CN112887347B - Dynamic migration method and device for edge calculation in industrial internet - Google Patents

Dynamic migration method and device for edge calculation in industrial internet Download PDF

Info

Publication number
CN112887347B
CN112887347B CN201911199046.XA CN201911199046A CN112887347B CN 112887347 B CN112887347 B CN 112887347B CN 201911199046 A CN201911199046 A CN 201911199046A CN 112887347 B CN112887347 B CN 112887347B
Authority
CN
China
Prior art keywords
cloud server
task
mobile cloud
time
migration
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
CN201911199046.XA
Other languages
Chinese (zh)
Other versions
CN112887347A (en
Inventor
王哲
时晓光
黄颖
汤立波
李璐
闫霞
于青民
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.)
China Academy of Information and Communications Technology CAICT
Original Assignee
China Academy of Information and Communications Technology CAICT
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 China Academy of Information and Communications Technology CAICT filed Critical China Academy of Information and Communications Technology CAICT
Priority to CN201911199046.XA priority Critical patent/CN112887347B/en
Publication of CN112887347A publication Critical patent/CN112887347A/en
Application granted granted Critical
Publication of CN112887347B publication Critical patent/CN112887347B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806—Task transfer initiation or dispatching
    • G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
    • G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005—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
    • G06F9/5038—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 considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5083—Techniques for rebalancing the load in a distributed system
    • G06F9/5088—Techniques for rebalancing the load in a distributed system involving task migration
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06316—Sequencing of tasks or work
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
    • G06Q10/087—Inventory or stock management, e.g. order filling, procurement or balancing against orders
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/14—Session management
    • H04L67/148—Migration or transfer of sessions
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/50—Network services
    • H04L67/60—Scheduling or organising the servicing of application requests, e.g. requests for application data transmissions using the analysis and optimisation of the required network resources
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2209/00—Indexing scheme relating to G06F9/00
    • G06F2209/48—Indexing scheme relating to G06F9/48
    • G06F2209/484—Precedence
    • 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
    • G06F2209/5021—Priority

Abstract

The application provides a dynamic migration method and a dynamic migration device for edge computing in an industrial internet, wherein the method comprises the following steps: when a request for calculating migration sent by a mobile terminal is received, system information of a dynamic system is obtained; acquiring the maximum value of a preset system time gain function according to the system information and the time delay carried by the request; determining an execution main body for the request distribution when the maximum value of a preset system time gain function is obtained, and responding the task execution main body to the mobile terminal; the task execution subject is a mobile cloud server or a central cloud server. The method can improve the calculation migration efficiency on the premise of obtaining the maximum system time gain.

