Cluster-based computing offloading system and the operating method

Abstract

The present invention provides a computing offloading system comprising a cluster head that, when a new task is assigned from either of the first or second IIoT devices, applies Q-learning reinforcement learning to each of the first and second IIoT devices and the MEC server to the first and second IIoT devices to calculate a first specific predicted value of the first IIoT device and a second specific predicted value of the second IIoT device, and compares the calculated values ​​with a set reference value, wherein the first and second specific predicted values ​​are calculated by applying Q-learning reinforcement learning to each of the first and second IIoT devices and the MEC information provided by the MEC server, respectively, and offloads the new task to either of the first and second IIoT devices and the MEC server.

Description

Technology Field

[0001] The present invention relates to a cluster-based computing offloading system and a method of operation thereof, and more specifically, to a cluster-based computing offloading system and a method of operation thereof that can determine an offloading strategy by applying reinforcement learning in a mobile edge computing (MEC) environment using D2D (Device-to-Device) communication. Background Technology

[0002] The Internet of Things (IoT) is a technology that has now become mainstream and is being applied in various fields, such as intelligent transportation and smart cities.

[0003] The Industrial Internet of Things (hereinafter referred to as 'IIoT') is part of the IoT domain, dealing with industrial devices, particularly in the manufacturing sector, where various tasks are dynamically generated due to small-batch production.

[0004] However, IIoT faces difficulties in performing tasks with only devices that have limited battery life and computing power, and to address this problem, it is adopting the Mobile Edge Computing (MEC) paradigm.

[0005] However, if there are many tasks to be processed on the MEC server (MECS), it may not be suitable to process all server tasks within latency constraints due to limited computing power and high network overhead.

[0006] Recently, IIoT has been researching technologies to determine the optimal offloading strategy in MEC environments using D2D (Device-to-Device) communication. The problem to be solved

[0007] The objective of the present invention is to provide a cluster-based computing offloading system and a method of operation thereof that can determine an offloading strategy by applying reinforcement learning in a mobile edge computing (MEC) environment using D2D (Device-to-Device) communication.

[0008] In addition, the objective of the present invention is to provide a cluster-based computing offloading system and a method of operation thereof that minimizes energy consumption and task execution delay of an IIoT device under specific delay constraints.

[0009] The objects of the present invention are not limited to those mentioned above, and other unmentioned objects and advantages of the present invention may be understood from the following description and will be more clearly understood by the embodiments of the present invention. Furthermore, it will be readily apparent that the objects and advantages of the present invention can be realized by the means and combinations thereof set forth in the claims. means of solving the problem

[0010] A cluster-based computing offloading system according to the present invention may include a Mobile Edge Computing (MEC) server providing MEC information, a first Industrial Internet of Things (IIoT) device providing first task information, a second Industrial Internet of Things (IIoT) device providing second task information within the same cluster as the first IIoT device, and a cluster head that, when a new task is assigned from either of the first or second IIoT devices, applies the first and second task information provided by the first and second IIoT devices and the MEC information provided by the MEC server to Q-learning reinforcement learning to compute a first specific predicted value of the first IIoT device and a second specific predicted value of the second IIoT device, respectively, and compares these values ​​with a set reference value, thereby offloading the new task to either of the first or second IIoT devices and the MEC server.

[0011] The first and second IIoT devices can provide the first and second task information and the new task by communicating with the cluster head via D2D (Device-to-Device).

[0012] The first task information includes first device information of the first IIoT device and a first task list for assigned tasks, and the second task information may include second device information of the second IIoT device and a second task list for assigned tasks.

[0013] The cluster head may include a communication unit that receives the new task, the first and second task information, and the MEC information; a learning unit that performs reinforcement learning on the first and second task information to calculate the first and second specific predicted values; and a decision unit that compares the first and second specific predicted values ​​with the reference value to offload the new task to either the first and second IIoT devices or the MEC server.

[0014] The first specific expected value includes energy consumption and execution delay time when a task is executed according to the first task list on the first IIoT device, and the second specific expected value may include energy consumption and execution delay time when a task is executed according to the second task list on the second IIoT device.

[0015] The above decision unit can offload the new task to an IIoT device that is lower than the reference value among the first and second specific expected values.

