CN112491964B

CN112491964B - Mobile assisted edge calculation method, apparatus, medium, and device

Info

Publication number: CN112491964B
Application number: CN202011212357.8A
Authority: CN
Inventors: 郭得科; 谷思远; 廖汉龙; 罗来龙; 李欣奕
Original assignee: National University of Defense Technology
Current assignee: National University of Defense Technology
Priority date: 2020-11-03
Filing date: 2020-11-03
Publication date: 2022-05-31
Anticipated expiration: 2040-11-03
Also published as: CN112491964A

Abstract

The invention provides a mobile assisted edge computing method, a device, a medium and equipment, which are applied to the Internet of things at least comprising a fixed edge node and a mobile edge node, wherein the method comprises the following steps: requesting an edge task from a fixed edge node according to a user request and resource distribution, wherein the edge task is a user request which cannot be responded by the fixed edge node; auctioning the edge tasks to a plurality of mobile edge nodes according to an auction mechanism, and determining a final pricing strategy according to the income of each mobile edge node and the expenditure of a fixed edge node; and determining the global distribution strategy of the edge task at each mobile edge node by combining the pricing strategy and the pricing of the mobile edge nodes. The invention firstly provides a mobile auxiliary edge computing framework, and improves the service quality of a fixed edge node by introducing a mobile edge node.

Description

Mobile assisted edge calculation method, apparatus, medium, and device

Technical Field

The exemplary embodiment of the invention relates to the technical field of internet of things, in particular to a mobile assisted edge computing method, device, medium and equipment.

Background

As internet of things devices have become a place in the market, more and more new internet of things applications such as mobile games, video processing, and image recognition have emerged and are frequently used. This type of application is typically delay sensitive, resource starvation, and computationally intensive. However, internet of things devices tend to be designed to be more flexible and mobile, which in turn limits the processing power of the device. To address this challenge, edge computing has become a promising paradigm that caches services and provides cloud-like resources through fixed edge nodes (also known as edge clouds, micro-clouds, carry-on clouds, fog) on the edge of the network. Each edge node is typically composed of an access point (or base station) and an edge server.

However, even with the help of fixed edge nodes, the low latency requirements of internet of things applications are still difficult to guarantee anytime and anywhere. According to observations, the root cause may be a supply-and-demand mismatch between the resource supply capabilities of the fixed edge nodes and the end user's needs. The reasons for this mismatch can be roughly classified into the following two categories: 1) the resources (e.g., computing and storage resources) of existing fixed edge nodes cannot meet the needs of the end user. From the fixed edge node deployment perspective, it is difficult for a single edge node to handle all surrounding user requests due to the limited service cache space and computational resources of the fixed edge node. From the construction of existing edge computing environments, increasing the resource capacity of individual nodes or adding new edge nodes results in additional budgets. If the budget does not meet the expected revenue, the construction will be abandoned. This results in the total amount of resources at the edge end not keeping up with the user's resource needs for a long time; 2) when the user's demand for a certain type of application peaks, the resources of some fixed edge nodes may not be able to meet the end user's demand. For example, a large number of application demands occur in a certain area, but the adjacent fixed edge nodes do not cache the corresponding application services in advance. This may require waiting for re-caching or selecting other solutions. Alternatively, even if an edge node pre-caches such services, the edge node's computing resources may still be scarce. The two reasons are different in that the former emphasizes the shortage of the entire resource in terms of temporal and spatial distribution, while the latter emphasizes the shortage of the resource due to dynamic demand.

To address the supply-demand mismatch problem, many solutions attempt to make up for the shortfall of fixed edge node resources. To date, some hot issues in the field of edge computing focus on how to prefetch and cache corresponding services from a remote cloud to a fixed edge node (also referred to as service provisioning, service placement) to quickly respond to user requests. However, frequent caching affects the stability of the network and creates additional costs. In addition, some work has considered using device-to-device collaboration, i.e., a d2d fog, to respond to some end user requests. However, it is difficult to ensure security and establish trust between the end user and other resource providing devices. In addition, there are efforts to schedule excessive user requests over the backhaul network that are temporarily not handled by the fixed edge node to neighboring fixed edge nodes for processing, which can result in unpredictable delays. In the worst case, when the total resources of the fixed edge node cannot meet the amount of resources required by the end user, many requests will be scheduled to the remote cloud for processing, which will result in greater delay.

In addition to the above observations, another observation is about mobile devices, including smart cars, drones, robots, and the like. They all have the following properties: 1) considerable resources. The deployment of mobile devices is extensive and bulky. Each mobile device has built-in services (also known as API toolkits, data and programs) and sufficient computing resources. More importantly, mobile devices typically have idle periods. 2) And (4) mobility. With the rise of smart cities and smart transportation, more and more devices are beginning to have great mobility, i.e. mobile devices. They can move their location continuously according to the needs of the user, providing their resources across the area.

Solving the supply and demand mismatch problem at the edge of the network is very challenging because: 1) in view of cost, it is important to stimulate potential mobile devices as mobile edge nodes to satisfy a user's request. 2) How to assign portions of the requests to the appropriate fixed edge nodes given a set of user requests, and how to schedule available mobile edge nodes to assist the appropriate fixed edge nodes to handle other overloaded requests. 3) The incentive for a large number of potential mobile edge nodes with a particular service to move to a particular area in response to a user request remains an open problem.

Disclosure of Invention

In view of this, an object of an exemplary embodiment of the present invention is to provide a method, an apparatus, a medium, and a device for mobile assisted edge computation, so as to solve the problem of insufficient computation of a fixed edge node in the current internet of things.

In view of the above, an exemplary embodiment of the present invention provides a mobile assisted edge computing method, applied to an internet of things including at least a fixed edge node and a mobile edge node, where the method includes:

requesting an edge task from a fixed edge node according to a user request and resource distribution, wherein the edge task is a user request which cannot be responded by the fixed edge node;

auctioning the edge tasks to a plurality of mobile edge nodes according to an auction mechanism, and determining a final pricing strategy according to the income of each mobile edge node and the expense of a fixed edge node;

and determining the global distribution strategy of the edge task at each mobile edge node by combining the pricing strategy and the pricing of the mobile edge nodes.

In another possible implementation manner of the embodiment of the present invention, in combination with the above description, the method further includes:

and acquiring the bid of each mobile edge node, the time for the mobile edge node to complete the edge task and the auction price of the fixed edge node, and ensuring the profit maximization of the fixed edge node through bipartite graph matching.

In another possible implementation manner of this embodiment of the present invention, before auctioning the edge task to a plurality of mobile edge nodes according to an auction mechanism, the method further includes:

predicting resources of the mobile equipment joining the Internet of things, and judging whether the mobile equipment can become the mobile edge node according to a prediction result;

when the resources of the mobile device are capable of running the edge task, the mobile device becomes the mobile edge node, otherwise it is discarded.

and performing task allocation according to the global allocation strategy, processing the allocated edge tasks by the mobile edge nodes which are won in the auction mechanism, and returning the processing results of the edge tasks to the end users who send the edge tasks.

during the auction mechanism, sending a participation proof to the fixed edge node by each mobile edge node;

when the mobile edge node wins the auction abandons the edge task, reducing the reputation parameter of the corresponding mobile edge node according to the participation proof;

and when the mobile edge node winning in the auction does not finish the distributed edge task within the guaranteed time range, determining that the corresponding mobile edge node makes compensation for the fixed edge node according to the participation certification and the compensation coefficient.

and when the edge tasks distributed according to the global distribution strategy are processed and finished, the fixed nodes pay the workload to the corresponding mobile edge nodes according to the distribution certification sent by the fixed edge nodes to the corresponding mobile edge nodes.