Description

Dynamic migration method and device for edge calculation in industrial internet
Technical Field
The invention relates to the technical field of data processing, in particular to a method and a device for dynamically migrating edge calculation in an industrial internet.
Background
Currently, industry transformation and industrial digital transformation are rapidly raised in a global range, and industrial internet edge computing is taken as a product of deep integration of information technology and manufacturing industry and becomes a key technical support of a new industrial revolution.
In the process of constructing an industrial internet architecture by an enterprise, due to the limitation of insufficient computing processing capacity and storage capacity, an equipment terminal cannot meet the computing capacity requirement of application.
For example, in the logistics storage industry, an Automated Guided Vehicle (AGV) cannot plan and design a driving path in real time according to its own application requirements, and if a related computation-intensive application task is migrated to a server with a strong computing power, the application completion time can be effectively reduced. However, the existing AGV control system based on cloud service is a centralized service, but due to the factors such as distance attenuation of wireless communication and network bandwidth limitation, the communication reliability between AGV and server is poor, and the data transmission time is long, so the AGV control system based on cloud service is not suitable for providing real-time service for large warehouses and a large number of AGVs.
Disclosure of Invention
In view of this, the present application provides a method and an apparatus for dynamic migration of edge computing in an industrial internet, which can improve the efficiency of computing migration on the premise of obtaining the maximum system time gain
In order to solve the technical problem, the technical scheme of the application is realized as follows:
in one embodiment, a method for dynamic migration of edge computing in an industrial internet is provided, which is applied to a computing scheduling CS in a dynamic system, and the method includes:
when a request for calculating migration sent by a mobile terminal is received, system information of a dynamic system is obtained;
acquiring the maximum value of a preset system time gain function according to the system information and the time delay carried by the request;
determining an execution main body for requesting allocation when the maximum value of a preset system time revenue function is obtained, and responding the task execution main body to the mobile terminal; the task execution main body is a mobile cloud server or a central cloud server.
In another embodiment, an edge computing dynamic migration apparatus in an industrial internet is provided, which is applied to a computing scheduling CS in a dynamic system, and includes: a receiving unit, an obtaining unit and a determining unit;
the receiving unit is used for receiving a request for calculating migration sent by the mobile terminal;
the acquiring unit is used for acquiring the system information of the dynamic system when the receiving unit receives a request for computing migration sent by the mobile terminal; acquiring the maximum value of a preset system time gain function according to the system information and the time delay carried by the request;
the determining unit is configured to determine that the acquiring unit acquires the maximum value of the preset system time gain function and then responds the task execution subject to the mobile terminal; the task execution subject is a mobile cloud server or a central cloud server.
In another embodiment, an electronic device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the steps of the method for edge computing live migration in the industrial internet.
In another embodiment, a computer readable storage medium is provided, on which a computer program is stored, which when executed by a processor, performs the steps of the method for edge computing live migration in industrial internets.
According to the technical scheme, when a request for calculating migration sent by a mobile terminal is received, the system information of the dynamic system is obtained, and the maximum value of the preset system time gain function, namely the maximum time gain of the system, is obtained according to the system information, so that the execution main body allocated for the task request is determined. According to the scheme, the calculation migration efficiency can be improved on the premise of obtaining the maximum system time benefit.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the description below are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a schematic diagram of a plant logistics edge calculation system;
FIG. 2 is a schematic diagram of a plant logistics edge calculation;
FIG. 3 is a schematic diagram illustrating an edge calculation live migration process in the industrial Internet according to an embodiment of the present disclosure;
FIG. 4 is a schematic diagram of system time gain under different conditions of calculating the arrival rate of requests for migration;
FIG. 5 is a diagram illustrating expected completion times of tasks under different conditions of computing request arrival rates for migration;
FIG. 6 is a schematic diagram of system time gains under different mobile cloud server leaving rate conditions;
FIG. 7 is a schematic diagram of system time yield of tasks under different mobile cloud server number conditions;
FIG. 8 is a schematic diagram of an apparatus for implementing the above technique in an embodiment of the present application;
fig. 9 is a schematic physical structure diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only some embodiments of the present application, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
The terms "first," "second," "third," "fourth," and the like in the description and in the claims, as well as in the drawings, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are, for example, capable of operation in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprising" and "having," as well as any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements explicitly listed, but may include other steps or elements not explicitly listed or inherent to such process, method, article, or apparatus.
The technical solution of the present invention will be described in detail with reference to specific examples. Several of the following embodiments may be combined with each other and some details of the same or similar concepts or processes may not be repeated in some embodiments.
The embodiment of the application provides an edge computing dynamic migration method in an industrial internet, which is applied to a Computing Scheduling (CS) in a dynamic system, and a small edge computing server is carried by using a specific mobile terminal.
The application scenarios of edge computing in the industrial internet are numerous, such as factory logistics, logistics warehousing industry, logistics management industry and the like. The following takes a plant logistics management application as an example, but the embodiments of the present application are not limited to such application.
Firstly, a dynamic system is described, wherein the dynamic system comprises a mobile cloud server, and the mobile cloud server is a mobile terminal carrying a small edge computing server and is hereinafter referred to as a mobile cloud server; a mobile terminal not equipped with a small edge computing server, hereinafter referred to as a mobile terminal, may also be included; and when the mobile cloud server corresponding to the CS leaves the dynamic system, re-electing the CS according to the preset rule.
Taking a plant logistics edge calculation system as an example, referring to fig. 1, fig. 1 is a schematic diagram of the plant logistics edge calculation system. In fig. 1, each mobile terminal is an AGV logistics vehicle, and each AGV logistics vehicle is configured with a vehicle-mounted communication unit, and the communication radius of the vehicle-mounted communication unit is R. Meanwhile, the AGV logistics vehicle with the path planning and other calculation requirements is called a mobile terminal, and the specific AGV vehicle loaded with the server with the strong calculation capacity is used as a mobile cloud server.
The mobile cloud server is used as the assistance of a central cloud server for processing the computing tasks and bears the computing tasks at the edge of the network, the mobile cloud server and the cloud computing resources can mutually borrow the computing resources to enhance the network performance, and the flexibility of the cloud is greatly improved. On the other hand, by interacting with the vehicle, shortage of communication resources at the network edge and bandwidth pressure of the backbone network can be alleviated.