[0016] The above decision unit can offload the new task to the IIoT device to which the new task was assigned among the first and second IIoT devices when the first and second specific expected values ​​are lower than the reference value.

[0017] The above decision unit can offload the new task to the MEC server if the first and second specific expected values ​​are higher than the reference values.

[0018] A method of operation of a computing offloading system according to the present invention may include: receiving first task information from a first Industrial Internet of Things (IIoT) device; receiving second task information from a second Industrial Internet of Things (IIoT) device; a cluster head applying the first and second task information provided by the first and second IIoT devices and the MEC information provided by the MEC server, respectively, to Q-learning reinforcement learning to compute a first specific predicted value of the first IIoT device and a second specific predicted value of the second IIoT device; and when a new task is assigned from either of the first and second IIoT devices, comparing the first and second specific predicted values ​​with a set reference value and offloading the new task to either of the first and second IIoT devices and the MEC server.

[0019] The first and second IIoT devices can provide the first and second task information and the new task by communicating with the cluster head via D2D (Device-to-Device).

[0020] The first task information includes first device information of the first IIoT device and a first task list for assigned tasks, and the second task information may include second device information of the second IIoT device and a second task list for assigned tasks.

[0021] The first specific expected value includes energy consumption and execution delay time when a task is executed according to the first task list on the first IIoT device, and the second specific expected value may include energy consumption and execution delay time when a task is executed according to the second task list on the second IIoT device.

[0022] The step of offloading the new task above may offload the new task to an IIoT device in which either of the first or second specific expected values ​​is lower than the reference value.

[0023] The step of offloading the new task may, when the first and second specific expected values ​​are lower than the reference values, offload the new task to the IIoT device to which the new task was assigned among the first and second IIoT devices.

[0024] The step of offloading the new task above may offload the new task to the MEC server when the first and second specific expected values ​​are higher than the reference values. Effects of the invention

[0025] The cluster-based computing offloading system and method of operation according to the present invention have the advantage of reducing energy consumption and execution latency when performing tasks for each of the plurality of IIoT devices by having a cluster head offload tasks for a plurality of IIoT devices grouped into a cluster.

[0026] In addition, the cluster-based computing offloading system and the method of operation according to the present invention have the advantage of reducing complexity compared to a centralized method of managing many devices at once by grouping multiple IIoT devices.

[0027] Meanwhile, the effects of the present invention are not limited to those mentioned above, and various effects may be included within the scope obvious to a person skilled in the art from the contents described below. Brief explanation of the drawing

[0028] FIG. 1 is a simplified system diagram illustrating a cluster-based computing offloading system according to the present invention. Figure 2 is a control block diagram showing the control configuration of the cluster head shown in Figure 1. FIG. 3 is a flowchart illustrating the operation method of a cluster-based computing offloading system according to the present invention. Specific details for implementing the invention

[0029] The present invention is susceptible to various modifications and may have various embodiments, and specific embodiments are illustrated in the drawings and described in detail. However, this is not intended to limit the invention to specific embodiments, and it should be understood that the invention includes all modifications, equivalents, and substitutions that fall within the spirit and scope of the invention. Similar reference numerals have been used for similar components in the description of each drawing.

[0030] Terms such as first, second, A, B, etc., may be used to describe various components, but said components should not be limited by said terms. These terms are used solely for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be named the second component, and similarly, the second component may be named the first component. The term "and / or" includes a combination of multiple related described items or any of the multiple related described items.

[0031] When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. On the other hand, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between.

[0032] The terms used in this application are used merely to describe specific embodiments and are not intended to limit the invention. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this application, terms such as "comprising" or "having" are intended to indicate the presence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0033] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the present invention pertains. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.

[0034] Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the attached drawings.

[0035] FIG. 1 is a simplified system diagram showing a cluster-based computing offloading system according to the present invention, and FIG. 2 is a control block diagram showing the control configuration of the cluster head shown in FIG. 1.

[0036] Referring to FIGS. 1 and 2, a cluster-based computing offloading system (100) may include a Mobile Edge Computing (MEC) server (110), first and second Industrial Internet of Things (IIoT) devices (120, 130), and a cluster head (140).

