KR102385087B1 - Method for latency minimization in a fuzzy-based mobile edge orchestrator and system using same - Google Patents

Abstract

According to an embodiment of the present invention, a mobile edge orchestrator for minimizing latency comprises: a fuzzification unit that generates a fuzzy result value based on a plurality of fuzzy rules; an aggregating unit that determines a manner in which the fuzzy result value is coupled in a fuzzy rule group; a fuzzy inference unit including a de-fuzzification unit that generates a de-fuzzification result value (ω); and a target determination unit for determining a target server for offloading a job based on the de-fuzzification result value (ω) outputted from the de-fuzzification unit.

Description

Method for latency minimization in a fuzzy-based mobile edge orchestrator and system using same}

An embodiment according to the concept of the present invention relates to a method for minimizing delay in a fuzzy-based mobile edge orchestrator (MEO) and a system using the same, and in particular, to a task managed by a fuzzy-based mobile edge orchestrator (MEO) A method of determining a target server that minimizes delay and a system using the same.

Recently, 5G technology is developing new and diverse applications such as video streaming analysis, augmented reality (AR), Internet of Things (IoT), Internet of Vehicles (IoV), and autonomous driving. Characteristics of these applications are that they generate vast amounts of data and require real-time processing or equivalent processing speed while the user is moving.

Typical mobile devices with limited processing speed and battery life have limitations in supporting compute-intensive and latency-sensitive applications. Thus, these applications or tasks can improve quality of service (QoS) by providing users' services by relying on computing offloading and improved communication infrastructure.

Cloud computing provides various services and data-intensive analysis for mobile users through a wide area network (WAN), but the rapidly increasing IoT application data becomes an issue in managing WAN traffic capacity. In addition, since cloud computing operations are centralized and operated remotely, it is difficult to support services tailored to the situation.

To solve these challenges, edge computing that utilizes distributed resources to provide timely and contextual services is emerging. An edge server installed in a base station (BS) can replace computing that requires computation, but the limited computing resources of edge computing cannot support all devices, so a selection process is required.

In order to improve the problem, in this specification, a device, an edge server, and a cloud form a hierarchical IoT architecture. Mobile Edge Orchestration (MEO) plays a role in meeting application requirements by controlling application deployment. Using network state information and application requirements, MEO decides which mobile edge server to handle the application.

When large amounts of data need to be processed in compute-intensive applications that involve compressing and uploading data to the cloud for analysis and storage, many studies have considered partially offloading the work from the edge cloud.

The IoT system also needs to be designed so that the edge computing system can efficiently process dynamic request flows. Typical IoT applications consist of partitioned applications (microservices) that provide logically independent and timely, contextual processes.

For augmented reality applications, for example, computationally intensive components (tracers, mappers and object recognition) can be offloaded from the IoT device to the edge server while the video renderer is running locally.

These tasks assume that the BS can accurately know the channel information of users, the amount of local computing energy consumed per bit, and the size of input data of all users that can be obtained through feedback. Using this information, the BS selects the users who need offloading and then determines the size of data to offload based on the minimum weighted sum of mobile energy consumption or the minimum system delay of all users.

However, under unpredictable load fluctuations (e.g., a VM's CPU utilization frequently changes depending on the tasks running on the VM), it can be difficult to determine which edge server to handle partitioned tasks. To solve this problem, this specification uses resource information and proposes an optimal split ratio for a task that can meet application requirements.

An object of the present invention is to provide a method for minimizing the delay of an offloading task in an IoT cloud environment having a hierarchical structure and a system using the same.

Another object of the present invention is to provide a method for selecting an edge server suitable for executing offloading tasks in an IoT cloud environment with a hierarchical structure, and for optimally dividing tasks and providing a system using the same.

A mobile edge orchestrator for minimizing delay according to an embodiment of the present invention according to an embodiment of the present invention for solving the above problems is based on a plurality of fuzzy rules A fuzzy inference unit comprising a fuzzy unit generating a fuzzy result value with and a target determination unit that determines a target server to offload a job based on the defuzzification result value (ω) output from the fuzzing unit.