In a factory logistics edge computing system, a queuing theory is utilized to model the arrival process of a mobile terminal computing migration request, the computing migration request of a user terminal is regarded as a migration task, and corresponding execution priority is given to the migration task according to the time delay requirements of different migration tasks. The concrete implementation is as follows:
when receiving a request for computing migration, the CS allocates priority to the request according to a preset rule and time delay carried by the request;
the preset rule is a mapping relationship between priorities and delays, for example, 10 priorities are configured, each priority configuration corresponds to a delay range, and the priority belongs to the delay range and corresponds to the delay range. .
And simulating and dividing the corresponding requests into corresponding queues according to the priority based on a queuing model.
Referring to fig. 2, fig. 2 is a schematic diagram of a plant logistics edge calculation. The CS in the dynamic system considers that a calculation migration request initiated by a newly arrived mobile terminal is regarded as the arrival of a new task in the queuing system, and supposing that the tasks with different time delay requirements of N types correspond to one type for each priority, wherein the task arrival rate of each task type obeys Poisson distribution.Meanwhile, the execution priority of each task is sorted by using the difference of the time delay (requirement) T of each task, and T = { T = { (T) 1 ，T 2 ，……，T N In which T 1 ＜T 2 ＜……＜T N 。
Further, with reference to some widely used wireless communication models, it is assumed that the wireless transmission rate between mobile terminals is fixed, and it is considered that the communication delay of the task input data is negligible compared to the execution time of the task.
The mobile cloud server forms an edge cloud with dynamically changed computing resources according to a clustering principle, and the edge cloud and the center cloud form a logistics edge computing system with a hierarchical structure.
The central cloud server has abundant computing resources, but the communication time delay is high, compared with the central cloud, the mobile cloud server is close to the user vehicle due to the position, the communication time delay is effectively reduced, and the user vehicle can reduce the time delay and improve the communication success rate by transferring the computing task to the mobile cloud server.
According to a rule of selecting cluster heads in a clustering principle, CS is screened out from a factory logistics edge computing system, and communication between the CS and each mobile cloud server and a central cloud is guaranteed. When a user vehicle initiates a computing migration request, the CS will decide whether the mobile cloud server receives the request or the central cloud server receives the request according to the computing resources available to the system and the task latency requirements for the request to be performed. If the mobile cloud server receives the request, the CS decides which mobile cloud servers are allocated to process according to the task priority; if the central cloud server executes the task, the mobile terminal needs to transmit the task input data to the CS, and then the CS forwards the migration request to the central cloud server for processing.
The specific implementation process of how the CS schedules the request on the premise of ensuring the maximum system time gain, that is, allocating the main body for executing the corresponding task, is as follows:
referring to fig. 3, fig. 3 is a schematic view illustrating a process of dynamic migration of edge calculation in the industrial internet according to an embodiment of the present application. The method comprises the following specific steps:
step 301, when receiving a request for computing migration sent by a mobile terminal, acquiring system information of a dynamic system.
The system information includes: the system comprises the number of current mobile cloud servers, the number of mobile cloud servers distributed by computing migration tasks with different priorities, request tasks, the completion rate of the tasks, the arrival rate and the departure rate of the mobile cloud servers, computing migration requests, events generated by the mobile terminals and the like.
Step 302, obtaining a maximum value of a preset system time gain function according to the system information and the time delay carried by the request.
The preset system time gain function in the embodiment of the application is as follows:
Figure BDA0002295399630000061
wherein, V π (s) System time gain at State s, V, when strategy π is taken π (s ') is the system time gain in state s' when strategy π is taken; r (s, a) is the system time gain in state s when the action a is executed, P (s '| s, a) is the transition probability of the system state from s to s' after the action a is taken, gamma is the influence factor of the system state time gain to which the system state is transitioned after the action a is executed on the strategy pi,
Figure BDA0002295399630000063
the strategy pi is a set of system states, and the allocation execution main body is a central cloud server or a mobile cloud server;
the behavior action a belongs to a behavior action set; the state s and the state s' belong to a system state set;
wherein the system state represents a state of a plant logistics edge computing system, the set of system states comprising: the method comprises the following steps of the number of mobile cloud servers in the current state, the number of mobile cloud servers allocated to tasks with priority i in a system, and one event in an event set, wherein the event set comprises: when a new computing migration request comes, the system completes a task with the priority i, the arrival of a mobile cloud server and the departure of the cloud server;
such as using
Figure BDA0002295399630000062
To represent a set of system states:
Figure BDA0002295399630000071
where C is the number of mobile cloud servers in the current state, and the value of C may change with the change of the state due to the vehicle mobility. l. the i Represents the number of mobile cloud servers allocated to the task of priority i in the system, and therefore the total number of mobile cloud servers already occupied in the system is
Figure BDA0002295399630000072
e represents an event in the event set epsilon, e epsilon = { A, D 1 ，D 2 ，……，D N ，F 1 ，F -1 Denotes that a new compute migration request comes, D i Indicating completion of the system to perform a task of priority i, F 1 And F -1 Respectively representing the arrival and departure of a mobile cloud server.
The system state set contains all possible situations, but the actual content corresponding to the current system state is obtained each time a request is received.
The set of behavioral actions includes: and the task with the priority i reaches the system and stays on the mobile cloud server for execution processing, the newly arrived task is forwarded to the central cloud server for processing, and the system state information is updated without executing behavior actions.
In the present embodiment, the CS will slave the set of system behaviors when given a system state s
Figure BDA0002295399630000074
Selecting the appropriate behavioral action a(s), i.e.
Figure BDA0002295399630000073
The method includes the steps that a(s) = i indicates that a task with the priority of i arrives at a system and is left on a mobile cloud service for execution processing, a mobile cloud server is allocated to the system, a(s) =0 indicates that a newly arrived task is forwarded to a central cloud for processing, and a(s) = -1 indicates that the system does not need to do any action under a current event, and only needs to update system state information, for example, when events such as completion of task execution, arrival and departure of the mobile cloud server and the like occur, the system does not need to make a migration decision, and only needs to update state information such as occupation conditions of the mobile cloud server and the like.
The step of determining P (s' | s, a) comprises:
and determining P (s' | s, a) according to the expected time interval between two continuous states, the arrival rate of the task, the completion rate of the task, the leaving rate and the arrival rate of the mobile cloud server.
The following gives a detailed procedure for determining P (s' | s, a) in connection with the embodiments of the present application:
taking action a(s) in state s can cause the state to transition, and in order to obtain the system transition probability, the average occurrence rate σ (s, a) of events in the mobile cloud computing system of the internet of vehicles needs to be obtained first, and the selected action a in state s is the reciprocal of the expected time interval τ (s, a) between two continuous states.