In a second aspect, an exemplary embodiment of the present invention further provides a mobile assisted edge computing apparatus applied to an internet of things including at least a fixed edge node and a mobile edge node, the apparatus including:

the edge task determining module is used for requesting an edge task from a fixed edge node according to a user request and resource distribution, wherein the edge task is a user request which cannot be responded by the fixed edge node;

the auction module is used for auctioning the edge tasks to a plurality of mobile edge nodes according to an auction mechanism, and determining a final pricing strategy according to the income of each mobile edge node and the expenditure of the fixed edge node;

and the distribution module is used for determining the global distribution strategy of the edge task at each mobile edge node by combining the pricing strategy and the pricing of the mobile edge nodes.

The above apparatus, further comprising:

and the calculation module is used for acquiring the bid of each mobile edge node, the time for the mobile edge node to finish the edge task and the auction price of the fixed edge node, and ensuring the profit maximization of the fixed edge node through bipartite graph matching.

In a third aspect, an exemplary embodiment of the present invention further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the above-mentioned method for calculating a moving auxiliary edge when executing the computer program.

In a fourth aspect, exemplary embodiments of the present invention also provide a non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform the above-described movement-assisted edge calculation method.

As can be seen from the foregoing, in order to solve the problems of mismatch between supply and demand and long service supply period during peak hours, the method according to the exemplary embodiment of the present invention adopts a mobile assisted edge computing framework, and uses a mobile edge node to improve the service quality of a fixed edge node, and in this process, the present invention also adopts a CRI (credible, reciprocal and incentive) auction mechanism to incentivize the mobile edge node to participate in the service requested by the user, thereby achieving higher task completion rate, profit maximization and computing efficiency.

Drawings

In order to more clearly illustrate the exemplary embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only exemplary embodiments of the present invention, and for those skilled in the art, other drawings may be obtained based on these drawings without inventive effort.

FIG. 1 is a schematic basic flow chart of a method for calculating a mobile assisted edge according to an exemplary embodiment of the invention;

FIG. 2 is a schematic diagram of an edge computing framework composition in accordance with an exemplary embodiment of the present invention;

FIG. 3 is a task allocation flow diagram of an exemplary embodiment of the invention;

FIG. 4(a) is an exemplary schematic diagram of an auction in accordance with an exemplary embodiment of the present invention;

FIG. 4(b) is a schematic diagram of an exemplary auction using a bipartite graph according to an exemplary embodiment of the invention;

FIG. 5(a) is a schematic supply and demand diagram of an exemplary embodiment of the present invention;

FIG. 5(b) is a parameter setting diagram of an exemplary embodiment of the present invention;

FIG. 6(a) is a schematic diagram of a change in task completion rate for an exemplary embodiment of the present invention;

FIG. 6(b) is a diagram illustrating the profit variation of an operator according to an exemplary embodiment of the present invention;

FIG. 7(a) is a diagram illustrating mobile edge node value and bidding that is being won in an exemplary embodiment of the present invention;

FIG. 7(b) is a diagram illustrating winning mobile edge node bidding and cost according to an exemplary embodiment of the present invention;

fig. 8(a) is a schematic diagram of CRI authenticity according to an exemplary embodiment of the present invention;

FIG. 8(b) is a schematic illustration of benefit maximization for an exemplary embodiment of the present invention;

FIG. 9 is a schematic diagram of a basic structure of a mobile assisted border computing apparatus according to an exemplary embodiment of the present invention;

fig. 10 is a schematic diagram of an apparatus according to an exemplary embodiment of the present invention.

Detailed Description

To make the objects, technical solutions and advantages of the present disclosure more apparent, the present disclosure will be described in further detail below with reference to specific embodiments and the accompanying drawings.

It should be noted that technical terms or scientific terms used in the exemplary embodiments of the present invention should have a general meaning as understood by those having ordinary skill in the art to which the present disclosure pertains, unless otherwise defined. The use of "first," "second," and similar terms in the exemplary embodiments of the invention are not intended to indicate any order, quantity, or importance, but rather are used to distinguish one element from another. The word "comprising" or "comprises", and the like, means that the element or item preceding the word comprises the element or item listed after the word and its equivalent, but does not exclude other elements or items. The terms "connected" or "coupled" and the like are not restricted to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "upper", "lower", "left", "right", and the like are used merely to indicate relative positional relationships, and when the absolute position of the object being described is changed, the relative positional relationships may also be changed accordingly.

Summary of the concept: although existing edge computing models aim to provide a low latency experience for users by pre-caching cloud-like service functionality at the edge of the network. However, there is still much room for improvement with respect to supply and demand mismatch. In order to achieve flexible service caching at the network edge, there is often one operator responsible for caching decisions. The operator is connected to all fixed edge nodes through a backhaul network. When an edge computing environment is overloaded and cannot handle certain user requests, operators tend to schedule it to a remote cloud with sufficient processing power. However, such a decision may result in some requests failing to meet the latency requirements. And this has deviated from the original intention of edge computing to provide low-latency services, losing the advantage over cloud computing. To solve this problem, the exemplary embodiments of the present invention first propose to fully utilize potential mobile edge nodes (i.e., mobile devices with idle resources), which are different from fixed edge nodes, such as fixed base stations or access points. Exemplary embodiments of the present invention attempt to integrate a mobile edge node into an existing edge computing environment to more efficiently address current supply and demand mismatch issues with its mobility.

The embodiment may be applied to an intelligent terminal/server with a central processing module for performing mobile assisted edge calculation, where the method may be executed by a device serving as a fixed edge node, where the device may be implemented by software and/or hardware, and may be generally integrated in a mobile terminal, as shown in fig. 1, which is a basic flow diagram of the mobile assisted edge calculation method of the present invention, and the method specifically includes the following steps:

in step 110, according to the user request and the resource distribution, requesting an edge task from a fixed edge node, where the edge task is a user request that the fixed edge node cannot respond to;

in step 120, auctioning the edge tasks to a plurality of mobile edge nodes according to an auction mechanism, and determining a final pricing strategy according to the earnings of the mobile edge nodes and the expenses of the fixed edge nodes;

in step 130, a global allocation policy of the edge tasks at each of the mobile edge nodes is determined in combination with the pricing policy and the pricing of the mobile edge nodes.

The following description is made with reference to specific embodiments:

as shown in fig. 2, the moving edge assistance framework in an exemplary embodiment of the present invention is an improvement on an existing edge computing framework, which includes: the system comprises a group of fixed edge nodes, wherein each node consists of an Access Point (AP) and an edge server. In addition, there are some end users, one operator, and remote clouds in the figure. Based on the existing edge computing framework, the exemplary embodiment of the present invention provides a mobile assisted edge computing framework, and as shown in fig. 1, the mobile assisted edge computing framework of the exemplary embodiment of the present invention further includes a set of mobile edge nodes, where the edge nodes include some mobile devices, such as smart cars, drones, and robots, which can generally receive and send information through a wireless network, and the mobile devices have functions or functions similar to those of access points, and have certain computing and storage resources, and can serve as servers of fixed edge nodes, and a set of mobile edge nodes is used to solve the above mismatch problem.

It should be noted that the category of the mobile edge node is not fixed, and is not limited to the smart car, the drone and the robot, and any mobile device with an access function, a certain processing capability and mobility may be used as the mobile edge node in the mobile assisted edge computing framework according to the exemplary embodiment of the present invention.

The method of the exemplary embodiment of the present invention, first, edge computing and mobile assistance, can further perform data processing and analysis faster and in real time, so that the data processing is closer to the source rather than to an external data center or cloud, further reducing the delay time, and greatly reducing the cost budget in cost prediction, and the cost of the data management solution of an enterprise on a local device is much lower than that of the cloud or data network center.