A method of latency minimization in a mobile edge orchestrator according to an embodiment of the present invention is a fuzzy step of generating a fuzzy decision based on a plurality of fuzzy rules, a plurality of An aggregation phase that determines how the results of a fuzzy rule are combined within a rule set, a defragment phase that performs a defragifier to produce a defragmentation result value (ω), and an operation based on the defudge result value (ω). and a target determining step of determining a target server from which to offload.

According to an embodiment of the present invention, there is an effect of minimizing the delay of each task managed by the MEO.

In addition, according to an embodiment of the present invention, it is possible to find an optimal task division ratio for the offloading task by utilizing the collected resource information and application requirements.

1 is a schematic diagram of a mobile edge cloud system according to an embodiment.
2 is a block diagram of a mobile edge orchestrator module according to an embodiment.
3 is a diagram illustrating an example of a fuzzy logic architecture and task partitioning strategy in a mobile edge orchestrator module according to an embodiment.
4 illustrates a membership function used in a mobile edge orchestrator module according to an embodiment.
5 shows an example of a central defragifier in a mobile edge orchestrator module according to an embodiment.
6 is a flowchart illustrating a mobile edge computing method using fuzzy-based mobile edge orchestration in a mobile edge cloud system according to an embodiment.
7 is a flowchart illustrating a method for determining a target server using fuzzy-based mobile edge orchestration according to an embodiment.

The features and effects of the present invention described above will become more apparent through the following detailed description in relation to the accompanying drawings, whereby those of ordinary skill in the art to which the present invention pertains can easily implement the technical idea of the present invention. will be able

Since the present invention can have various changes and can have various embodiments, specific embodiments are illustrated in the drawings and will be described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and it should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

Terms such as those defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present application. shouldn't

The suffix module, block, and part for the components used in the following description are given or mixed in consideration of only the ease of writing the specification, and do not have distinct meanings or roles by themselves.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings so that those of ordinary skill in the art can easily implement them. In the following description of embodiments of the present invention, if it is determined that a detailed description of a related known function or a known configuration may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted.

Hereinafter, a mobile edge cloud system using a mobile edge orchestration (MEO) according to the present invention will be described.

1 is a schematic diagram of a mobile edge cloud system 10 according to an embodiment. The mobile edge cloud system 10 using the edge orchestrator 100 may form a hierarchical IoT architecture by integrating the device 200 , the edge server 300 , and the cloud server 400 .

The mobile edge cloud system 10 may use an edge orchestrator 100 to manage information such as distributed mobile edge servers, available services and available resources for each server, instantiated applications, and network topologies.

The device 200 is a device of a mobile user, and includes mobile user applications such as wearables, virtual reality streaming, and mobile social media, and may be a device that requires general-purpose computational capability and continuous data processing.

The device 200 may be, for example, a cellular phone, a smart phone, a tablet PC, a Head Mounted Display (HMD) device, a smart watch, a desktop computer, a laptop computer, a digital television, a set-top box, a navigation device, a personal use device. It may include at least one of a digital assistant (PDA, Personal Digital Assistant), a portable game machine, an artificial intelligence sound reproducing device, and/or various devices capable of inputting and modifying other codes.

The edge server 300 serves as an intermediary between the cloud server 400 and the mobile user as a server equipped with computing power and storage for data processing and analysis near the mobile device 200 . Since it is located relatively closer to the user (device) than the cloud server 400 , it is possible to provide better quality of service (QoS) through edge computing.

The cloud server 400 may provide dynamic services and data analysis of mobile users using a wide area network (WAN). Cloud computing uses a cloud server or multiple remote servers to provide centralized computation and storage services. Users can use virtualized servers to access remote storage and computing facilities. Data stored in the cloud server can be retrieved anytime, anywhere, regardless of the computing power and storage capacity of the mobile terminal.

In this case, the WAN may be a local-area network (LAN) or other network communicating with the cloud server 400 .

Hereinafter, a fuzzy-based mobile edge orchestrator (MEO) according to the present invention will be described with reference to FIGS. 2 and 3 .

2 is a block diagram of a mobile edge orchestrator (or mobile edge orchestration module) 100 according to an embodiment. The mobile edge orchestrator 100 according to the present invention includes a fuzzy inference unit 110 and a target determiner 140, and the fuzzy inference unit 110 includes a fuzzy unit 111, an aggregator 112, and a defuzz It may include a fire unit 113 .