Figure BDA0002295399630000081
Wherein λ is j Indicating the arrival rate of tasks with priority j, the total arrival rate of tasks can be expressed as
Figure BDA0002295399630000082
Meanwhile, the completion rate of a task may be expressed as
Figure BDA0002295399630000083
Several expressions among the expressions of σ (s, a) will be described below, one for each:
when a new mobile cloud server is added into the system, the total number of mobile cloud servers in the system is increased by 1.
When there is a mobile cloud server in the system to leave, the total number of mobile cloud servers in the system is reduced by 1.
When the task with the priority j reaches the system and is left in the system for processing, the number of occupied mobile cloud servers in the system is
Figure BDA0002295399630000084
And when the task is forwarded to the central cloud processing, the number of occupied mobile cloud servers in the system is kept unchanged.
When a task with the priority of j in the system is completed and executed, the correspondingly distributed mobile cloud servers are idle, and the number of occupied mobile cloud servers is reduced to
Figure BDA0002295399630000085
The corresponding task completion rate is reduced to
Figure BDA0002295399630000086
According to different situations, when the system takes the action a, the transition probability P (s '| s, a) of the system state from s to s' can be further expressed as:
in the first case:
s＝(C，l 1 ，l 2 ，...，l N and A) then
Figure BDA0002295399630000091
In the second case:
s＝(C，l 1 ，l 2 ，...，l N ，D i ) Then, then
Figure BDA0002295399630000101
In a third case:
s＝(C，l 1 ，l 2 ，...，l N ，F 1 ) Then, then
Figure BDA0002295399630000102
In the fourth case, s = (C, l) 1 ，l 2 ，...，l N ，F -1 ) Then, then
Figure BDA0002295399630000111
Wherein λ is f For a leaving rate of a mobile cloud server, μ f Is the arrival rate of the mobile cloud server.
A step of determining r (s, a), comprising:
determining the difference value of the migration instant time benefit and the system time cost calculated by the system as r (s, a);
the system calculates the migration instant time gain and determines the migration instant time gain according to the difference value of the completion time of the task and the time delay carried in the request; the system time cost is determined based on a desired time interval between two successive states and a time cost per unit time of the system.
Based on the system state and the system behavior, a corresponding system time gain calculation mode can be obtained. The system needs to consider not only the time gain brought by the plant logistics edge computing system to execute the migration task, but also the time cost brought by the occupation of computing and communication resources.
The system time benefit r (s, a) is determined by the calculated migration time benefit h (s, a) and the system cost g (s, a) of the system taking action a in state s, and can be expressed as
r(s，a)＝h(s，a)-g(s，a)
The system calculates the migration instant time gain h (s, a) which is mainly determined by the difference between the completion time of the task and the corresponding time delay requirement, and can be calculated as
Figure BDA0002295399630000121
The time-of-day benefit h (s, a) is explained in detail as follows:
when a computing migration request is received by the mobile cloud computing system of the Internet of vehicles, the system can obtain
Figure BDA0002295399630000122
Wherein when X is not less than 0, [ X ]] + = X, when X < 0, [ X] + = - ∞, eta represents the gain per unit time, omega 1 Representing the transfer time, T, of the migration of task input data to the computation schedule CS and the return of the execution results i Indicating a task latency requirement of priority i.
For a task with priority i, the processing flow of the task can be regarded as a queuing model, because the arrival of each task is subject to the poisson process, and the arrival and the processing of the task are relatively independent, so that the expected completion time W of the task with priority i can be obtained i ，
Figure BDA0002295399630000123
Wherein, the first and the second end of the pipe are connected with each other,
Figure BDA0002295399630000124
representing the utilization in the queuing model, the probability of no task with priority i in the system can be expressed as:
Figure BDA0002295399630000125
(1) In this way, when a compute migration request is forwarded to the central cloud for processing (e.g., there are no mobile cloud servers in the system that are idle), the system can obtain
Figure BDA0002295399630000131
In time of day, wherein, ω 2 And the transmission time for forwarding the computation migration request to the center cloud and returning the result after the processing is completed is represented by the computation scheduling CS. Because the traditional central cloud has strong processing capacity, the processing time delay of the task in the central cloud processor can be ignored.
(2) When the task of computing migration is completed by the system, the system does not get any time gain.
(3) When the system has idle computing resources, the mobile cloud server leaves the system, and the system can not obtain time benefits.
(4) When a new mobile cloud server is added into the system, the system can not obtain time benefits.
(5) When all the computing resources in the system are occupied, the mobile cloud server leaves, which indicates that the task execution of the mobile cloud server which is already allocated before cannot be guaranteed, user experience is affected, and the system loses phi, wherein the phi represents a specific loss value.
Further, the system cost g (s, a) may be expressed as:
g(s，a)＝o(s，a)τ(s，a)
where τ (s, a) represents the desired time interval between two consecutive states and o (s, a) represents the cost per unit time of the system, which can be characterized by the total number of mobile cloud servers occupied in the system, i.e.
Figure BDA0002295399630000132
In summary, since the system state does not change between two consecutive decision moments, the expected discount system time benefit r (s, a) can be obtained to consider the influence of the future system state on the current system state and behavior, and then determine the system time benefit as follows:
Figure BDA0002295399630000133
wherein, alpha is a discount factor of the duration and can be set according to actual needs.
On the premise that each value in the preset system time gain function can be calculated, the method for obtaining the maximum value of the preset system time gain function specifically comprises the following steps:
iterating a preset system time gain function;
and when the difference value between the value of the preset system time gain function obtained at the K +1 th time and the value of the preset system time gain function obtained at the K th time is smaller than the preset value, determining the value of the system time gain obtained at the K +1 th time as the maximum value of the preset system time gain function.
Considering the characteristics of a request of computing migration and dynamic change of a mobile cloud server, the behavior of each system can influence the state of the next system, and the influence of the future state on the current decision time is considered by using a value function of a semi-decision markov decision process (SMDP) in the embodiment of the application.
Through a state transition probability equation, the optimal preset system time gain time function can be expressed in a bellman mode, and the method specifically comprises the following steps:
Figure BDA0002295399630000141
wherein, pi * Indicating the optimal migration mechanism, i.e. what behavior to perform to what system state. And the process of solving the optimal migration mechanism is to improve the value function by continuously carrying out iterative updating according to a dynamic programming Bellman formula. And when the value function is close to the optimum through the bounded error measurement, namely the Bellman errors before and after the updating of all the value functions in each iteration are smaller than the epsilon, the value function values are converged to obtain a migration decision. Otherwise, the value function will continue to iterate until convergence.
The specific iterative process is given below:
in a specific iteration process, before the first iteration, V(s) corresponding to each state in the system state set has no value or a value of 0, which is equivalent to an initialization. s 'is also a state in the set, so V (s') is also 0.
The first iteration, for any state s,
Figure BDA0002295399630000142
the maximum function of the state s can be obtained, and corresponding to the execution of the action a, pi is iterated for the first time * (s)＝a。
We can also get V 1 (s'), which is essentially V for each state in the set of states 1 (xxx) We can get that xxx generally refers to any state.