Secondly, the method can reduce network flow, and with the increase of the Internet of things equipment, the speed of generating data in a geometric data level index is increased, the network bandwidth becomes more limited, and cloud is overwhelmed, so that a larger data bottleneck is caused.

Thirdly, the method of the invention improves the efficiency of the application program, can continuously learn by reducing the delay and by edge calculation, and can adjust the model according to the individual requirements to bring personalized experience.

The exemplary embodiment of the present invention emphasizes that the operator makes the service buffering decision in a fixed time frame and the request scheduling decision in a time slice. The time slices are much smaller than the time frames, both controlled by the operator. When the entire edge computing environment is overloaded or some requested services are not pre-cached from the remote cloud, some user requests will not be processed at the network edge, but the operator will give these end users another opportunity to be processed at the network edge instead of the remote cloud in the new edge computing framework employed by the present invention.

Within this framework, exemplary implementations of the inventionIn an example, a fixed edge node set is denoted by N ═ 1, 2. The starting time point of each time slice is described using the consecutive integer T ═ {0,1, 2. }. Note that the continuous time is divided into equal time slices, and the time slices are continuous. In an exemplary embodiment of the present invention, all services that may be requested in an edge computing environment are also denoted by S ═ 1, 2. At time slice t, in an exemplary embodiment of the present invention, M^t＝{1^t,2^t,...,m^t,...,M^tDenotes the moving edge node and is denoted by U^t＝{1^t,2^t,...,U^tDenotes user requests that cannot be handled in time at the fixed edge node, i.e. waiting user requests. For each user request i^t∈U^tThe operator will give him an opportunity to handle at the network edge once with the help of CRI.

For ease of reference, the prime notation herein is listed in table 1 in the exemplary embodiments of the present invention.

Similar resources involved in this mobile assisted edge computing framework for fixed edge nodes and mobile edge nodes are as follows:

1) network link: it must be emphasized that each end user or mobile edge node in the framework can establish a cellular link with the AP. In addition, it is assumed in the exemplary embodiment of the present invention that the bandwidth of the era 5G is sufficient for them to establish a link. Furthermore, it is also contemplated in exemplary embodiments of the present invention that the operator connects to the fixed edge node through a backhaul network, which provides for considerably faster routing. Remote cloud caching all services required at the edge of the network are typically deployed remotely from the end user. It can be connected to fixed edge nodes through the backbone network, but the latency is large and unacceptable for latency sensitive user requests. In an exemplary embodiment of the invention, scheduling to a remote cloud is a very time consuming option.

2) The caching capacity is as follows: all fixed edge nodes design flexible service caching capabilities at the network edge. They implement a fast cache function through technologies such as containers or virtual machines. This functionality means the use of toolboxes, settings, algorithms, etc. to cache different services quickly. Unlike a fixed edge node, a mobile edge node has a specific service type and therefore can only implement specific functions. Given that the mobile edge nodes have pre-deployed specific services, this appears to greatly limit the application scenarios, as there is no guarantee in exemplary embodiments of the present invention that there will always be some suitable mobile smart cars, drones, and robots that need to move into proximity. However, the total number of choices is better than the number of choices. Preferably, in natural disaster emergencies, battlefields and future urban scenarios, a large number of mobile devices will have greater support for existing edge servers. In addition, the remote cloud caches all needed services with rich storage resources. In contrast, the storage resources of the fixed edge node and the mobile edge node are rather limited and cannot cache too much service. It is emphasized in exemplary embodiments of the present invention that an edge node may pre-cache some services per time frame with its flexible caching capabilities.

3) Computing power: in general, the computational power of the fixed edge node and the mobile edge node refers to the ability to process user requests. This is related to the performance of CPU, GPU, TPU etc. It relates to the amount of tasks processed per unit time or the time required to process the same request. In the exemplary embodiment of the invention, the user request U which cannot be processed by the fixed edge node in each time slice is processed^tThe tasks to be processed are considered to be potential mobile edge nodes.

In order to incentivize voluntary participation by mobile edge nodes in edge computing environments, a trusted, reciprocal, and incentivized CRI auction mechanism is devised in exemplary embodiments of the invention. CRI enables a reciprocal relationship between mobile edge nodes and operators in a rewarding manner, thus forming a marketing mechanism. A basic observation is that as a member of the market environment, the mobile edge node is always seeking maximization of its own benefits. Since computing resources are idle, there are incentive mechanisms in the market environment if profitable, and mobile edge nodes tend to handle operator-scheduled tasks.

In an edge computing environment, five types of participants are cautiously considered in exemplary embodiments of the present invention, including end users, service providers, infrastructure providers, mobile edge nodes, and remote clouds:

1) the end user seeks the best experience on behalf of the holder of the user equipment.

2) Service providers are responsible for providing a wide range of services to end users. The service provider may deploy the service at a fixed edge node, a mobile edge node, a remote cloud, and the like.

3) Infrastructure providers, such as fixed edge nodes and remote clouds, primarily provide computing, storage, communication and other network resources.

4) The mobile edge node may help the fixed edge node to handle the corresponding service request, i.e. some tasks. In processing tasks from fixed edge nodes, the mobile edge nodes need to be responsible for some of their own traffic, including calculating task completion times (see equation (1) below), and placing bids and planning corresponding movement paths. The system does not know which mobile device will move to which edge area, instead the mobile edge node can plan its own trajectory (e.g., stay in place or move to a particular location) based on the received tasks and predicted task completion times. After the mobile device moves according to the planned path, the mobile device can predict the position of the mobile device when the task is completed.

5) The remote cloud caches all services required by the end user and is considered to have relatively unlimited network resources. In an edge computing environment, end users pay service providers and infrastructure providers a fee to respond to a service.

In order to stimulate the mobile edge node to handle some user requests, the operator needs to pay for it. In the exemplary embodiment of the present invention it must be emphasized that each fixed edge node auctions locally generated user request tasks within each time slice, the task of the auction being determined by the operator. On this basis, details of CRI design are presented in an exemplary embodiment of the invention, as shown in fig. 3:

(one) user request generation and edge task determination

The end user continuously generates user requests and sends the user requests to the operator through the AP. Then, in each time slice, the operator will decide how to schedule these requests according to the conditions of the edge computing environment. Scheduling details that have been extensively studied before are omitted from the exemplary embodiments of the present invention.

For user requests for which fixed edge nodes cannot respond within each time slice, the operator will inform these fixed edge nodes to auction them as tasks. In the exemplary embodiment of the invention with U^tRepresenting the task at time slice t, which is also the corresponding set of user requests. Then, for each task i^t∈U^tBy parameter tuples in an exemplary embodiment of the invention

To indicate.

Is task i^tPre-cached services are required.

Is the data size of the input (output) of the task.

Is task i^tThe amount of computing resources (e.g., CPU cycles) required.

Where is the maximum number of time slices required to process the task (i.e.,

representing task i^tT is the length of each time slice).

Is to cover task i^tThe fixed edge node (containing the corresponding end user).

Note that the tuples in the exemplary embodiments of the present invention may be further extended by introducing more parameters, such as considering other types of resources (e.g., GPU, bandwidth). Here, the required communication resources are not considered in the exemplary embodiments of the present invention, because 5G will provide sufficient bandwidth.

(II) identifying candidate tasks and moving edge nodes

For an auction, both the fixed edge node and the mobile edge node need to ensure that the auction is profitable. In addition, the delay requirement of the user request is also critical. Therefore, a candidate set of mobile edge nodes and tasks that meet the above requirements must first be selected. The basic idea is seen in algorithm 1.