[0037] In the embodiment, the first and second IIoT devices (120, 130) are shown and described as two, but the number is not limited.

[0038] First, the MEC server (110) can provide a task to be executed on the first and second IIoT devices (120, 130) in an Industrial Internet of Things environment.

[0039] At this time, the MEC server (110) can communicate with the cluster head (140) to transmit a task.

[0040] Additionally, the MEC server (110) can control the operation of the first and second IIoT devices (120, 130) and the cluster head (140) located within the same cluster, but is not limited thereto.

[0041] The first and second IIoT devices (120, 130) are fixed within a certain distance and can be clustered into the same cluster depending on the distance.

[0042] The first and second IIoT devices (120, 130) can provide first and second task information (D1, D2) by communicating D2D (Device-to-Device) with the cluster head (140).

[0043] First, the first IIoT device (120) can provide first task information (D1), including first device information of the first IIoT device and a first task list for assigned tasks, to the cluster head (140) in real time or at set time intervals.

[0044] Here, the first device information is information of a processor required to execute an assigned task, and the first task list may represent the order of tasks or a task queue of tasks offloaded from the first IIoT device (120).

[0045] The second IIoT device (130) can provide second task information (D2), including second device information of the second IIoT device and a second task list for assigned tasks, to the cluster head (140) in real time or at set time intervals.

[0046] Here, the second device information is information of a processor required to execute an assigned task, and the second task list may represent the order of tasks or a task queue of tasks offloaded from the second IIoT device (130).

[0047] The first and second IIoT devices (120, 130) can offload tasks assigned to each other via D2D (Device-to-Device) communication according to the offloading strategy of the cluster head (140), but are not limited thereto.

[0048] Additionally, the first and second IIoT devices (120, 130) can have tasks directly offloaded from the MEC server (110) according to the offloading strategy of the cluster head (140).

[0049] The cluster head (140) may include a communication unit (142), a learning unit (144), and a decision unit (146).

[0050] The communication unit (142) communicates with the MEC server (110) via a wireless communication network and can perform D2D communication with the first and second IIoT devices (120, 130).

[0051] The communication unit (142) can receive a New Task from either of the first or second IIoT devices (120, 130).

[0052] Additionally, the communication unit (142) can receive first and second task information (D1, D2) from each of the first and second IIoT devices (120, 130) and MEC information from the MEC server (110).

[0053] The learning unit (144) can apply the first and second task information and the MEC information, respectively, to Q-learning reinforcement learning to calculate the first and second specific predicted values ​​(Q1, Q2).

[0054] Here, the first and second IIoT devices (120, 130) can generate tasks with delay constraints given by a Poisson distribution in time slots, and it is assumed that the tasks are not assigned and are offloaded locally within the cluster.

[0055] The learning unit (144) can learn the first and second task lists assigned to each of the first and second IIoT devices (120, 130) by dividing them into time slots. That is, the learning unit (142) can be composed of a tuple having three attributes for each task (cd(t)) generated by the first and second IIoT devices (120, 130) in the time slot, in the set of tasks generated in the time slot.

[0056] That is, cd(t)=(vd(t),wd(t),Lreq), and

[0057] Here, vd(t) is the task size, wd(t) is the number of processor (CPU) cycles used per bit to process the task, and Lreq may be the task's delay constraint.

[0058] Task execution delay is defined differently depending on the task offloading strategy, and to calculate the task execution delay, the task transmission speed is applied as an essential factor when the task is transmitted to the MEC server (110) or another IIoT device.

[0059] The above task transfer speed may vary depending on the offload location of the first and second IIoT devices (120, 130). The task transfer speed is calculated using the Shannon-Hartley formula, and

[0060] The task transfer speed in a time slot can be explained as shown in [Equation 1] below.

[0061]

[0062] Here, Bcell(t) is the channel bandwidth of wireless communication, e.g., cellular communication, in time slot (t), pcelld,m(t) is the power consumed by the IIoT device to transmit data to the MEC server (110) with minimum load, hcelld,m(t) is the channel gain between the IIoT device and the MEC server (110), and σ2 may be the noise power.