Since the initialization V(s) =0, it corresponds to V 0 (s) =0, for each state s, as long as | V 1 (s)-V 0 (s) | is a number greater than e, iteration continues.
In the second iteration of the process,
Figure BDA0002295399630000151
where s' is a state belonging to a set of states,
Figure BDA0002295399630000152
and representing all the states in the traversal state set, and then summing the results.
In the second iteration process, V corresponding to each state has its own value, and new pi can be obtained * (s) that the optimal decision-making action has changed.
For each state s in the state set, | V is utilized 2 (s)-V 1 (s) | is determined, and if again greater than e, iteration continues until convergence.
And 303, determining an execution main body which is allocated for the request when the maximum value of the preset system time gain function is obtained, and responding the task execution main body to the mobile terminal.
The task execution subject is a mobile cloud server or a central cloud server.
When the CS determines to obtain the maximum value of a preset system time gain function and then determines to be the execution subject requested to be distributed, determining whether the execution subject is a mobile cloud server or a central cloud server;
when the execution subject is determined to be the central cloud server, directly responding the execution subject to the mobile terminal;
and when receiving the response of the CS, the mobile terminal sends the calculation task to the CS, and the CS forwards the calculation task to the central cloud server.
When the execution main body is determined to be the mobile cloud server, determining the priority corresponding to the time delay carried by the request according to the mapping relation between the priority and the time delay; and determining the mobile cloud server corresponding to the priority according to the mapping relation between the priority and the mobile cloud server.
The mapping relationship between the low configuration priority and the time delay, for example, 10 priorities are configured, each priority configuration corresponds to a time delay range, and the priority belongs to the time delay range and corresponds to the time delay range;
and local mapping relation between the priority and the mobile cloud server is configured, if the priority is 1, the mobile cloud server with the number of 1 to 3 corresponds to the local mapping relation, or the mobile cloud server with the number of 1 corresponds to the local mapping relation.
After such processing, there may be one or more mobile cloud servers responding to the mobile terminal.
When the mobile terminal receives an execution main body of a CS response, if the mobile terminal corresponds to a plurality of mobile cloud servers, selecting one mobile cloud server to establish connection according to a preset rule to carry out a request of a computing task;
the preset rule can be that a mobile cloud server with a large number is selected from the mobile cloud servers, or a mobile cloud server with a small number is selected; the selection may also be random, and in the embodiment of the present application, there is no limitation on the rule for selecting the mobile cloud server in the execution subject of the response.
If the selected mobile cloud server cannot process the calculation task because of busy calculation task, selecting the next mobile cloud server according to a preset rule;
and waiting until a mobile cloud server capable of performing task computing is selected to send a computing task to perform task computing, or when no mobile cloud server capable of performing task computing exists, and sending the computing task to the mobile cloud server to perform task computing when a command capable of performing task computing is received from the mobile cloud server.
Take the CS responding to the mobile terminal with two mobile cloud servers (mobile cloud server 1 and mobile cloud server 2) as an example.
The mobile terminal selects one mobile cloud server according to a preset rule, and if the mobile cloud server 1 is selected, a request of a computing task is sent to the mobile cloud server 1.
When receiving the request sent by the mobile terminal, the mobile cloud server 1 notifies the mobile terminal that the processing of the computing task is temporarily disabled if it is determined that the mobile terminal is in a busy computing stage.
And the mobile terminal selects the mobile cloud server 2 again and sends a request of a computing task to the mobile cloud server 2.
When receiving a request sent by the mobile terminal, the mobile cloud server 2 informs the mobile terminal that the processing of the computing task of the mobile terminal cannot be performed temporarily if the mobile cloud server is determined to be in a busy computing stage; and if the calculation task can be processed, processing the calculation task for the mobile terminal.
If the mobile terminal determines that the mobile server 2 cannot process the calculation task, the mobile terminal waits; and sending the computing task to a corresponding mobile cloud server until the mobile cloud server 1 or the mobile cloud server 2 sends an instruction that the computing task of the mobile terminal can be processed is idle.
The request sent to the CS when the execution subject needs to be determined to be the mobile cloud server or the central cloud server is a request for computing migration, and the request sent after the mobile cloud server or the central cloud server is allocated to the mobile terminal is a request for computing tasks.
The performance of the proposed compute migration mechanism was compared with a compute migration mechanism based on first come first served and a compute migration mechanism based on a preemption override by changing the system compute migration arrival rate. Regardless of which compute migration mechanism is used, the system expectation time gain decreases as the compute migration request arrival rate increases. This is because as more and more mobile cloud servers will be occupied, newly arriving tasks will need to wait longer or forward to a central cloud server, which will both increase the completion time of the task and reduce the system expected time gain.
Compared with all the computer migration mechanisms, the mechanism provided by the invention can always obtain the highest expected benefit of the system. This is because it can dynamically adjust the computational migration decision according to task latency requirements of different priorities, and takes into account the effect of future states on the expected revenue of the system using the properties of the semi-markov decision process SMDP.
Referring to fig. 4, fig. 4 is a schematic diagram illustrating system time gains under different conditions of calculating the request arrival rate of migration.
As can be seen from fig. 4, when the request arrival rate of the system computing migration is low, the computing resources in the system are relatively sufficient, and the computing migration mechanism based on first-come first-served can obtain the same performance as the computing migration mechanism proposed by the present invention. And as the arrival rate of the computation migration requests increases, the performance advantages of the computation migration mechanism provided by the invention are more and more obvious. In addition, when the arrival rate of the computing migration request is greater than 1.2, the expected revenue of the system based on the computing migration mechanism with the high priority will be reduced to 0.5 and remain unchanged, which is also the lowest time revenue of the mobile cloud computing system in the internet of vehicles.
Referring to fig. 5, fig. 5 is a diagram illustrating expected completion times of tasks under different conditions of calculating request arrival rates for migration.
Fig. 5 shows the variation trend of the completion time of different types of tasks along with the migration arrival rate calculated by the system, and it can be seen that the calculation migration mechanism provided by the present invention can always obtain the average task completion time lower than the delay requirement for different types of tasks. When the arrival rate of the system computing migration requests is between 0.2 and 0.8, the average completion time of the 3 types of tasks is the same, because there are more idle mobile cloud servers in the system at this time, and all tasks newly arriving at the system can be migrated to the mobile cloud servers for processing, so the average completion time is the same. And as the arrival rate of the system computing migration requests increases, the mobile cloud servers which are idle in the system are not enough to meet all the computing migration requests, and the computing scheduling CS forwards the tasks with low priority to the central cloud. As can also be seen from fig. 5, when the system computing migration request arrival rate is greater than 0.8, the average completion time of the task with the priority of 3 is the processing time for forwarding the task to the central cloud server.