In an exemplary embodiment of the invention, a set of candidate tasks is first selected

It is U^tRepresents all candidate tasks that need to be auctioned at time slice t. Then, in the exemplary embodiment of the present invention, i will be^tIs constructed as

It is M^tRepresents a subset of candidate user requests i^tAll candidate mobile edge nodes of (2). Condition of line 5

Wherein

Is covering m^tEnsures that the auction process is performed locally. For the condition

Wherein

Is m^tIllustrating the mobile edge node m therein^tOnly tasks that invoke the corresponding pre-caching service are considered. On this basis, the mobile edge node will calculate the task completion time and give a corresponding quote. Unlike the D2D model, where tasks of the D2D model can be scheduled directly between devices, the mechanism in the exemplary embodiment of the invention requires that tasks be routed to the nearest fixed edge node first before auctioning. Thus, the completion time is determined by the case of processing a task in the mobile edge node in the exemplary embodiment of the present invention, as follows:

wherein i^tTasks representing respective end users, m^tRepresenting the mobile edge node for handling the task.

Is tuple information from the task.

Is m^tAvailable processing power (in cycles/frame). Here, the available processing capacity refers to the processing capacity of the mobile device for processing the corresponding task, and can be predicted according to the resource usage of the mobile device in a period of time.

In the exemplary embodiment of the invention, the resources of the mobile equipment joining the internet of things are predicted, and whether the mobile equipment can become the mobile edge node is judged according to the prediction result; when the resources of the mobile device are capable of running the edge task, the mobile device becomes the mobile edge node, otherwise it is discarded. That is, when a mobile device without computing resources joins the internet of things, it generally does not serve as a mobile edge node to carry a task, whereas when a mobile device with idle computing resources joins the internet of things, it may serve as a mobile edge node to carry an edge task and obtain possible benefits.

Although a mobile device changes over time due to its own on-device demand, e.g., the mobile device must satisfy its own requests of low latency applications before being able to service other requests, the exemplary embodiments of the present invention assume that the mobile device is still able to provide the same guaranteed resources as it predicted. Since the mobile device can at least ensure that the guaranteed resources are only used to process tasks within the allocated task processing time once the task is received.

In an exemplary embodiment of the present invention,

representing slave end users i^tTo the edge node

The 'user-edge' transmission rate.

Representing slave edge nodes

To the mobile edge node m^tThe 'edge-moving edge' transmission rate. In addition to this, the transmission rate can also be asymmetrical, i.e. the transmission rate can be asymmetrical

The task completion time is composed of a task uploading time, a task processing time and a result downloading time. Therefore, to the right of equation (1), the second and third terms are used in the exemplary embodiment of the present invention

Representing task upload time, first item

For task processing time, fourth and fifth items

Indicating the download time of the task processing result. Therefore, the present invention can be used in exemplary embodiments

The total time to complete the task in the mobile edge node is described.

In addition, the completion time of processing tasks in the remote cloud is expressed as follows in the exemplary embodiment of the present invention:

wherein, r represents a remote cloud,

is the 'edge-cloud' transmission rate. Notably, the processing power c of the cloud data center_rIs very large. In line 8 of Algorithm 1, conditions are used in exemplary embodiments of the present invention

To emphasize only moving edgesThe auction is advisable when the time to process the task in the edge node is shorter than the time to process in the remote cloud. This makes this important considering that edge nodes can be connected to remote clouds over high-speed wired networks, and that remote clouds have more computing power to reduce computing time, which may not always be the case.

In addition, even if the deadline of the task completion time is exceeded (in use), it is found in the exemplary embodiment of the present invention

Show), auctions are still a better choice. If so, a penalty function is used in exemplary embodiments of the present invention because it takes more time to exceed the deadline

Representing additional costs. In the input of Algorithm 1, B^tIs a bid set, V^tIs a set of values. In particular, use in exemplary embodiments of the invention

To represent a moving edge node m^tTo task i^tThe asking price of (2). Lines 10-18 of Algorithm 1 show if task i^tValue to fixed edge node (by)

Represents) greater than treatment i^tThe maximum possible cost is generated, wherein the cost is

When it is satisfied with

I satisfying the condition^t(m^t) Will be added to the candidate set

In (1). Based on all of the above conditions, Algorithm 1 will eventually generate a set of candidate tasks

And candidate moving edge node set

(III) task Allocation

In an implementation manner of the exemplary embodiment of the present invention, performing task allocation according to the global allocation policy, processing, by a mobile edge node that wins out of the auction mechanism, an allocated edge task, and returning a processing result of the edge task to an end user who issued the edge task includes:

considering a set of candidate tasks

And moving edge node sets

The pricing strategy found in the exemplary embodiments of the present invention is not simple, and is particularly shown in the following aspects:

1) at each time period, it is not possible to simply use a bivalent auction strategy (a well-known single item auction pricing strategy) because multiple tasks participate in a bid auction.

2) In addition to the bids of each mobile edge node, the edge nodes also need to consider profit maximization, which also depends on the task completion time of each mobile edge node. Here, in the exemplary embodiment of the present invention, task i will be described in the following manner^tAuction to m^tProfit of fixed edge node:

wherein

Is task i^tThe price of (a) is,

is i^tThe value for a fixed edge node. Suppose that

Is known only to the auctioning party.

M is also calculated in the exemplary embodiment of the present invention by the following method^tBidding task i^tProfit of (2):

wherein

Is the mobile edge node processes task i^tThe cost of (a). It typically reflects the lowest payment that the bidder is willing to bid, i.e.

From the above observations, it is not easy for a fixed edge node to achieve profit maximization with a simple pricing strategy.

3) It is assumed in the exemplary embodiment of the present invention that each mobile edge node can provide multiple services at the same time, but cannot handle multiple tasks of the same service at the same time. Although in a practical edge computing system, multiple tasks may be assigned to one service deployed on a mobile edge node with higher computing power. In particular, in edge computing systems, when a high computing power node is far more efficient than other available low computing power mobile edge nodes, the operator typically assigns multiple tasks to the node.

In one implementation manner of the exemplary embodiment of the present invention, obtaining bids of each mobile edge node, a time for the mobile edge node to complete the edge task, and an auction price of the fixed edge node, and ensuring profit maximization of the fixed edge node through bipartite graph matching includes:

it can still be assumed in the exemplary embodiments of the present invention that each mobile edge node cannot handle multiple tasks of the same service at the same time, because the CRI design in the exemplary embodiments of the present invention can ensure this, making it easier to deploy tasks on the mobile edge nodes. The above difficulties make it difficult to distribute tasks to mobile edge nodes and guarantee maximum profit. In fact, to more clearly state the problem, the exemplary embodiment of the present invention translates this problem into a maximum matching problem on a weighted bipartite graph (see theorem 2).

It is a brief explanation of how the auction problem translates into a bipartite graph matching problem. The vertical dashed lines of figure 4(b) are used to distinguish different auction nodes. The ellipse of fig. 4(b) contains Y nodes generated by the same moving edge node.

Specifically, as shown in FIG. 4(a), an edge computing environment is considered in an exemplary embodiment of the invention, which includes three services, two fixed edge nodes, six tasks, and seven mobile edge nodes. Each task requires a service, and it is assumed in the exemplary embodiment of the present invention that the mobile edge node with the requested service is a candidate mobile edge node. Since each mobile edge node cannot simultaneously handle multiple tasks of the same service, a bidder is defined in an exemplary embodiment of the invention as a particular mobile edge node having a particular service. For example, the mobile edge node 1 in fig. 4(a) with two embedded services may be considered as two bidders in fig. 4 (b). Thus, in fig. 4(b), a weighted bipartite graph (X, Y, E) is constructed in an exemplary embodiment of the invention, where a set of nodes X represents a set of tasks, a set of nodes Y represents a set of bidders, and a set of weighted links E represents a relationship between X and Y.