[0063] The operation transmission speed between the first and second IIoT devices (120, 130) in a time slot (t) is explained by [Equation 2] below.

[0064]

[0065] Here, Bd2d(t) may be the channel bandwidth of D2D communication in time slot (t), pd2di,j(t) may be the power consumed by i, the first IIoT device (120), to transmit data to j, the second IIoT device (130), at minimum load, and hd2di,j(t) may be the channel gain of the first and second IIoT devices (120, 130).

[0066] The execution delay time (Ld(t)) of a task generated by either of the first or second IIoT devices (120, 130) in time slot (t) can be determined by the following [Equation 3].

[0067]

[0068] Here, od(t) is an offloading strategy for cd(t), and od(t) indicates that it is determined by the cluster head (140) of the cluster where the task occurred.

[0069] od(t)=0 indicates that cd(t) is executed locally within the cluster on the IIoT device where it first occurred, and od(t)=1 indicates that cd(t) is offloaded to the MEC server (110) with minimal load. Additionally, od(t)=2 indicates that cd(t) is executed by the IIoT device with minimal load.

[0070] sd(t) is wd(t)vd(t), which is the number of processor (CPU) cycles of the IIoT device required to execute the task, and fd, fj, and fm may be the computing power of the first and second IIoT devices (120, 130) and the MEC server (110), respectively.

[0071] Additionally, Wd(t), Wj(t), and Wm(t) are each the waiting delays of the first and second IIoT devices (120, 130) and the MEC server (110) in time slot (t), and the energy consumption of the first and second IIoT devices (120, 130) can be defined differently according to the offloading strategy based on the execution delay time.

[0072] When a task generated by the first and second IIoT devices (120, 130) in time slot (t) is executed according to the offloading strategy od(t), the energy consumption for the first and second IIoT devices (120, 130) is represented as [Equation 4] below.

[0073]

[0074] Here, ε·(fd) 2 represents the energy consumption per processor (CPU) cycle of each of the first and second IIoT devices (120, 130), and ε is a constant that varies depending on the hardware architecture.

[0075] As described above, the learning unit (144) can calculate first and second specific predicted values ​​(Q1, Q2) including execution delay time and energy consumption when executing a task according to [Equation 1] to [Equation 4] based on first and second device information and first and second task lists included in first and second task information (D1, D2).

[0076] The decision unit (146) can compare the first and second specific expected values ​​(Q1, Q2) with the set reference values ​​to offload a new task to either the first and second IIoT devices (120, 130) or the MEC server (110).

[0077] That is, the decision unit (146) can offload a new task to an IIoT device that is lower than the reference value among the first and second specific expected values ​​(Q1, Q2).

[0078] Additionally, the decision unit (146) can offload the New Task to the IIoT device to which the New Task was assigned among the first and second IIoT devices (Q1, Q2) when the first and second specific expected values ​​(Q1, Q2) are lower than the reference value.

[0079] Additionally, the decision unit (146) can offload a new task to the MEC server (110) if the first and second specific expected values ​​(Q1, Q2) are higher than the reference value.

[0080] The decision unit (146) can offload the New Task so that task blocking and queue delay by the first and second IIoT devices (120, 130) are reduced when executing the New Task.

[0081] FIG. 3 is a flowchart illustrating the operation method of a cluster-based computing offloading system according to the present invention.

[0082] Referring to FIG. 3, the cluster head (140) of the cluster-based computing offloading system (100) receives first task information (D1) from the first IIoT device (120) (S110) and can receive second task information (D2) from the second IIoT device (130) (S120).

[0083] That is, the communication unit (142) included in the cluster head (140) can receive first and second task information (D1, D2) from each of the first and second IIoT devices (120, 130).

[0084] The cluster head (140) can apply the first and second task information (D1, D2) provided by the first and second IIoT devices (120, 130) to Q-learning reinforcement learning to compute the first specific predicted value (Q1) of the first IIoT device (120) and the second specific predicted value (Q2) of the second IIoT device (130) (S130).

[0085] That is, the learning unit (144) included in the cluster head (140) can learn the first and second task lists assigned to each of the first and second IIoT devices (120, 130) by dividing them into time slots.