3 shows one embodiment of a fuzzy logic architecture and task partitioning strategy in a mobile edge orchestrator 100 in accordance with the present invention.

The mobile edge orchestration module 100 performs a role of satisfying application requirements by controlling application deployment. The mobile edge orchestrator 100 may determine a target server to process the application by using the network state information and application requirements.

Specifically, the fuzzy logic-based MEO recognizes the state of network resources and communication, and searches for the target (target) server of the job by using the corresponding information as an input variable. The input variables are as follows.

where w is the WAN bandwidth, l is the length of the receive operation, v is the VM utilization rate of the edge server, and d is the delay sensitivity of the related operation. The job length l is used to calculate the job execution time, and the delay sensitivity d of the job shows that the job is tolerated for a longer execution time due to network conditions or server utilization level. w and v are determined by network and resource information. In the case of WAN communication latency, network congestion is an important indicator of w for when tasks are offloaded to the cloud server. VM utilization v provides information on the remaining compute capacity of the edge server. If v is greater than the threshold level, the edge server can consider it congested.

The fuzzy inference unit 110 of the MEO may set membership functions used by the fuzzy unit 111 and the defuzzifier 113 according to the input variable. For example, the fuzzy inference unit 110 of the MEO may set four membership functions used in the fuzzy and defuzzy steps according to four input variables.

4 illustrates a membership function used by the mobile edge orchestrator module 100 according to an embodiment. 4 , the linguistic variables of w, l, and d are low (L), medium (M), and high (H), and the linguistic variables of v are light (L), normal (N) and is heaviness (H).

The fuzzy unit 111 connects the grade to each linguistic term using the membership function or fuzzification function of Equation 1 to convert a crisp value into a fuzzy value. can be converted

Equation 1

The fuzzy unit 111 may generate a fuzzy decision value selected according to each fuzzy rule based on the fuzzy rule. The fuzzy rule may be defined as an If and then rule having a condition and a conclusion. In order to determine the optimal fuzzy rule, computational experiments are repeated and a relatively better fuzzy rule is used.

The purge unit 111 may be a singleton purge unit.

The aggregation unit 112 determines a method in which fuzzy results of a plurality of fuzzy rules are combined in a fuzzy rule set by using a min (min) max (max) function according to the number of fuzzy rules. In this case, the number of fuzzy rules may be, for example, 81 (=3 ⁴ ) when the fuzzy unit uses four membership functions having three language terms.

Specifically, a fuzzy value for selecting an edge server or a cloud server is calculated using Equations 2 to 3 below.

Equation 2

Equation 3

The aggregation and activation steps use the following Equations 4 to 5.

Equation 4

Equation 5

where x, y, z, v are the crisp input parameters for a fuzzy inference system.

The defragmentation unit 113 generates a defragmentation result value. The defuzzification unit 113 may calculate a crisp output value (ω) based on μ _edge and μ _cloud . The defuzzifier 113 may be a centroid defuzzifier for calculating inference.

5 shows an embodiment of a central defuzzifier according to the present invention. Referring to FIG. 5 , the MEO uses a centroid defuzzifier that returns the center of gravity (COG) of an area under a curve.

The sharp output value ω is calculated using Equation 6 below.

Equation 6

The target determination unit 140 divides the job by the value output after the defragmentation in the defragmentation unit 113, and decides which edge server 300 or cloud server 400 to offload the divided job to. Decide which target to offload work to.

For example, the target determination unit 140 of the MEO may determine whether the task should be processed by the edge server or jointly processed by the edge server and the cloud server.

In other words, the target determination unit 140 of the MEO may make an offloading decision according to the clear output value ω. For example, if ω is greater than 50, the job will be offloaded to the cloud server, otherwise it may run on the edge server.

In addition, the mobile edge orchestrator 100 (MEO) aims to find an optimal task splitting ratio by minimizing the delay of each task managed by the MEO.

In order to minimize the delay of each task managed by the MEO, the optimization problem can be formulated as follows.

Equation 7

는 i번째 디바이스에서 작업의 전체 지연(delay)을 의미하며, 다음의 식을 이용하여 연산된다.here,

is the total delay of the task in the i-th device, and is calculated using the following equation.