Referring to fig. 6, fig. 6 is a schematic diagram of system time yield under different mobile cloud server leaving rate conditions.
Fig. 6 shows that 3 computing migration mechanisms show a non-increasing trend along with the increase of the leaving rate of the mobile cloud server, wherein the computing migration mechanism provided by the present invention can always obtain better performance because it can reasonably make a computing migration decision according to the delay requirements of different tasks, and on the premise of meeting the task delay requirements, the mobile cloud server preferentially serves the tasks with high priority, and meanwhile, the situation that the preemption priority reduces the system gain does not occur. While the available computing resources in the system are continuously reduced, the system benefit of the computing migration mechanism based on the high priority is always at a minimum and remains unchanged. This is because, in the case of a shortage of available mobile cloud servers, the low-priority task is interrupted by the high-priority task that comes in the process of execution, and thus, corresponding benefits cannot be obtained. For a first-come-first-serve-based computing migration mechanism, when available mobile cloud servers in a system are limited, the mechanism cannot effectively allocate the mobile cloud servers according to different delay requirements, and only can obtain corresponding benefits of tasks with loose delay requirements.
Referring to fig. 7, fig. 7 is a schematic diagram of system time yield of tasks under different mobile cloud server number conditions.
Fig. 7 compares the performance of 3 computing migration mechanisms under different numbers of mobile cloud servers, and it can be seen that the expected revenue of the system of the 3 computing migration mechanisms increases with the increase of the available computing resources in the system.
To sum up, in the embodiment of the present application, when a request for computing migration sent by a mobile terminal is received, system information of a dynamic system is obtained, and according to the system information, a maximum value of a preset system time revenue function, that is, a maximum time revenue of the system, is obtained, so as to determine an execution subject allocated for the task request. According to the scheme, the calculation migration efficiency can be improved on the premise of obtaining the maximum system time benefit.
Based on the same inventive concept, the embodiment of the application also provides an edge computing dynamic migration device in the industrial internet, which is applied to a computing scheduling CS in the dynamic system. Referring to fig. 8, fig. 8 is a schematic structural diagram of an apparatus applied to the above technology in the embodiment of the present application. The device comprises: a receiving unit 801, an acquisition unit 802, and a determination unit 803;
a receiving unit 801, configured to receive a request for computing migration sent by a mobile terminal;
an obtaining unit 802, configured to obtain system information of the dynamic system when the receiving unit 801 receives a request for computing migration sent by the mobile terminal; acquiring the maximum value of a preset system time gain function according to the system information and the time delay carried by the request;
a determining unit 803, configured to determine that the acquiring unit 802 acquires the maximum value of the preset system time revenue function, and respond the task execution body to the mobile terminal; the task execution subject is a mobile cloud server or a central cloud server.
Preferably, the first and second electrodes are formed of a metal,
the obtaining unit 802 is specifically configured to, when obtaining a maximum value of a preset system time revenue function, include:
iterating a preset system time gain function;
and when the difference value between the value of the preset system time gain function obtained at the K +1 th time and the value of the preset system time gain function obtained at the K th time is smaller than the preset value, determining the value of the system time gain obtained at the K +1 th time as the maximum value of the preset system time gain function.
Preferably, the preset system time gain function is:
Figure BDA0002295399630000191
wherein, V π (s) System time gain at State s, V, when strategy π is taken π (s ') is the system time gain in state s' when strategy π is taken; r (s, a) is the system time gain at state s when action a is performed, P (s '| s, a) is the transition probability of the system state transitioning from s to s' after action a is taken, γ is the influence factor of the system state time gain transitioning to after action a is performed on the strategy π,
Figure BDA0002295399630000192
the strategy pi is a set of system states, and the allocation execution main body is a central cloud server or a mobile cloud server;
the behavior action a belongs to a behavior action set; wherein the set of system states comprises: the method comprises the following steps of counting the number of mobile cloud servers in the current state, counting the number of mobile cloud servers allocated to a task with priority i in the system, and counting one event in an event set, wherein the event set comprises the following steps: when a new calculation migration request comes, the system completes a task with the priority i, the arrival of the mobile cloud server and the departure of the mobile cloud server;
the state s and the state s' belong to a system state set; wherein the set of behavioral actions includes: and the task with the priority i reaches the system and stays on the mobile cloud server for execution processing, the newly arrived task is forwarded to the central cloud server for processing, and the system state information is updated without executing behavior actions.
Preferably, the first and second electrodes are formed of a metal,
determining the difference value of the migration instant time gain and the system time cost calculated by the system as r (s, a);
the system calculates the migration instant time gain and determines the migration instant time gain according to the difference value of the completion time of the task and the time delay carried in the request; the system time cost is determined based on a desired time interval between two successive states and a time cost per unit time of the system.
Preferably, the first and second electrodes are formed of a metal,
and determining P (s' | s, a) according to the expected time interval between two continuous states, the arrival rate of the task, the completion rate of the task, the leaving rate and the arrival rate of the mobile cloud server.
The units of the above embodiments may be integrated into one body, or may be separately deployed; may be combined into one unit or may be further divided into a plurality of sub-units.
In another embodiment, an electronic device is further provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement the steps of the method for edge computing live migration in the industrial internet.
In another embodiment, a computer readable storage medium is also provided, on which computer instructions are stored, which when executed by a processor, can implement the steps in the method for edge computing live migration in the industrial internet.
Fig. 9 is a schematic physical structure diagram of an electronic device according to an embodiment of the present invention. As shown in fig. 9, the electronic device may include: a processor (processor) 910, a communication Interface (Communications Interface) 920, a memory (memory) 930, and a communication bus 940, wherein the processor 910, the communication Interface 920, and the memory 930 are coupled for communication via the communication bus 940. Processor 910 may invoke logic instructions in memory 930 to perform the following method:
when a request for calculating migration sent by a mobile terminal is received, system information of a dynamic system is obtained;
acquiring the maximum value of a preset system time gain function according to the system information and the time delay carried by the request;
determining an execution main body for requesting allocation when the maximum value of a preset system time revenue function is obtained, and responding the task execution main body to the mobile terminal; the task execution subject is a mobile cloud server or a central cloud server.
Furthermore, the logic instructions in the memory 930 may be implemented in software functional units and stored in a computer readable storage medium when the logic instructions are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one position, or may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment may be implemented by software plus a necessary general hardware platform, and may also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims (9)