[0086] The learning unit (144) can calculate first and second specific predicted values ​​(Q1, Q2) including execution delay time and energy consumption when executing a task according to [Equation 1] to [Equation 4] based on first and second device information and first and second task lists included in first and second task information (D1, D2).

[0087] The cluster head (140) is assigned a new task from either of the first and second IIoT devices (120, 130) (S140), and by comparing the first and second specific expected values ​​(Q1, Q2) with a set reference value, the new task can be offloaded to either of the first and second IIoT devices (120, 130) and the MEC server (110) (S150).

[0088] That is, the decision unit (146) included in the cluster head (140) can compare the first and second specific expected values ​​(Q1, Q2) with a set reference value and offload a new task to either the first and second IIoT devices (120, 130) or the MEC server (110).

[0089] That is, the decision unit (146) can offload a new task to an IIoT device that is lower than the reference value among the first and second specific expected values ​​(Q1, Q2).

[0090] Additionally, the decision unit (146) can offload the New Task to the IIoT device to which the New Task was assigned among the first and second IIoT devices (Q1, Q2) when the first and second specific expected values ​​(Q1, Q2) are lower than the reference value.

[0091] Additionally, the decision unit (146) can offload a new task to the MEC server (110) if the first and second specific expected values ​​(Q1, Q2) are higher than the reference value.

[0092] In an embodiment, the cluster head (140) can collect device information and MEC information by time slot, learn new tasks generated in the slot, and then determine offloading.

[0093] The features, structures, effects, etc. described in the embodiments above are included in at least one embodiment of the present invention and are not necessarily limited to only one embodiment. Furthermore, the features, structures, effects, etc. exemplified in each embodiment may be combined or modified and implemented in other embodiments by a person skilled in the art to which the embodiments belong. Therefore, details regarding such combinations and modifications should be interpreted as being included within the scope of the present invention.

[0094] Furthermore, although the above description has focused on exemplary embodiments, this is merely illustrative and does not limit the invention. Those skilled in the art will understand that various modifications and applications not exemplified above are possible within the scope of the essential characteristics of the exemplary embodiments. For instance, each component specifically shown in the exemplary embodiments may be modified. Furthermore, differences related to such modifications and applications should be interpreted as being included within the scope of the invention as defined in the appended claims.