Examples of the inventionUse in the illustrative embodiment

To represent the weight of each link. In this example, the goal in the exemplary embodiment of the present invention is to maximize the operator's profit, that is, the total profit of all fixed edge nodes is maximized. FIG. 4(b) clearly shows that each fixed edge node conducting the auction is independent because there is no link between their X and Y nodes. Accordingly, the operator's profit maximization in the exemplary embodiment of the present invention may be achieved by maximizing the profit of each fixed edge node.

The structure of the drawing: the weighted bipartite graph (X, Y, E) is then extended in exemplary embodiments of the invention to a more general scenario, which corresponds to the motion-assisted edge computation framework in exemplary embodiments of the invention. Therefore, in each time slice, X is used to represent in the exemplary embodiments of the present invention

In addition, in the exemplary embodiments of the present invention

To represent the elements of Y. In addition, the link of E is used to indicate whether the corresponding mobile edge node from Y is a candidate mobile edge node for the task from X. From the observation of fig. 4(a) and 4(b), the original graph (X, Y, E) can be split into multiple sub-graphs (X, Y, E) in the exemplary embodiment of the present inventionⁿ,Yⁿ,Eⁿ) N ∈ N, and the following conclusion is reached: for subgraph (X)ⁿ,Yⁿ,Eⁿ) And N belongs to N, which are independent of each other and jointly form an original graph (X, Y and E).

To verify that the profit maximization problem of the operator is equivalent to the maximum matching problem of the weighted bipartite graph (X, Y, E), the relevant knowledge in definition 1 and definition 2 is given in the exemplary embodiment of the present invention, and

theorems

1 and 2 are proposed as follows:

definition 1: match in the graph is a set of independent links where no links shareThe same node. The value of the match is the sum of the weights of all links in the match. For bipartite graphs (X, Y, E) and their matching E_mIf | X | ═ E_m| or | Y | ═ E_mI, then E_mIs a perfect match.

Definition 2: for a bipartite graph (X, Y, E), the maximum match is the match that maximizes the value in the graph.

Theorem 1: the maximum profit for the operator is equal to the value of the maximum matching problem for the corresponding bipartite graph (X, Y, E). If the corresponding link is a match from the maximum matching problem, the corresponding task is assigned to the corresponding mobile edge node.

And (3) proving that: the maximum match problem can be said to be finding the most valued match, where the value of each link in the match corresponds to the profit of the operator. Thus, the maximum profit for the operator is equal to the value of the maximum matching problem for the corresponding bipartite graph (X, Y, E). When the maximum match is found, all its links represent the allocation of the corresponding task

Theorem 2: the maximum matching problem of (X, Y, E) can be solved by solving all subgraphs (X)ⁿ,Yⁿ,Eⁿ) And N is the maximum matching problem of N. The value of the previous maximum match is equal to the sum of the values of the maximum matches of the following subgraphs.

And (3) proving that: since all subgraphs of (X, Y, E) are independent of each other, the subgraph generated by each auction node can be computed separately in exemplary embodiments of the invention. The sum of the maximum matching values of the subgraphs is equal to the maximum matching value of (X, Y, E).

Based on

theorems

1 and 2, the operator profit maximization problem can be solved in the exemplary embodiment of the present invention by solving a series of graph matching problems. Therefore, the exemplary embodiments of the present invention will be described in detailThe points are placed on the maximum matching problem of a single node and the solution is given in algorithm 2. For

Hypothetical pricing in exemplary embodiments of the invention

Equating bids in task allocation process

As shown in Algorithm 2, this process is divided into two steps in the exemplary embodiment of the present invention: graph construction and maximum matching problem solving. In a first step (cf. lines 1-6), the exemplary embodiment of the invention is according to what is known

S^t，

To construct a weighted bipartite graph G. In a second step (see lines 7-14), the well-known Kuhn-Munkres (KM) algorithm is used in the exemplary embodiment of the present invention to determine the complexity of O (max { | X³,|Y|³}).

Is EⁿA subset of (a) that represents a perfect match is also the largest match found in the exemplary embodiment of the present invention. Further, P in row 15^tRepresenting the maximum profit for the operator in time slice t. Finally, the invention has been obtained in exemplary embodiments

This is the set of tasks (mobile edge nodes) for the winner (winning the auction).

In order to use the KM algorithm, some related concepts need to be known in the exemplary embodiments of the present invention. For a weighted bipartite graph G ═ (X, Y, E), where link (X, Y) ∈ E has a weight_xyIf, ifThe presence of a real-valued function f on XU Y causes the weight to be greater than or equal to f (X) + f (Y) ≧ weight in the exemplary embodiment of the invention_xy,

Y ∈ Y, then G is called the feasible node label of G. Given a bipartite graph G, a feasible node label can always be found in exemplary embodiments of the invention. One simple feasible node flag is to set f (x) max_y∈Yweight_xy,

f(y)＝0,

Given a feasible label f in graph G, the link set { (x, y) | f (x) + f (y) ═ weight in the exemplary embodiment of the invention_xyX, y ∈ E } is expressed as E_fWhere (X, Y) is referred to as a feasible link. Drawing G_f＝(X,Y,E_f) Called the equal subgraph of G. The KM algorithm attempts to find a perfect match on the graph by carefully adjusting the node labels on the graph, gradually considering more links.

In the process of auction mechanism, each mobile edge node sends participation certification to the fixed edge node;

6. The method of claim 1, further comprising:

Thus, in algorithm 2, in the second step in the exemplary embodiment of the present invention, the time complexity of O (N × max { | X) can be obtainedⁿ|³,|Yⁿ|³}) which is larger than O (max { | X³,|Y|³}) are much better (i.e., the time to solve the G-map). The maximum profit value for row 15 is then calculated, resulting in the winner task and mobile edge nodes in row 16.

(IV) task processing and Payment

After the task allocation is completed, the winner (i.e. the mobile edge node) will process the corresponding user request and return the result to the end user. At this stage it is worth noting whether the mobile edge node will perform processing operations and how to ensure this.

In one implementation of the exemplary embodiment of the present invention, during an auction mechanism, a participation certificate is sent by each of the mobile edge nodes to the fixed edge node;

This process includes the following:

as shown in fig. 3, in addition to generating the candidate set, the mobile edge node needs to submit a participation certificate, PoP (proof of participation), to the fixed edge node in section two of chapter three. The PoP may be a device identification number that may be used to identify and track unique mobile edge nodes. Then if the mobile edge node that submitted the PoP and became the winner abandons the mission, it would compromise its own reputation and may even be subject to additional indemnity and legal liability. In addition, if the mobile edge node does not complete the task within the guaranteed time, the edge node will be compensated. Thus, PoP may be used in exemplary embodiments of the present invention to ensure the honesty of the behavior of the mobile edge node.

When all winner tasks are processed

The fixed edge node should win the winner

Paying for its workload. To ensure that the fixed edge node remunerates the mobile edge node that completed the task, the exemplary embodiment of the present invention uses a PoA (proofs of distribution) that is sent to the winner before the task is processed. The PoA may record the fact that the fixed edge node assigned the task to the mobile edge node, on the basis of which the fixed edge node has to pay the mobile edge node for the assigned task after completing the task processing.

Algorithm complexity

In conjunction with the details of the CRI design, it is proposed to implement the main functionality of the CRI design in each time slice (see algorithm 3). The algorithm includes the entire auction process, i.e., candidate set generation (refer to algorithm 1) and task allocation determination phase (refer to algorithm 2). At line 9 of Algorithm 1, the

The mobile edge nodes in (1) are sequenced, and the complexity is taken as

The time of (c). For Loop in line 10, time complexity of

Thus, the temporal complexity of the for loop in row 6 is

Thus, algorithm 1 takes time

In one implementation of the exemplary embodiments of this invention, the method further comprises:

when the edge task distributed according to the global distribution strategy is processed and completed, the fixed node pays the workload to the corresponding mobile edge node according to the distribution certification sent by the fixed edge node to the corresponding mobile edge node, and the process comprises the following steps:

in algorithm 2, the temporal complexity of the KM algorithm is O (N × max { | X)ⁿ|³,|Yⁿ|³}) (please see section three of section three). In addition, the ordering X node in line 2 takes time

Ordering Y node in line 3 requires time

Thus, the time complexity of algorithm 2 is

Thus, the overall temporal complexity of the CRI in algorithm 3 is

In other words, the CRI can converge to the final allocation and pricing result within the polynomial time.