1. An edge computation dynamic migration method in an industrial internet is applied to a Computation Scheduling (CS) in a dynamic system, and the method comprises the following steps:
when a request for calculating migration sent by a mobile terminal is received, system information of a dynamic system is obtained;
acquiring the maximum value of a preset system time gain function according to the system information and the time delay carried by the request;
determining a task execution main body for the request distribution when the maximum value of a preset system time gain function is obtained, and responding the task execution main body to the mobile terminal; the task execution subject is a mobile cloud server or a central cloud server;
wherein, when determining to obtain the maximum value of the preset system time gain function as the task execution subject requested to be allocated, the method further comprises:
when the task execution main body is determined to be the mobile cloud server, determining the priority corresponding to the time delay carried by the request according to the mapping relation between the priority and the time delay; determining a mobile cloud server corresponding to the priority according to the mapping relation between the priority and the mobile cloud server;
the responding the task execution main body to the mobile terminal comprises:
and responding the mobile cloud server corresponding to the determined priority to the mobile terminal.
2. The method of claim 1, wherein obtaining the maximum value of the predetermined system time gain function comprises:
iterating a preset system time gain function;
and when the difference value between the value of the preset system time gain function obtained at the K +1 th time and the value of the preset system time gain function obtained at the K th time is smaller than the preset value, determining the value of the system time gain obtained at the K +1 th time as the maximum value of the preset system time gain function.
3. The method of claim 1, wherein the predetermined system time gain function is:
Figure FDA0003757834980000011
wherein, V π (s) System time gain at State s, V, when strategy π is taken π (s ') is the system time gain in state s' when strategy π is taken; r (s, a) is the system time gain at state s when action a is performed, P (s '| s, a) is the transition probability of the system state transitioning from s to s' after action a is taken, γ is the influence factor of the system state time gain transitioning to after action a is performed on the strategy π,
Figure FDA0003757834980000021
the strategy pi is a set of system states, and the task allocation execution main body is a central cloud server or a mobile cloud server;
the behavior action a belongs to a behavior action set; wherein the set of system states comprises: the method comprises the following steps of the number of mobile cloud servers in the current state, the number of mobile cloud servers allocated to tasks with priority i in a system, and one event in an event set, wherein the event set comprises: when a new computing migration request comes, the system completes a task with the priority i, the arrival of a mobile cloud server and the departure of the cloud server;
the state s and the state s' belong to a system state set; wherein the set of behavioral actions includes: and the task with the priority i reaches the system and stays on the mobile cloud server for execution processing, the newly arrived task is forwarded to the central cloud server for processing, and the system state information is updated without executing behavior actions.
4. The method of claim 3, wherein the step of determining r (s, a) comprises:
determining the difference value of the migration instant time benefit and the system time cost calculated by the system as r (s, a);
the system calculates the migration instant time gain and determines the migration instant time gain according to the difference value of the completion time of the task and the time delay carried in the request; the system time cost is determined from the desired time interval between two successive states and the time cost per unit time of the system.
5. The method of claim 4, wherein the step of determining P (s' | s, a) comprises:
and determining P (s' | s, a) according to the expected time interval between two continuous states, the arrival rate of the task, the completion rate of the task, the leaving rate and the arrival rate of the mobile cloud server.
6. The method according to any one of claims 1-5, wherein the method further comprises:
when the mobile terminal receives a task execution main body responded by the CS, if the mobile terminal corresponds to a plurality of mobile cloud servers, selecting one mobile cloud server to establish connection according to a preset rule to carry out a request of computing a task;
if the selected mobile cloud server cannot process the calculation task because of busy, selecting the next mobile cloud server according to a preset rule;
and waiting until a mobile cloud server capable of performing task computing is selected to send a computing task for performing task computing or no mobile cloud server capable of performing task computing exists, and sending the computing task to the mobile cloud server for performing task computing when a command capable of performing task computing is received from the mobile cloud server.
7. An edge computation dynamic migration apparatus in an industrial internet, which is applied to a computation scheduling CS in a dynamic system, the apparatus comprising: the device comprises a receiving unit, an acquiring unit and a determining unit;
the receiving unit is used for receiving a request for calculating migration sent by the mobile terminal;
the acquiring unit is used for acquiring the system information of the dynamic system when the receiving unit receives a request for computing migration sent by the mobile terminal; acquiring the maximum value of a preset system time gain function according to the system information and the time delay carried by the request;
the determining unit is configured to determine that the task execution main body requested to be allocated is obtained when the obtaining unit obtains the maximum value of the preset system time revenue function, and respond the task execution main body to the mobile terminal; the task execution subject is a mobile cloud server or a central cloud server;
the determining unit is further configured to determine, when the maximum value of the preset system time gain function is obtained and the task execution subject allocated to the request is determined, a priority corresponding to the time delay carried by the request according to a mapping relationship between the priority and the time delay when the task execution subject is determined to be the mobile cloud server; determining a mobile cloud server corresponding to the priority according to the mapping relation between the priority and the mobile cloud server; and responding the mobile cloud server corresponding to the determined priority to the mobile terminal.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method according to any of claims 1-6 when executing the program.
9. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the method of any one of claims 1 to 6.
CN201911199046.XA 2019-11-29 2019-11-29 Dynamic migration method and device for edge calculation in industrial internet Active CN112887347B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911199046.XA CN112887347B (en) 2019-11-29 2019-11-29 Dynamic migration method and device for edge calculation in industrial internet