Claims

Claim 1 A Mobile Edge Computing (MEC) server providing MEC information; a first Industrial Internet of Things (IIoT) device providing first task information; and a second Industrial Internet of Things (IIoT) device providing second task information within the same cluster as the first IIoT device; When a new task is assigned from either of the first and second IIoT devices, the cluster head includes a first specific predicted value of the first IIoT device and a second specific predicted value of the second IIoT device, which are computed by applying Q-learning reinforcement learning to each of the first and second task information provided by the first and second IIoT devices and the MEC information provided by the MEC server, respectively, and offloading the new task to either of the first and second IIoT devices and the MEC server; the first task information includes first device information of the first IIoT device and a first task list for assigned tasks; the second task information includes second device information of the second IIoT device and a second task list for assigned tasks; and the cluster head includes a communication unit that receives the new task, the first and second task information, and the MEC information; and a component that computes the first and second specific predicted values ​​by applying reinforcement learning to the first and second task information and the MEC information. Learning Department;A computing offloading system comprising a decision unit that compares the first and second specific expected values ​​with the reference value to offload the new task to either the first or second IIoT device or the MEC server, wherein the first specific expected value includes energy consumption and execution delay time when a task is executed according to the first task list on the first IIoT device, and the second specific expected value includes energy consumption and execution delay time when a task is executed according to the second task list on the second IIoT device, wherein the energy consumption is calculated by Equation 4 and the execution delay time is calculated by Equation 3. [Equation 4]; Here, is the energy consumption of an industrial IoT device d in time slot t, and is the computational capability of the above device d, and is a constant that depends on the hardware architecture of the above device d, and is the number of processor cycles of the above device d required to execute the task, and is the task's work size, and is the minimum load power consumption for transmitting data from the device d to the MEC server in time slot t, and is the task transfer rate from the device d to the MEC server in time slot t, and is the computational power of industrial IoT device j, and is the minimum load power consumption for transmitting data from device d to device j in time slot t, and is the task transmission rate from device d to device j in time slot t. Also, is an offloading strategy for tasks, and If this is 0, it is executed locally on the above device d where the first task occurred, and If this is 1, the task is offloaded from the device d to the MEC server with minimum load, and If this is 2, the task is offloaded from device d to device j with minimum load.[Equation 3] Here, is the execution delay time of the device d at time slot t, and is the computing power of the above MEC server, and , , is the waiting delay of the device d, the device j, and the MEC server at time slot t, respectively. Claim 2 In claim 1, the computing offloading system wherein the first and second IIoT devices communicate D2D (Device-to-Device) with the cluster head to provide the first and second task information and the new task. Claim 3 delete Claim 4 delete Claim 5 delete Claim 6 In claim 1, the determining unit offloads the new task to an IIoT device lower than the reference value among the first and second specific expected values, a computing offloading system. Claim 7 A computing offloading system according to claim 1, wherein the determining unit offloads the new task to the IIoT device to which the new task is assigned among the first and second IIoT devices when the first and second specific expected values ​​are lower than the reference value. Claim 8 In claim 1, the determining unit offloads the new task to the MEC server when the first and second specific expected values ​​are higher than the reference value, a computing offloading system. Claim 9 A step of receiving first task information from a first Industrial Internet of Things (IIoT) device; a step of receiving second task information from a second Industrial Internet of Things (IIoT) device; a step in which a cluster head applies the first and second task information provided by the first and second IIoT devices, respectively, and the MEC information provided by the MEC server to Q-learning reinforcement learning to compute a first specific predicted value of the first IIoT device and a second specific predicted value of the second IIoT device; A method of operation of a computing offloading system, wherein when a new task is assigned from either of the first or second IIoT devices, the method includes the step of offloading the new task to either of the first or second IIoT devices and an MEC server by comparing the first or second specific expected values ​​with a set reference value, wherein the first task information includes first device information of the first IIoT device and a first task list for assigned tasks, the second task information includes second device information of the second IIoT device and a second task list for assigned tasks, the first specific expected value includes energy consumption and execution delay time when a task is executed according to the first task list in the first IIoT device, and the second specific expected value includes energy consumption and execution delay time when a task is executed according to the second task list in the second IIoT device, wherein the energy consumption is calculated by Equation 4 and the execution delay time is calculated by Equation 3. [Equation 4] Here, is the energy consumption of an industrial IoT device d in time slot t, and is the computational capability of the above device d, and is a constant that depends on the hardware architecture of the above device d, and is the number of processor cycles of the above device d required to execute the task, and is the task's work size, and is the minimum load power consumption for transmitting data from the device d to the MEC server in time slot t, and is the task transfer rate from the device d to the MEC server in time slot t, and is the computational power of industrial IoT device j, and is the minimum load power consumption for transmitting data from device d to device j in time slot t, and is the task transmission rate from device d to device j in time slot t. Also, is an offloading strategy for tasks, and If this is 0, it is executed locally on the above device d where the first task occurred, and If this is 1, the task is offloaded from the device d to the MEC server with minimum load, and If this is 2, the task is offloaded from device d to device j with minimum load.[Equation 3] Here, is the execution delay time of the device d at time slot t, and is the computing power of the above MEC server, and , , is the waiting delay of the device d, the device j, and the MEC server at time slot t, respectively. Claim 10 In claim 9, the method of operation of a computing offloading system wherein the first and second IIoT devices communicate D2D (Device-to-Device) with the cluster head to provide the first and second task information and the new task. Claim 11 delete Claim 12 delete Claim 13 In claim 10, the step of offloading the new task is to offload the new task to an IIoT device in which either of the first or second specific expected values ​​is lower than the reference value, a method of operation of a computing offloading system. Claim 14 In claim 10, the step of offloading the new task is to offload the new task to the IIoT device to which the new task is assigned among the first and second IIoT devices when the first and second specific expected values ​​are lower than the reference values, a method of operation of a computing offloading system. Claim 15 In claim 10, the step of offloading the new task is to offload the new task to the MEC server when the first and second specific expected values ​​are higher than the reference values, a method of operation of a computing offloading system.

Images