Equation 8

는 모바일 디바이스에서의 전송 지연,

는 엣지 서버의 계산 지연,

는 엣지-클라우드 전송 지연, 그리고

는 클라우드 서버의 계산 지연을 나타낸다.

is the transmission delay on the mobile device,

is the computational delay of the edge server,

is the edge-to-cloud transmission delay, and

represents the computational delay of the cloud server.

)은 i번째 디바이스(사용자)가 계산 작업을 j번째 엣지 노드로 오프로드하기 위한 평균 전송 지연으로 아래의 수학식 9를 이용하여 연산된다.Specifically, the transmission delay in the mobile device (

) is the average transmission delay for the i-th device (user) to offload the calculation task to the j-th edge node, and is calculated using Equation 9 below.

Equation 9

는 작업 길이(task length)를 나타내며 지수 분포 랜덤 변수로 추정될 수 있다.where T represents the length of a time division multiple access (TDMA) frame and R is the average data rate for upload. The average data rate R can be used to calculate the average transmission delay matching the time scale of task offloading.

represents the task length and can be estimated as an exponentially distributed random variable.

)은 아래의 수학식 10를 이용하여 연산된다.Edge server computational delay (

) is calculated using Equation 10 below.

Equation 10

는 GIPS(giga-instructions per second)에서 j 번째 엣지 노드가 i번째 장치에 할당하는 계산 리소스를 나타내고,

는 작업 분할 비율(the task-splitting ratio)로, 엣지 서버에서 실행되는 데이터 부분을 나타내며,

는 해당 작업의 1 비트 데이터를 계산하는 데 필요한 CPU 사이클 수를 나타낸다.

denotes the computational resources allocated to the i-th device by the j-th edge node in giga-instructions per second (GIPS),

is the task-splitting ratio, representing the portion of data running on the edge server,

represents the number of CPU cycles required to compute one bit of data for that task.

)에 의해 결정될 수 있다 In FIG. 1 , all edge nodes are connected to a cloud server through a WAN, and all edge nodes cover a wide area geographically from a mobile edge computing (MEC) server to a cloud server. Because the high-speed WANs have minimal transmission time and do not account for data loss rates, edge-to-cloud transmission delays for long-distance WANs depend on the propagation time (the propagation time,

) can be determined by

)은 아래의 수학식 11를 이용하여 연산된다.Cloud server computational delay (

) is calculated using Equation 11 below.

Equation 11

는 GIPS(giga-instructions per second)에서 i번째 장치에 대하여 클라우드 서버에서 제공하는 클라우드 계산 리소스(the cloud computation resources)를 나타낸다.Upon receiving a task from the edge node, the MEO allocates the available computational resources of the cloud server to the task. At this time, the MEO can use the network information for parallel computing and allocate computational resources to the task by matching it with the task requirements. here,

represents the cloud computation resources provided by the cloud server for the i-th device in giga-instructions per second (GIPS).

Equation 7b means that the computational resource required for the job should not exceed the maximum CPU resource provided by the virtual machine (VM) by the j-th edge server (Bj).

)이 1이면 작업을 클라우드 서버(400)로 오프로드하지 않고 엣지 서버(300)에서 처리하며, 작업 분할 비율(

)이 0이면 작업을 모두 클라우드 서버(400)로 오프로드하는 것을 의미한다.In Equation 7c, the task split ratio (

) is 1, the job is processed on the edge server 300 without offloading to the cloud server 400, and the job split ratio (

) is 0, it means that all jobs are offloaded to the cloud server 400 .

)은 다음의 수학식 12을 이용하여 연산된다.Optimal work split ratio (

) is calculated using Equation 12 below.

Equation 12

는 엣지 서버 집합에서 최대 CPU 리소스가 있는 j번째 엣지 서버를 나타낸다.

represents the jth edge server with the maximum CPU resource in the edge server set.

)이 엣지 클라우드의 전송 지연(

)보다 작으면 최적 작업 분할 비율(

)은 1보다 커진다. 즉, 모든 작업이 엣지 노드와 연결되며, 클라우드 서버(400)로 오프로드하지 않고 엣지 서버(300)에서 처리된다.Edge server computational delay (

) is the edge cloud's transmission delay (

) is less than the optimal task split ratio (

) is greater than 1. That is, all tasks are connected to the edge node and are processed by the edge server 300 without being offloaded to the cloud server 400 .

)에 따라 MEO는 최적화 제약 조건(the optimization constraints)을 충족하는 적절한 엣지 서버를 선택하기 위해 어플리케이션 요구사항과 네트워킹 리소스를 활용할 수 있다.In addition, the optimal task split ratio (

), the MEO can utilize application requirements and networking resources to select an appropriate edge server that meets the optimization constraints.

Hereinafter, a mobile edge computing method using mobile edge orchestration (MEO) in a mobile edge cloud system according to the present invention will be described.

6 is a flowchart of a mobile edge computing method using mobile edge orchestration (MEO) according to an embodiment of the present invention.

First, each device 200 may directly send an application profile to an edge node connected through a wireless channel. The mobile edge orchestration module 100 may receive and update application information from a Metropolitan Area Network (MAN) (S110).

Next, the MEO 100 may select a target server using a fuzzy-based approach (S120). You can do this by using the resource information to match your application requirements. When the cloud server processes a job jointly with the edge server, the MEO determines the optimal job split ratio for the job (S130).

Next, each device 200 according to the determination of the MEO 100 offloads the job to the edge server 300 or the cloud server 400 (S140).

The edge server 300 or the cloud server 400 transmits the calculation result back to each device 200 (S150).

Hereinafter, with reference to FIG. 7 , a method for determining a target server for minimizing delay using mobile edge orchestration (MEO) according to the present invention will be described.

7 is a flowchart illustrating a method for minimizing delay time using mobile edge orchestration (MEO) and a method for determining a target server using the same according to an embodiment of the present invention.

Fuzzy logic-based MEO recognizes the state of network resources and communication, and uses that information as an input variable to discover the target server for the job.

The MEO sets membership functions used by the fuzzy unit and the defuzzifier according to the input variable (S210).

The fuzzy unit of the MEO may convert a crisp value into a fuzzy value by connecting the rating to each linguistic term using the membership functions or fuzzification function.

The fuzzy unit of the MEO may generate a fuzzy decision selected according to each fuzzy rule based on the fuzzy rule (S220). The fuzzy rule may be defined as an If and then rule having a condition and a conclusion. In order to determine the optimal rule, we repeat the computational experiment and use a relatively better fuzzy rule.

The aggregation unit of the MEO determines a method in which fuzzy results of a plurality of fuzzy rules are combined in a fuzzy rule set by using a min (min) max (max) function according to the number of fuzzy rules (S230).

The defuzzification unit of the MEO generates a defragmentation result value ω. The defuzzifier may calculate a crisp output value (ω) based on μ _edge and μ _cloud . The defuzzifier may be a centroid defuzzifier for calculating inferences.

The target determination unit of the MEO divides the job by the value (ω) output after the defuzzification in the defuzzification unit, and decides which edge server or cloud server to offload the divided job from, the target to offload the job from (S250). For example, the MEO's target determiner can decide whether a task should be processed on an edge server or jointly on an edge server and a cloud server.

In the above, it includes a fuzzy inference unit and a target determination unit using a fuzzy technique according to the present invention, and includes a mobile edge orchestrator and mobile edge orchestrator that minimizes delay (latency minimization), a device, an edge server, and a cloud server. Various embodiments of the mobile edge cloud system to be minimized have been described. In addition, various embodiments of a fuzzy-based delay minimization method using the same have been described.

Since the present invention can have various changes and can have various embodiments, specific embodiments are illustrated in the drawings and will be described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and it should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention.

According to the software implementation, not only the procedures and functions described in this specification but also the design and parameter optimization for each component may be implemented as a separate software module. The software code may be implemented as a software application written in a suitable programming language. The software code may be stored in a memory and executed by a controller or a processor.

Claims (16)

A fuzzy unit that generates a fuzzy result value based on a plurality of fuzzy rules;
an aggregator that determines how the fuzzy result values are combined in a fuzzy rule set; and
A fuzzy inference unit comprising a defuzzification unit generating a defragmentation result value ω, and
and a target determination unit that determines a target server to offload a job based on the defragmentation result value (ω) output from the defragmentation unit,
The target server includes an edge server or a cloud server,
latency minimization,
The delay is characterized in that it is calculated using the following equation,
A mobile edge orchestrator for latency minimization.