(II) desired Properties

To analyze the proposed CRI auction mechanism, we demonstrated its ideal properties using the following theorem.

Theorem 3: CRI satisfies individuality.

And (3) proving that: CRI satisfies individual rational number and only number

Row 11-22 of Algorithm 1 and in equation (3)

In combination, we find that in algorithm 2 for the operator (or fixed edge node), we can do so

Auction task i^tMust be profitable, i.e.

Thus, the auctioneer's profit is non-negative.

For the auction participants (i.e., the mobile edge nodes participating in the auction), the exemplary embodiment of the present invention uses the equation (4)

To represent the profit of each auctioneer. Let us now turn to an exemplary embodiment of the invention

Then

Always hold, so the bidder's profit is non-negative. From the perspective of the bidder/mobile edge node, the non-negative profit actually promotes efficient utilization of idle resources. Thus, potential mobile edge nodes can be fully utilized.

The above two conclusions together demonstrate that CRI satisfies individuality.

Theorem 4: the CRI satisfies authenticity.

And (3) proving that:the particularity of the CRI design is that the exemplary embodiment of the present invention seeks to maximize the profit of the auctioneer only, not the auctioneer. In this design, bidders will decide their own bid B^tSo that they can obtain some profit. However, due to the sealed auction, the bidder is unaware of the counterparty's bids, computing power, and other information. If this were to upset the price, it would likely result in the bidder being excluded from the candidate set and no revenue being obtained. Thus, it may be inferred in exemplary embodiments of the present invention that their bids are substantially reasonable. In B^tOn the basis of the above, in order to maximize the profit of the whole domain, each auctioneer (fixed edge node) decides winning bidders by using the KM algorithm to maximize their profit. Since the pricing strategy is based on bids of each bidder, the decision to fix the edge node does not compromise the bidder's profit. From the above reasoning, it can be seen that both auctioneers and auction owners must be truthfully credited to maximize their own benefits.

Theorem 5: CRI satisfies profit maximization.

And (3) proving that: from the proof of theorem 3, it can be seen that only when in equation (4)

The fixed edge nodes can only achieve maximum profit. If it is not

The benefits of the auctioneer are compromised and should be avoided. If it is not

Then the auctioneer's confidence, and ultimately the long-term interest, will be reduced in terms of the presence of PoA. Thus, the CRI design in exemplary embodiments of the present invention ensures profit maximization.

The economic characteristics are combined with the computational efficiency, and the feasibility of CRI design is theoretically guaranteed. Then, in the exemplary embodiment of the present invention, from the experimental point of view, the feasibility of the simulation result in the second and third sections of chapter five is ensured.

Fifth, performance evaluation

(one) Experimental setup

Network construction: the evaluation of CRI design does not assume the requirements of additional user requests, the value of the task and the bids of the mobile edge nodes. Thus, the numerical results in the exemplary embodiments of the present invention are valid for any possible data set. Since there is no statistical data about the value of the user request and the cost of the mobile edge node to process the tasks in a real edge computing environment, for the sake of generality, the value of the user request and the corresponding bid for each task are randomly generated in exemplary embodiments of the present invention, all uniformly distributed within (0, 1).

Fig. 5(a) (b) are data graphs illustrating the supply-demand mismatch and the corresponding ratios α and β.

In addition, a parameter α is introduced in exemplary embodiments of the present invention, which is the ratio of candidate tasks to all user requests generated in the edge computing environment. In fact, α reflects the resource shortage of the fixed edge node, wherein the larger the value of α, the more scarce the resource. The candidate task set is generated using alpha randomness in the exemplary embodiment of the invention. Furthermore, the parameter β, i.e. the ratio of candidate moving edge nodes to all covered moving edge nodes, is also introduced in the exemplary embodiments of the present invention. And beta can be used for reflecting the emergency capacity, namely the larger the value is, the stronger the emergency capacity is. Both α and β dynamically change over time, and the goal in exemplary embodiments of the present invention is to generate α and β to approximate the edge computing environment, which means that the supply and demand are substantially balanced. In the experimental apparatus in the exemplary embodiment of the present invention, the supply and demand mismatch condition in fig. 5(a) was simulated in the exemplary embodiment of the present invention. As described above, in the exemplary embodiment of the present invention, the data set of the taxi-hire user is used to simulate the requirement of the user request, and a capability index (e.g., the processing capability of the fixed edge node) is set.

From the observation of fig. 5(a), it is found in the exemplary embodiment of the present invention that the supply-demand mismatch is very serious because the resources of the fixed edge node are over-allocated during certain periods of the day (e.g., 0-8 o' clock) and the resources are in short supply during certain periods (e.g., 18:00 to 21: 00). Then, the ratios α and β are expressed as follows in the exemplary embodiment of the present invention:

α＝max{(Demand-Supply)/Supply,0} (5)

β＝max{(Demand-Supply)/Supply+σ,0} (6)

fig. 5(b) shows that the value of β varies with the value of α, and the value of σ is set according to a uniform distribution within [0,0.1] in the exemplary embodiment of the present invention.

And (3) comparison algorithm: to demonstrate the performance of the CRI algorithm, it is compared in an exemplary embodiment of the invention with two reference algorithms:

(i) greedy Assignment (GA): the algorithm uses a greedy assignment algorithm (refer to lines 4-10 of algorithm 4) instead of the KM algorithm of algorithm 3. In line 9

Is EⁿRepresents a matching term of the greedy algorithm search. It can be seen that the temporal complexity of the greedy allocation algorithm in exemplary embodiments of the invention is O (N X | X)ⁿ×Yⁿ|²log₂|Xⁿ×Yⁿ|)。

(ii) Edge cloud mode (AC): the intent is to respond to all users' requests with the capability of a fixed edge node and a remote cloud, which has been studied in a number of jobs. In this case, it is found in the exemplary embodiment of the present invention that although the operator does not pay the mobile edge node cost, the operator still loses economic profit due to low task completion rate.

Performance of the (II) algorithm

To demonstrate the performance of the algorithm, it was compared with two references (i.e., GA and AC in the first section of chapter five) in the exemplary embodiment of the present invention. In the exemplary embodiment of the present invention, three indicators are strictly analyzed and generously evaluated, as shown in fig. 6(a) and fig. 6(b), wherein fig. 6(a) shows the change of CRI/GA/AC corresponding to the completion rate of the task, and fig. 6(b) shows the change of CRI/GA/AC corresponding to the profit of the operator.

Task Completion Ratio (TCR): figure 6(a) shows the mean task completion rate for the benchmark algorithm and CRI for different numbers of candidate mobile edge nodes and tasks. For example, 350 × 500 in fig. 6(a) means that 350 candidate moving edge nodes and 500 candidate tasks are randomly generated in the exemplary embodiment of the present invention. The task completion rate of the Y-axis represents the completion rate of the candidate task. The results show that the TCR of CRI and GA are much higher than that of AC, suggesting that introducing an auction mechanism in a marginal computing environment can significantly improve TCR. Meanwhile, the TCR found to be CRI in exemplary embodiments of the present invention is slightly higher than the TCR of GA, and the KM algorithm found to be CRI in exemplary embodiments of the present invention is looking for a complete match of a drama-equivalent subgraph, whereas the greedy assignment algorithm of GA may not search for such a match.