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911199046.XA CN112887347B (en) 2019-11-29 2019-11-29 Dynamic migration method and device for edge calculation in industrial internet

Publications (2)

Publication Number Publication Date
CN112887347A CN112887347A (en) 2021-06-01
CN112887347B true CN112887347B (en) 2022-10-28

Family

ID=76038745

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911199046.XA Active CN112887347B (en) 2019-11-29 2019-11-29 Dynamic migration method and device for edge calculation in industrial internet

Country Status (1)

Country Link
CN (1) CN112887347B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115049347B (en) 2022-08-17 2022-12-06 成都秦川物联网科技股份有限公司 Industrial Internet of things system for AGV control and control method thereof

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107734558A (en) * 2017-10-26 2018-02-23 北京邮电大学 A kind of control of mobile edge calculations and resource regulating method based on multiserver
CN108509276A (en) * 2018-03-30 2018-09-07 南京工业大学 A kind of video task dynamic migration method in edge calculations environment
CN109831522A (en) * 2019-03-11 2019-05-31 西南交通大学 A kind of vehicle connection cloud and mist system dynamic resource Optimal Management System and method based on SMDP
CN109947545A (en) * 2019-03-11 2019-06-28 重庆邮电大学 A kind of decision-making technique of task unloading and migration based on user mobility

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107734558A (en) * 2017-10-26 2018-02-23 北京邮电大学 A kind of control of mobile edge calculations and resource regulating method based on multiserver
CN108509276A (en) * 2018-03-30 2018-09-07 南京工业大学 A kind of video task dynamic migration method in edge calculations environment
CN109831522A (en) * 2019-03-11 2019-05-31 西南交通大学 A kind of vehicle connection cloud and mist system dynamic resource Optimal Management System and method based on SMDP
CN109947545A (en) * 2019-03-11 2019-06-28 重庆邮电大学 A kind of decision-making technique of task unloading and migration based on user mobility

Also Published As

Publication number Publication date
CN112887347A (en) 2021-06-01

Similar Documents

Publication Publication Date Title
CN109669768B (en) Resource allocation and task scheduling method for edge cloud combined architecture
CN107911478B (en) Multi-user calculation unloading method and device based on chemical reaction optimization algorithm
CN112416554B (en) Task migration method and device, electronic equipment and storage medium
CN113254197A (en) Network resource scheduling method and system based on deep reinforcement learning
CN111124662B (en) Fog calculation load balancing method and system
CN109788046B (en) Multi-strategy edge computing resource scheduling method based on improved bee colony algorithm
CN107734558A (en) A kind of control of mobile edge calculations and resource regulating method based on multiserver
JP6687788B1 (en) Decentralized robot scheduling determination method, device, system, electronic device, and storage medium
CN113032120B (en) Industrial field big data task cooperative scheduling method based on edge calculation
Misra et al. Multiarmed-bandit-based decentralized computation offloading in fog-enabled IoT
CN113132456B (en) Edge cloud cooperative task scheduling method and system based on deadline perception
CN110489218A (en) Vehicle-mounted mist computing system task discharging method based on semi-Markovian decision process
CN110519370A (en) A kind of edge calculations resource allocation methods based on Facility Location Problem
CN113645273B (en) Internet of vehicles task unloading method based on service priority
Huang et al. Toward decentralized and collaborative deep learning inference for intelligent iot devices
CN115292032A (en) Task unloading method in multi-user accessed intelligent edge computing system
CN114205353A (en) Calculation unloading method based on hybrid action space reinforcement learning algorithm
CN112887347B (en) Dynamic migration method and device for edge calculation in industrial internet
CN115629865A (en) Deep learning inference task scheduling method based on edge calculation
CN114970834A (en) Task allocation method and device and electronic equipment
Wang Edge artificial intelligence-based affinity task offloading under resource adjustment in a 5G network
CN113190342A (en) Method and system architecture for multi-application fine-grained unloading of cloud-edge cooperative network
CN116886703A (en) Cloud edge end cooperative computing unloading method based on priority and reinforcement learning
Huang et al. Intelligent task migration with deep Qlearning in multi‐access edge computing
CN114172558B (en) Task unloading method based on edge calculation and unmanned aerial vehicle cluster cooperation in vehicle network

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