(here,

is the total delay of the operation on the i-th device,

is the transmission delay on the mobile device,

is the computational delay of the edge server,

is the edge-to-cloud transmission delay, and

represents the computational delay of the cloud server) delete According to claim 1,
When the target determiner includes the target server includes the edge server and the cloud server,
It is characterized by further calculating the optimal work split ratio by utilizing the application requirements and networking resources,
The optimal work division ratio is calculated using the following equation,
A mobile edge orchestrator for latency minimization.

(here,

is the computational resource allocated to the i-th device by the j-th edge node,

is the cloud computing resource provided by the cloud server for the i-th device,

is the edge-to-cloud transmission delay,

is the working length,

is the number of CPU cycles required to compute 1-bit data for that operation,

is the edge-to-cloud transmission delay, and

represents the jth edge server with the maximum CPU resource in the edge server set) 4. The method of claim 3,
The fuzzy inference unit uses at least one of WAN bandwidth, job length, edge server virtual machine (VM) usage rate, and job delay sensitivity as input variables.
A mobile edge orchestrator for latency minimization. 5. The method of claim 4,
wherein the fuzzy inference unit sets membership functions used by the fuzzy unit and the defuzzifier according to the input variable
A mobile edge orchestrator for latency minimization. delete As a latency minimization method in a mobile edge orchestrator consisting of a fuzzy inference unit and a target determination unit,
a fuzzy step in which the fuzzy inference unit generates a fuzzy decision based on a plurality of fuzzy rules;
an aggregation step in which the fuzzy inference unit determines how results of a plurality of fuzzy rules are combined in a rule set;
a defragmentation step of generating a defuzz result value (ω) by the fuzzy inference unit performing a defragfire; and
a target determination step of determining, by the target determination unit, a target server from which to offload a job based on the deferging result value (ω) generated in the defragmentation step; and
The target server includes an edge server or a cloud server as a server that minimizes operation delay,
The delay is characterized in that it is calculated using the following equation,
A method of latency minimization in MEO.

(here,

is the total delay of the operation on the i-th device,

is the transmission delay on the mobile device,

is the computational delay of the edge server,

is the edge-to-cloud transmission delay, and

represents the computational delay of the cloud server) delete 8. The method of claim 7,
When the target server includes the edge server and the cloud server,
It is characterized by further calculating the optimal work split ratio by utilizing the application requirements and networking resources,
The optimal work division ratio is calculated using the following equation,
A method of latency minimization in MEO.

(here,

is the computational resource allocated to the i-th device by the j-th edge node,

is the cloud computing resource provided by the cloud server for the i-th device,

is the edge-to-cloud transmission delay,

is the working length,

is the number of CPU cycles required to compute 1-bit data for that operation,

is the edge-to-cloud transmission delay, and

represents the jth edge server with the maximum CPU resource in the edge server set) 10. The method of claim 9,
The fuzzy inference unit uses at least one of WAN bandwidth, job length, edge server virtual machine (VM) usage rate, and job delay sensitivity as input variables.
A method of latency minimization in MEO. 11. The method of claim 10,
The fuzzy inference unit sets membership functions used in the fuzzy step and the defuzz step according to the input variable.
A method of latency minimization in MEO. delete mobile edge orchestration module,
user device,
Edge servers with computing power and storage for data processing and analysis; and
including cloud servers;
The mobile edge orchestration module
A fuzzy unit that generates a fuzzy result value based on a plurality of fuzzy rules;
an aggregator that determines how the fuzzy result values are combined in a fuzzy rule set; and
A fuzzy inference unit comprising a defuzzification unit generating a defragmentation result value ω, and
and a target determination unit that determines a target server to offload a job based on the defragmentation result value (ω) output from the defragmentation unit,
The target server includes an edge server or a cloud server,
latency minimization,
The delay is characterized in that it is calculated using the following equation,
A mobile edge cloud system that minimizes latency.

(here,

is the total delay of the operation on the i-th device,

is the transmission delay on the mobile device,

is the computational delay of the edge server,

is the edge-to-cloud transmission delay, and

represents the computational delay of the cloud server)
delete delete delete

Images