In addition, in the exemplary embodiment of the present invention, it can be seen that the TCR increases with the increase of the candidate moving edge nodes and decreases with the increase of the candidate tasks, which can be interpreted as that the more candidate moving edge nodes (the fewer candidate tasks to be responded to), the more tasks that can be responded to (the higher TCR). Therefore, in the exemplary embodiments of the present invention, it is concluded that the CRI design can significantly improve the quality of service of the fixed edge node.

Profit: fig. 6(b) shows the average profit that a fixed edge node can obtain by varying the number of candidate mobile edge nodes and tasks. The results show that the yield of the CRI algorithm is higher than that of the GA and the AC because the KM algorithm of the CRI pursues the maximum matching, and the greedy algorithm of the GA may not obtain the optimal solution. While the profit above the AC is because although more profit is available for each task being responded to, the overall profit is reduced by the reduction in TCR. Thus, the profit of the CRI design in the exemplary embodiment of the present invention is still better than the other two benchmarks. Furthermore, all profits in fig. 6(b) increase with the increase of candidate mobile edge nodes due to the increase of the synchronized TCR. However, as the number of candidate tasks increases, the profit increases and then decreases. The reason can be understood as that TCR is decreasing but the cardinality of candidate tasks is increasing. Therefore, in exemplary embodiments of the present invention, it is concluded that CRI design is economically advantageous for fixed edge nodes.

Calculating efficiency: to confirm the analysis of time complexity in the exemplary embodiment of the present invention, CRI run times at different settings are recorded in table 2 in the exemplary embodiment of the present invention. For each setting, 50 instances were randomly generated and the average result was recorded in the exemplary embodiment of the present invention. All tests were run in the exemplary embodiment of the invention using python3.6 and Intel (R) core (TM) i7-8750H processors, 16gb memory processors. Algorithm 3, which verifies the CRI in the exemplary embodiment of the present invention, converges within polynomial time for the candidate task set and the moving edge node set, respectively. Table 3 shows the average run times of the GAs, and the results show that both CRI and GA are computationally efficient, and that GA is no better than CRI. With respect to the AC, while it need not take the time of the auction process, exemplary embodiments of the present invention subsequently show that it may take other metrics.

Overall, the CRI algorithm in the exemplary embodiment of the present invention has better performance on all three criteria as compared to the other two benchmarks. These metrics reflect the advantages of CRI design in terms of quality of service, economic efficiency and computational efficiency. Thus, in exemplary embodiments of the present invention, it may be concluded that: applying CRI to the mobile assisted edge computing framework in exemplary embodiments of the present invention can improve QoS of fixed edge nodes by achieving higher TCR and higher profit margins.

Individuality: theorem 3 proves that CRI is individual rational, meaning that the value of each winning bid task is higher than the charged price (see fig. 7(a)), while each winning mobile edge node receives a payment that is no lower than its cost of processing the corresponding task (see fig. 7 (b)).

The value and price of the winning bid task in fig. 7(a) and the bid and cost of winning the mobile edge node in fig. 7(b) are given in the exemplary embodiment of the present invention. The results show that the winning mobile edge node gets a positive benefit, i.e. benefits from CRI. In addition, the operator may still make a profit because the utility of each winning task is positive. Thus, mobile edge nodes are willing to support fixed edge nodes, and operators also tend to auction tasks that some fixed edge nodes cannot handle.

Authenticity: to verify the authenticity of CRI in theorem 4, a mobile edge node is randomly chosen in the exemplary embodiment of the present invention to study its change in benefit when bids are not on the same time. As shown in fig. 8(a), a shows that the mobile edge node is a winner when its bid is 0.448 and cost is 0.295. Thus, the mobile edge node obtains a utility/profit of 0.153. The bid retest of this mobile edge node is then changed in the exemplary embodiment of the present invention. The results show that when the bid is below 0.448, the mobile edge node can still win the auction, but when the bid exceeds $ 0.448, the node will lose the auction. It can be seen that the mobile edge node cannot improve its utility no matter how much it bids. However, from tests on all moving edge nodes, it is found in the exemplary embodiments of the present invention that there are very few cases like b: when the bid is $ 0.557 and the cost is $ 0.306, the mobile edge node can still increase its utility by raising the bid. Here, it is emphasized in the exemplary embodiment of the present invention that although all mobile edge nodes can flexibly adjust their bids, there is also a great risk of bid failure, and the failure rate of the experimental test is as high as 97.87%. Thus, a rational mobile edge node will really bid.

Furthermore, the authenticity of edge nodes may be inferred in two cases in exemplary embodiments of the present invention: 1) if the price is lower than the bid of the successful bidder, the reputation of the fixed edge node is damaged, which is catastrophic; 2) if the price is higher than the bid of the winning bidder, the profit of the fixed edge node is compromised. In summary, CRI can guarantee the authenticity of mobile edge nodes and fixed edge nodes, since unreal behavior cannot improve utility. Thus, CRI can be liberated from interference by unrealistic participants (fixed edge nodes and mobile edge nodes) who attempt to benefit themselves at the expense of the benefit of others.

And (3) maximizing profit: to verify that the profit of theorem 5 is maximized, case c in fig. 8(b) is used in the exemplary embodiment of the present invention, in which the winning bidder bids 0.44 and the value of the winner mission is 0.55. If price < 0.44, the reputation of the fixed edge node is compromised (see theorem 4), which is avoided to the utmost. If price ≧ 0.44, the fixed edge node cannot obtain further benefit from the task (see equation (3)). Thus, CRI ensures that the profit that the fixed edge node gets from each winner task is maximized.

Through the simulation experiments, the consistency of the experimental results and theoretical proofs is verified, and the simulation of a real scene is considered to be necessary in future work.

In exemplary embodiments of the present invention, it will be discussed whether CRI design in exemplary embodiments of the present invention can bring about other performance improvements, such as potential long-term profit growth for the operator. From the result of the simulation experiment, the design in the exemplary embodiment of the invention is beneficial to improving the task completion rate, so that the user experience of emerging internet of things application is further improved, and more users are attracted to send requests to the edge computing environment. This is a potential long-term benefit of applying CRI in practical edge-computing scenarios. Preliminary attempts have been made in exemplary embodiments of the present invention to move the assisted border calculation framework.

Efficient solutions for providing ultra-low latency services for emerging internet of things applications in edge computing environments are being actively researched and applied. For example, service placement and request scheduling are combined, so that the utilization rate of edge node resources is improved; GSP-ORS allows an edge computing system operator to globally allocate resources of fixed edge nodes, optimizes the request of a scheduling user and realizes efficient utilization of the resources; computation offload in relation to edge computation; describing a distributed computing offload decision problem among mobile device users as a multi-user computing offload gaming problem; the computing power of a mobile device is increased by borrowing processing resources from the cloud. However, none of the above approaches consider the transfer of tasks from a fixed edge node to a mobile device. CRI therefore complements these efforts, and currently, the mismatch problem at the network edge remains of little concern.

The exemplary embodiments of the present invention are concerned with a unique emergency (supply-demand mismatch), auction item (user request) and buyer (mobile edge node). The work in the exemplary embodiments of the present invention also differs from a general auction mechanism that employs a double auction and an off-center auction to achieve an optimized configuration of resources. However, CRI focuses on reasonable, real quotations based on mobile edge nodes, without considering the optimal quotation strategy of mobile edge nodes. CRI can also be considered as a gaming mechanism design with credit vouching certificates (see PoP, PoA in fig. 2).

While the goal of edge computing is to provide end users with "anytime and anywhere" low latency services, the limited resources of fixed edge nodes and the bursty peak demand present challenges to existing solutions. These challenges are described in exemplary embodiments of the present invention as a supply and demand mismatch problem at the edge of the network. The exemplary embodiment of the present invention first proposes a mobile assisted edge computing framework, which improves the quality of service of a fixed edge node by introducing a mobile edge node. Meanwhile, CRI is designed in the exemplary embodiment of the present invention to stimulate the mobile edge node to reduce the supply and demand mismatch. Experiments have shown that CRI has good individual reasonableness, reality and profit maximization, and numerical results show the superiority of the mobile assisted edge computing framework, which has prompted more research in exemplary embodiments of the present invention. The possible future development directions in the exemplary embodiments of the present invention can be briefly summarized as: 1) applying the CRI design to an actual scene; 2) evaluating the long-term yield of the CRI; 3) and researching and improving the QoE of the user.

Fig. 9 is a schematic structural diagram of a mobile assisted edge computing apparatus according to an embodiment of the present invention, where the apparatus may be implemented by software and/or hardware, generally integrated in an intelligent terminal, and may be implemented by a mobile assisted edge computing method. As shown in the figure, the embodiment may provide a mobile assisted edge computing device based on the above embodiments, which is applied to the internet of things including at least a fixed edge node and a mobile edge node, and mainly includes an edge task determining module 910, an auction module 920, and an allocating module 930.

The edge task determining module 910 is configured to request an edge task from a fixed edge node according to a user request and resource distribution, where the edge task is a user request that cannot be responded by the fixed edge node;

an auction module 920, configured to auction the edge tasks to a plurality of mobile edge nodes according to an auction mechanism, and determine a final pricing policy according to the profit of each mobile edge node and the expenditure of a fixed edge node;

an allocation module 930 configured to determine a global allocation policy of the edge task at each of the mobile edge nodes according to the pricing policy and pricing of the mobile edge nodes.

In one implementation of the exemplary embodiments of this invention, the apparatus further comprises:

In an implementation manner of the exemplary embodiment of the present invention, before auctioning the edge task to a number of mobile edge nodes according to an auction mechanism, the apparatus further includes a determining module configured to:

In an implementation manner of the exemplary embodiment of the present invention, the apparatus further includes a task allocation sub-module, configured to:

In one implementation of the exemplary embodiments of this invention, the apparatus further includes an evaluation module configured to:

when the mobile edge node wins the auction abandons the edge task, the reputation parameter of the corresponding mobile edge node is reduced according to the participation proof;

The apparatus provided in the foregoing embodiments may be configured to perform the method for calculating a moving auxiliary edge provided in any embodiment of the present invention, and have corresponding functional modules and advantageous effects for performing the method.

It should be noted that the method of the exemplary embodiment of the present invention may be executed by a single device, such as a computer or a server. The method of the embodiment can also be applied to a distributed scene and completed by the mutual cooperation of a plurality of devices. In such a distributed scenario, one of the devices may perform only one or more steps of the method of the exemplary embodiments of the present invention, and the devices may interact with each other to complete the method.

The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.

Fig. 10 is a schematic diagram illustrating a more specific hardware structure of an electronic device according to this embodiment, where the electronic device may include: a processor 1010, a memory 1020, an input/output interface 1030, a communication interface 1040, and a bus 1050. Wherein the processor 1010, memory 1020, input/output interface 1030, and communication interface 1040 are communicatively coupled to each other within the device via bus 1050.

The processor 1010 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits, and is configured to execute related programs to implement the technical solutions provided in the embodiments of the present disclosure.

The Memory 1020 may be implemented in the form of a ROM (Read Only Memory), a RAM (Random Access Memory), a static storage device, a dynamic storage device, or the like. The memory 1020 may store an operating system and other application programs, and when the technical solution provided by the embodiment of the present disclosure is implemented by software or firmware, the relevant program codes are stored in the memory 1020 and called by the processor 1010 to execute the mobile assisted edge computing method according to the embodiment of the present disclosure.

The input/output interface 1030 is used for connecting an input/output module to input and output information. The i/o module may be configured as a component in a device (not shown) or may be external to the device to provide a corresponding function. The input devices may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output devices may include a display, a speaker, a vibrator, an indicator light, etc.

The communication interface 1040 is used for connecting a communication module (not shown in the drawings) to implement communication interaction between the present apparatus and other apparatuses. The communication module can realize communication in a wired mode (such as USB, network cable and the like) and also can realize communication in a wireless mode (such as mobile network, WIFI, Bluetooth and the like).

Bus 1050 includes a path that transfers information between various components of the device, such as processor 1010, memory 1020, input/output interface 1030, and communication interface 1040.

It should be noted that although the above-mentioned device only shows the processor 1010, the memory 1020, the input/output interface 1030, the communication interface 1040 and the bus 1050, in a specific implementation, the device may also include other components necessary for normal operation. In addition, those skilled in the art will appreciate that the above-described apparatus may also include only those components necessary to implement the embodiments of the present description, and not necessarily all of the components shown in the figures.

Computer-readable media of the present embodiments, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, programs, modules of the programs themselves, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device to perform the above-described aspects of embodiments of the present invention.

Those of ordinary skill in the art will understand that: the discussion of any embodiment above is meant to be exemplary only, and is not intended to intimate that the scope of the disclosure, including the claims, is limited to these examples; within the idea of the present disclosure, also technical features in the above embodiments or in different embodiments may be combined, steps may be implemented in any order, and there are many other variations of the different aspects of the exemplary embodiments of the present invention as described above, which are not provided in detail for the sake of brevity.

In addition, well-known power/ground connections to Integrated Circuit (IC) chips and other components may or may not be shown within the provided figures for simplicity of illustration and discussion, and so as not to obscure the exemplary embodiments of the invention. Furthermore, devices may be shown in block diagram form in order to avoid obscuring exemplary embodiments of the present invention, and this also takes into account the fact that specifics with respect to implementation of such block diagram devices are highly dependent upon the platform within which the exemplary embodiments of the present invention are to be implemented (i.e., specifics should be well within purview of one skilled in the art). Where specific details (e.g., circuits) are set forth in order to describe example embodiments of the disclosure, it should be apparent to one skilled in the art that the example embodiments of the disclosure can be practiced without, or with variation of, these specific details. Accordingly, the description is to be regarded as illustrative instead of restrictive.

While the present disclosure has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations thereof will be apparent to those skilled in the art in light of the foregoing description. For example, other memory architectures (e.g., dynamic ram (dram)) may use the discussed embodiments.

The exemplary embodiments of the invention are intended to embrace all such alternatives, modifications and variances that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, substitutions, improvements, and the like that may be made without departing from the spirit and principles of the exemplary embodiments of the invention are intended to be included within the scope of the disclosure.

Claims

1. A mobile assisted edge computing method applied to the Internet of things at least comprising fixed edge nodes and mobile edge nodes is characterized by comprising the following steps:

auctioning the edge tasks to a plurality of mobile edge nodes according to an auction mechanism, and determining a final pricing strategy according to the income of each mobile edge node and the expenditure of a fixed edge node;

2. The method of claim 1, further comprising:

3. The method of claim 1, wherein prior to auctioning the edge tasks to a number of mobile edge nodes according to an auction mechanism, the method further comprises:

4. The method of claim 1, further comprising:

5. The method of claim 4, further comprising:

6. The method of claim 1, further comprising:

and when the edge tasks distributed according to the global distribution strategy are processed and finished, the fixed edge nodes pay the workload to the corresponding mobile edge nodes according to the distribution certification sent by the fixed edge nodes to the corresponding mobile edge nodes.

7. A mobile assisted edge computing device for use in an internet of things including at least a fixed edge node and a mobile edge node, the device comprising:

8. The apparatus of claim 7, further comprising:

9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of motion assisted edge calculation as claimed in any one of claims 1 to 6 when executing the program.

10. A non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method of motion-assisted edge computation of any one of claims 1 to 6.