN112351503A (en) * | 2020-11-05 | 2021-02-09 | 大连理工大学 | Task prediction-based multi-unmanned-aerial-vehicle-assisted edge computing resource allocation method |
| CN112995023A (en) * | 2021-03-02 | 2021-06-18 | 北京邮电大学 | Multi-access edge computing network computing unloading system and computing unloading method thereof |
Non-Patent Citations (5)
| Title |
|---|
| TIANTIAN YANG 等: "Latency Optimization-based Joint Task Offloading and Scheduling for Multi-user MEC System", 《2020 29TH WIRELESS AND OPTICAL COMMUNICATIONS CONFERENCE (WOCC)》 * |
| XIN HE等: "Offloading Deadline-aware Task in Edge Computing", 《 2020 IEEE 13TH INTERNATIONAL CONFERENCE ON CLOUD COMPUTING (CLOUD)》 * |
| 陈玺: "智能边缘系统中的训练数据收集及任务调度研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 陈贵海等: "面向边缘计算的资源优化技术研究进展", 《大数据》, vol. 5, no. 2 * |
| 马惠荣;陈旭;周知;于帅;: "绿色能源驱动的移动边缘计算动态任务卸载", 计算机研究与发展, no. 09 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113791878B (en) | 2023-11-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108509276B (en) | Video task dynamic migration method in edge computing environment | |
| CN111240701B (en) | Task unloading optimization method for end-side-cloud collaborative computing | |
| CN109862592B (en) | Resource management and scheduling method in mobile edge computing environment based on multi-base-station cooperation | |
| CN110109745B (en) | Task collaborative online scheduling method for edge computing environment | |
| US9596281B2 (en) | Transport accelerator implementing request manager and connection manager functionality | |
| CN109343904B (en) | Lyapunov optimization-based fog calculation dynamic unloading method | |
| US9813529B2 (en) | Effective circuits in packet-switched networks | |
| CN110489176B (en) | Multi-access edge computing task unloading method based on boxing problem | |
| CN109951849B (en) | Method for combining resource allocation and content caching in F-RAN architecture | |
| US9930097B2 (en) | Transport accelerator systems and methods | |
| WO2009026321A2 (en) | Media streaming with online caching and peer-to-peer forwarding | |
| CN110072130B (en) | HTTP/2-based HAS video slice pushing method | |
| Zhong et al. | Age-aware scheduling for asynchronous arriving jobs in edge applications | |
| CN111741249B (en) | Network congestion detection method and device | |
| US10965613B2 (en) | Multi-pipe bandwidth control in hosted systems | |
| US11729241B2 (en) | Maximum sustainable encoding bit rates for video downloads | |
| CN110113140B (en) | Calculation unloading method in fog calculation wireless network | |
| CN107708152A (en) | The task discharging method of isomery cellular network | |
| CN111935783A (en) | Edge cache system and method based on flow perception | |
| CN116366576A (en) | Method, device, equipment and medium for scheduling computing power network resources | |
| CN113687876A (en) | Information processing method, automatic driving control method and electronic equipment | |
| CN113778675A (en) | Calculation task distribution system and method based on block chain network | |
| CN113791878B (en) | Distributed task unloading method for perceiving expiration date in edge calculation | |
| CN109688421B (en) | Request message processing method, device and system, server and storage medium | |
| CN111278039B (en) | User perception suppression identification method, device, equipment and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |