EEH-attached green multiple access edge computing demand response method based on contract system

Through the EEH strategy based on the contract system, using time-varying hypergraphs and social contract theory, the problems of network dynamic changes and energy consumption management in edge computing are solved, and the demand response and deception of green multiple access edge computing are realized. Improves the sustainability and energy efficiency of the system.

CN119003168APending Publication Date: 2024-11-22CHUZHOU UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411071206.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

Existing edge computing technologies face challenges such as dynamic changes in network topology, energy consumption management and fraud prevention when implementing green multi-access edge computing, especially the difficulty in adapting to the impact of high-density base station deployment on sustainability and fraudulent behavior.

Method used

Adopting the EEH (Edge Energy Collection and Management) strategy based on the contract system, formalize the MEC scenario through time-varying hypergraphs, design edge efficiency models and energy collection models, and construct the MEC contract system in combination with social contract theory to realize dynamic node management and Optimize energy supply to prevent fraudulent behavior.

Benefits of technology

In the context of green MEC, the overall energy consumption of the system is minimized, the network changes are dynamically responded to network changes, the energy utilization rate is improved, the edge node fraud behavior is prevented, and the efficient computing task is ensured.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119003168A_ABST
    Figure CN119003168A_ABST
Patent Text Reader

Abstract

The invention discloses a green multiple access edge computing demand response method with EEH based on a contract system, and relates to the technical field of edge computing, and the method comprises the following steps: formalizing an MEC scene based on a time-varying hypergraph; designing an edge efficiency model, namely an EU model, and designing an edge energy collection model; establishing a joint optimization model; under the support of a social contract theory, constructing an MEC contract system based on the time-varying hypergraph; establishing an EEH strategy based on a contract system; according to the method, an MEC scene is formalized by using a time-varying hypergraph, multi-type relation characteristics of edge devices and users of different domains in an ECN along with time are described, in addition, concepts of an incidence matrix and a hypergraph cardinal number are provided to manage edge nodes, and the edge efficiency is improved. A cardinal number is calculated through an operation set in the hypergraph, and an incidence matrix is synchronized in real time, so that edge nodes are dynamically started and stopped to adapt to the change of edge users along with time.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of edge computing technology, and in particular to a green multi-access edge computing demand response method with EEH based on a contract system. Background Art

[0002] With the advent of the 5G era, the proliferation of Internet of Things (IoT) devices and users has grown exponentially. Traditional cloud computing has been struggling to handle the terminal data generated by these IoT devices 24 / 7, as well as the needs of online applications and clients, such as online games, financial services, and healthcare applications. As a remedy to these limitations, multi-access edge computing (MEC) has emerged, which can extend the cloud to the edge of the network to meet the strict timeliness requirements and intensive computing needs of emerging IoT applications. The existing MEC strategy deploys a large number of base stations (BSs) to increase the coverage of the network and provide users with higher throughput, with a maximum of 50 edge servers per square kilometer at present. However, the deployment of such a large and high-density BSs will pose a significant obstacle to achieving sustainable green computing. Therefore, how to minimize the overall energy consumption of the system in the context of green MEC has received widespread attention from scholars. The existing technology still has the following shortcomings:

[0003] 1. The network topology in the edge computing environment may change frequently over time, such as the joining and exit of new and old users, which will affect the performance and stability of the entire network. Existing modeling methods may find it difficult to adapt to such dynamic changes, and more flexible modeling strategies are needed to cope with the uncertainty of network topology.

[0004] 2. Existing studies have proposed energy harvesting technology to help edge devices save energy, but these studies do not change strategies for different nodes, nor can they power edge devices while effectively preventing fraudulent behavior.

[0005] 3. At present, relevant research mainly focuses on external network deception strategies, but how to reduce the deception behavior originating from the edge nodes themselves in MEC requires in-depth research. In addition, although the current research on deception behavior can prevent network deception, it lacks relevant theoretical background support.

[0006] Therefore, a green multi-access edge computing demand response method with EEH based on a contract system is needed to solve the above problems. Summary of the invention

[0007] The purpose of the present invention is to provide a green multi-access edge computing demand response method with EEH based on a contract system to solve the problems existing in the prior art proposed in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] The green multi-access edge computing demand response method with EEH based on the contract system includes the following steps:

[0010] S1: Formalize the MEC scenario based on time-varying hypergraph;

[0011] S2: Design the edge efficiency model, namely EU model, and design the edge energy collection model;

[0012] S3: Establish a joint optimization model;

[0013] S4: With the support of social contract theory, we build a MEC contract system based on time-varying hypergraph;

[0014] S5: Establish EEH strategy based on contract system;

[0015] S6: Use the EEH strategy based on the contract system to formulate demand response logic, minimize the edge efficiency model, and prevent edge node deception, ultimately achieving the goal of green MEC.

[0016] Preferably, the MEC scenario construction steps of the time-varying hypergraph in step S1 are:

[0017] (1) Using triple TH t = <V t ,B t ,F t > to describe the time-varying hypergraph, of which the three subset components are as follows:

[0018] -V t = {U t ,S t ,N t} is a finite set of edge servers, edge nodes, and edge users at t;

[0019] - and m={1,2,...,M} is V t The non-empty subsets at time t correspond to different edge communities in real edge scenarios. In addition, they also meet the following restrictions:

[0020] -F t Represents the changes in real edge scenarios and demand response strategies in the future. t ={wake_node,off_node,add_user,delete_user} and contains V t Operations;

[0021] (2) In order to perform real-time demand response analysis on the edge system, the association matrix of the edge node is set to Describing TH in terms of triples of time-varying hypergraphs t = <V t ,B t ,F t >, incidence matrix The following conditions are met:

[0022] - The line with EH t The vertices of are related, that is, edge nodes;

[0023] - The columns and EH t The hyperedge correlation in ;

[0024] (3) In order to better manage the node relationships in the edge community, the concept of cardinality is introduced, namely where the cardinality in the hyperedge is the number of edge nodes in the edge community m at time t in the time-varying hypergraph, Calculate as formula (1):

[0025]

[0026] In the formula is the incidence matrix at time t The number of columns, Y(*) is the indicator function, when Y(*) is true, its value is 1, otherwise it is 0;

[0027] Will m={1,2,...,M} represents the cardinality set of all edge node rows in the hypergraph.

[0028] Preferably, the steps of establishing the edge effectiveness model in step S2 are:

[0029] The parameters of the EU model include task completion time CT, task failure rate FR, total data rate DR and system energy consumption EC. The formal expressions of each parameter are as follows:

[0030] (1) CT consists of three parts: task transmission time, computation time, and waiting time. The transmission time Γ of task i from the end user device to the edge node is calculated. tran (t) is shown in formula (2):

[0031]

[0032] In the formula is the wireless channel bandwidth from the terminal device to the edge node, θ ik is the channel gain of task i arriving at node k for the first time, θ ki is the channel gain of the last return of task i to node k, ρ is the noise power spectral density;

[0033] The total computation time Γ of task i com (t) is shown in formula (3):

[0034]

[0035] In summary, the total CT of all computing tasks is shown in formula (4):

[0036]

[0037] (2) FR represents the proportion of edge tasks that fail to be successfully scheduled within a specific time frame of MEC or under specific operating conditions and a measure of execution efficiency, which can be calculated by formula (5):

[0038]

[0039] Where K i and Represent the number of resources required and remaining resources of MEC respectively. If the edge task The number of required resources is greater than the number of remaining resources, or the waiting time is greater than the computing time, indicating that the user is on the edge of the task Scheduling failed;

[0040] (3) DR is a measure of the total amount of data transmission between edge nodes and edge servers within a certain period of time. It represents the total data transmission rate and takes into account the upload and download speeds. DR can be calculated by formula (6), where Indicates DR, represents the amount of data transmitted by edge node k in time period t;

[0041]

[0042] (4) In the present invention, the EC of the edge device consists of two parts: the energy consumption when the local CPU is awakened and the energy consumption during operation. The power consumption of the local CPU awakening is mainly consumed by the device CPU. k is the effective switching capacitance of the CPU of edge device k, which is determined by the chip structure of the edge device. To represent the energy consumed by the local CPU execution, as shown in formula (7):

[0043]

[0044] According to the above task transmission time, let η(t) be the transmission power of edge node k in time slot t, then calculate the operating energy consumption As shown in formula (8), and defined is the total EC of the system;

[0045]

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] 1. This paper formalizes the MEC scenario by using a time-varying hypergraph to describe the multi-type relationship characteristics of edge devices and users in different domains in the ECN over time. In addition, the concepts of association matrix and hypergraph cardinality are proposed to manage edge nodes. The cardinality is calculated through the operation set in the hypergraph, and the association matrix is ​​synchronized in real time, so as to dynamically start and stop edge nodes to adapt to the changes of edge users over time.

[0048] 2. This paper proposes a scenario modeling EU model based on time-varying hypergraph, which is an important parameter model for evaluating user QoS and overall system performance. Secondly, the traditional EH model is improved and optimized, and an EEH model is proposed, which can classify energy supply according to the situation, making it more suitable for EDR problems. Finally, an optimization model is proposed to achieve an adjustable trade-off between deception perception and the evaluation parameters mentioned in the EU model.

[0049] 3. The present invention uses social contract theory to solve the EDR problem and proposes a new GEC-CS algorithm based on the contract system. The edge servers will be divided into three categories: contract nodes, abnormal nodes and dormant nodes according to the historical scheduling behavior or user needs of the hyper-edge. In addition, the EEH strategy adopts different energy supply strategies according to different types of edge nodes to ensure maximum energy utilization. The GEC-CS method starts and stops the working status of edge nodes through real-time analysis and judgment, thereby realizing green MEC. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a multi-access edge computing (MEC) model diagram based on a time-varying hypergraph of the present invention.

[0051] Figure 2 This is a structural model diagram of the edge energy harvesting (EEH) technology of the present invention.

[0052] Figure 3 This is a diagram of the EEH mechanism based on the contract system of the present invention.

[0053] Figure 4 This is a comparison chart of the energy supply of EEH components at different nodes of the present invention.

[0054] Figure 5 This is a comparison chart of the algorithm effects under different benchmark methods of the present invention (Set#1).

[0055] Figure 6 This is a comparison chart of the algorithm effects under different benchmark methods of the present invention (Set#2).

[0056] Figure 7This is a comparison chart of the algorithm effects under different benchmark methods of the present invention (Set#3).

[0057] Figure 8 This is a comparison chart of the number of algorithm iterations under different experimental parameters of the present invention.

[0058] Fig. 9 This is a comparison chart of total time consumption under different benchmark algorithms of the present invention (small-scale scenario).

[0059] Fig.10 This is a comparison chart of total time consumption under different benchmark algorithms of the present invention (large-scale scenario). DETAILED DESCRIPTION

[0060] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0061] See also Figure 1-10 , the present invention provides the following technical solutions:

[0062] The present invention provides an ECN based on a time-varying hypergraph, in which, in a green MEC scenario with deceptive behavior, the joint optimization objective is maximized with the assistance of a contract system and EEH while achieving EDR, and edge users are ensured to efficiently complete edge computing tasks while avoiding deceptive behavior.

[0063] The green multi-access edge computing demand response method with EEH based on the contract system includes the following steps:

[0064] First, the edge tasks and edge devices are formally described:

[0065] (1) i={1,2,...,n} represents the computing task generated by the i-th edge user at time t; and Represents the set U t A sub-element of , where the sextuplet is defined as follows:

[0066] -Li represents the edge community that can schedule the i-th edge task, -Wi is the waiting time of the i-th edge task, -Di represents the size of data that needs to be downloaded by the i-th task, -Di' represents the amount of data that needs to be uploaded by the i-th task, that is, the size of the uploaded data, and -Ki represents the number of edge resources required to complete the i-th edge task. is the data transmission volume of edge node i at time t;

[0067] (2) and j = {1, 2, ..., J} denotes edge servers deployed in different edge communities, which allocate computing resources to edge nodes that are physically close to each other;

[0068] (3) and represents the edge nodes deployed around the jth edge server. It is worth noting that the edge servers in the same community can deploy multiple edge nodes. Represents an octet describing an edge node. The set is defined as follows:

[0069] - is the set of edge communities to which the kth edge node belongs;

[0070] - is the set of real-time social attribute values ​​of the kth node at time t, where and Represent the super reputation value, super bonding coefficient and super selfishness value respectively,

[0071] represents the resources available for task scheduling at the kth node at time t, is the clock frequency of the kth edge node at time t, is the transmission power of the kth edge node, is the data transmission volume of edge node k at time t, is the number of tasks that the kth node needs to process, The current operating status of the EEH device configured for the kth node.

[0072] S1: Formalize the MEC scenario based on time-varying hypergraph; edge devices and users in the edge computing network have multi-type relationship characteristics of different domains that evolve over time, which makes it difficult to model ECN in traditional graph theory. Therefore, the hypergraph with time dimension is extended to the MEC environment, which is called time-varying hypergraph. The MEC scenario construction steps of the time-varying hypergraph are as follows:

[0073] (1) Using triple TH t = <V t ,B t ,F t > to describe the time-varying hypergraph, of which the three subset components are as follows:

[0074] -V t = {U t ,S t ,N t} is a finite set of edge servers, edge nodes, and edge users at t;

[0075] - and m={1,2,...,M} is V t The non-empty subsets at time t correspond to different edge communities in real edge scenarios. In addition, they also meet the following restrictions:

[0076] -F t Represents the changes in real edge scenarios and demand response strategies in the future. t ={wake_node,off_node,add_user,delete_user} and contains V t operations; the functional semantics of these operations are described in Table 1.

[0077] Table 1. Semantic description of the operation set

[0078]

[0079]

[0080] 1)

[0081] 2) B t ∈V t

[0082] 3)

[0083] 4)

[0084] (2) In order to perform real-time demand response analysis on the edge system, the association matrix of the edge node is set to Describing TH in terms of triples of time-varying hypergraphs t = <V t ,B t ,F t >, incidence matrix The following conditions are met:

[0085] - The line with EH t The vertices of are related, that is, edge nodes;

[0086] - The columns and EH t The hyperedge correlation in ;

[0087] -The information represented by the correlation matrix is ​​shown in Table 2:

[0088] Table 2. Correlation matrix information table

[0089]

[0090] (3) In order to better manage the node relationships in the edge community, the concept of cardinality is introduced, namely where the cardinality in the hyperedge is the number of edge nodes in the edge community m at time t in the time-varying hypergraph, Calculate as formula (1):

[0091]

[0092] In the formula is the incidence matrix at time t The number of columns, Y(*) is the indicator function, when Y(*) is true, its value is 1, otherwise it is 0;

[0093] Will m={1,2,...,M} represents the cardinality set of all edge node rows in the hypergraph.

[0094] To illustrate how to use a time-varying hypergraph to model an ECN that evolves over time, Figure 1 (a), (b) and (c) show the demand response strategies of the edge system at time t1, t2 and t3 respectively; Figure 1 As shown, the time-varying hypergraph considered by the present invention contains a total of 5 hyperedges, each of which contains an edge server and multiple edge nodes scattered in the hyperedge. At the same time, 10 edge users randomly scattered in the ECN are considered and task requests are issued; as time changes, the demand response strategy uses wake_node and off_node operations to save the energy consumption of the entire system according to the needs of real-time edge users. On the other hand, the present invention will suspend abnormal nodes to effectively avoid node deception. In order to minimize system energy consumption, 6 operations are performed: and The correlation matrix of t1 and t2 and As shown below:

[0095]

[0096] After a period of time, at time t3, if Figure 1 As shown in (c), it can be found that the user and When exiting ECN at time t3, the super edge {e3(Medical Community)} without edge users also disappears; then, three new users u x 、u y and u z At t3, ECN is added. These processes can use the operation set To achieve: delete_user(EH,u4,t3), delete_user(EH,u7,t3), delete_user(EH,u 10 ,t3)、add_user(EH,u x ,t3)、add_user(EH,u y ,t3) and add_user(EH,u z ,t3). Finally, we get the correlation matrix

[0097]

[0098] S2: Design the edge efficiency model, namely EU model, and design the edge energy collection model;

[0099] The steps for establishing the edge effectiveness model are:

[0100] The parameters of the EU model include task completion time CT, task failure rate FR, total data rate DR and system energy consumption EC. The formal expressions of each parameter are as follows:

[0101] (1) CT consists of three parts: task transmission time, computation time, and waiting time. The transmission time Γ of task i from the end user device to the edge node is calculated. tran (t) is shown in formula (2):

[0102]

[0103] In the formula is the wireless channel bandwidth from the terminal device to the edge node, θ ik is the channel gain of task i arriving at node k for the first time, θ ki is the channel gain of the last return of task i to node k, ρ is the noise power spectral density;

[0104] The total computation time Γ of task i com (t) is shown in formula (3):

[0105]

[0106] In summary, the total CT of all computing tasks is shown in formula (4):

[0107]

[0108] (2) FR represents the proportion of edge tasks that fail to be successfully scheduled within a specific time frame of MEC or under specific operating conditions and a measure of execution efficiency, which can be calculated by formula (5):

[0109]

[0110] Where K i and Represent the number of resources required and remaining resources of MEC respectively. If the edge task The number of required resources is greater than the number of remaining resources, or the waiting time is greater than the computing time, indicating that the user is on the edge of the task Scheduling failed;

[0111] (3) DR is a measure of the total amount of data transmission between edge nodes and edge servers within a certain period of time. It represents the total data transmission rate and takes into account the upload and download speeds. DR can be calculated by formula (6), where Indicates DR, represents the amount of data transmitted by edge node k in time period t;

[0112]

[0113] (4) In the present invention, the EC of the edge device consists of two parts: the energy consumption when the local CPU is awakened and the energy consumption during operation. The power consumption of the local CPU awakening is mainly consumed by the device CPU. k is the effective switching capacitance of the CPU of edge device k, which is determined by the chip structure of the edge device. To represent the energy consumed by the local CPU execution, as shown in formula (7):

[0114]

[0115] According to the above task transmission time, let η(t) be the transmission power of edge node k in time slot t, then calculate the operating energy consumption As shown in formula (8), and defined is the total EC of the system;

[0116]

[0117] like Figure 2 As shown in the figure, the edge server can obtain energy from the nearby environment through two ways: external energy station (ees) and internal energy station (ies), so as to achieve the purpose of green MEC; specifically, the edge node will be equipped with intelligent sensors and control algorithms to sense the current node status and battery capacity; when the edge node needs energy supply, it can use the energy absorption device deployed in the EEH to obtain energy from the ees. The ees mainly include several nearby TV transmission towers and various base stations. When the edge node battery is about to run out, the EEH will automatically trigger the IES-based wireless reverse charging to provide energy for the node, ensuring that the edge node can normally participate in subsequent resource allocation tasks.

[0118] The method for establishing the edge energy collection model is:

[0119] Define the energy obtained by edge node k from ees and ies as and Will define is the power received by edge node i from the nearby ees RF source. According to the Friis equation, It can be defined by formula (9):

[0120]

[0121] In the formula, and Indicates the lth node close to the kth edge node The transmit power of the ambient RF source and the transmit antenna gain, represents the receiving antenna gain of edge node k, c represents the propagation speed of electromagnetic waves in free space, d k,l represents the distance between the kth edge node and the lth ambient RF source nearby, f l t is the emission frequency of the lth environmental RF source;

[0122] Calculate the energy obtained by edge node k from ees in time slot t As shown in formula (10):

[0123]

[0124] Among them, the parameters is the energy collection efficiency of edge node k, T ees represents the time during which k continuously obtains energy from ees;

[0125] Introducing the energy obtained by edge node k from ies through wireless reverse charging Edge nodes k and ies are both equipped with magnetic induction coils. Therefore, there is mutual inductance between the receiving coil at edge node k and the transmitting coil at edge node ies. The magnitude of the mutual inductance can be calculated by formula (11):

[0126]

[0127] where μ represents the magnetic permeability of free space, represents the number of turns of the transmitting coil in ies. Similarly, represents the number of turns of the receiving coil of edge node k, and Represent the radii of the transmitting and receiving coils respectively, and the denominator Indicates the distance between the receiving coil and the transmitting coil;

[0128] Based on the existing research, the present invention considers the feasibility of EEH in real edge scenarios and its related resistance materials, and finds that the transmitting coil resistance of ies should be and the receiving coil resistance at edge node k Designed to meet daily maintenance requirements while minimizing energy consumption. and The size of is shown in formula (12) and formula (13):

[0129]

[0130]

[0131] Where R0 is the fixed resistance per unit length of the induction coil. In addition, since the self-inductance is affected by many factors, the self-inductance of the transmitting coil of ies is and the self-inductance of the receiving coil at edge node k They are respectively expressed as formula (14) and formula (15):

[0132]

[0133]

[0134] In the unit time slot t, we try to match the impedance to maximize the received power of edge node k, so we set the load impedance value to It is defined as the complex conjugate of the output impedance of the secondary circuit, as shown in equation (16):

[0135]

[0136] Where ω t is the angular frequency, which can be expressed as express, represents the operating frequency of time slot t, and the letter j represents the imaginary part of the complex number;

[0137] In addition, the induced voltage of the ies receiving coil As shown in formula (17):

[0138]

[0139] in is the power supply voltage. According to the equivalent circuit principle, the received power at time slot t is Equal to the load impedance The power consumption is shown in formula (18):

[0140]

[0141] Assume that the total time for edge node k to be wirelessly charged by ies is Ties , then the total energy obtained by edge node k through ies As shown in formula (19):

[0142]

[0143] Finally, according to and The total energy provided by the described EEH technology to the edge node k is shown in formula (20):

[0144]

[0145] At the same time, in order to prevent EEH from over-discharging, its battery discharge is constrained, as shown in formula (21):

[0146]

[0147] in and are the minimum and maximum values ​​of the power supply energy per unit time respectively;

[0148] In summary, we consider how to minimize the energy consumed by edge nodes in each time slot t to achieve a more energy-efficient GEC-CS strategy. Based on existing research, the energy consumption formula is optimized according to EEH, and the total energy consumption of edge node k in time slot t is As shown in formula (22):

[0149]

[0150] Therefore, in the overall edge scenario architecture, the total energy consumption of all edge nodes in time slot t is As shown in formula (23):

[0151]

[0152] S3: Establish a joint optimization model;

[0153] The steps for establishing the joint optimization model are:

[0154] Assume TH t is a hypergraph about ECN, TH t There are M super communities, randomly distributed in the super communities edge nodes and n edge users; in addition, the edge servers in the super community assign the edge tasks initiated by the end users to the edge nodes near the physical location for offloading. At the same time, the constrained optimization problem COP model of the EU minimization problem is constructed based on the ECN hypergraph. In the offloading process, the EU model jointly gives four measurable indicators, which are CTΓ total(t), FRγ, DR and EC In addition, weight coefficients α, β, γ, and δ are introduced to balance the interests between each other. The constraint relationship of the weight coefficients is α+β+γ+δ=1 and {α, β, γ, δ}>0. In summary, formula (24) is the optimization goal of this study:

[0155]

[0156]

[0157]

[0158]

[0159]

[0160]

[0161]

[0162] Formula (25) constrains the scope, which limits the edge nodes and edge tasks from appearing in multiple communities, requiring that tasks generated by end users can only be scheduled by edge nodes in the same community; Formula (26) constrains deception behavior, limiting the actual resources of the node to be greater than the resources required to complete the computing task; Formulas (27) and (28) jointly constrain energy to prevent EEH from oversupplying energy and limit excessive charging and discharging of edge nodes, where is the maximum energy possessed by the edge node in time slot t; Formula (29) can ensure that the EEH works normally for a long time, and the maximum transmission power of the lth environmental RF source near the edge node k shall not exceed Formula (30) constrains the data rate of each terminal user assigned to slot t by edge node k to be no less than Λ min .

[0163] S4: With the support of social contract theory, a MEC contract system based on a time-varying hypergraph is constructed; the contract system considered in this invention is based on the hypergraph ECN model, and analyzes the demand, supply, and social relations between edge nodes and end users in multiple iterable hypercommunities. From the perspective of social contract theory, an algorithm for constructing a contract system based on social relations is proposed, and the following three definitions are related to the algorithm. The steps for establishing the MEC contract system based on a time-varying hypergraph are:

[0164] (1) Super reputation Hr refers to the reputation level of edge nodes in ECN based on their past behaviors and interactions. In each edge community, the corresponding edge server is responsible for real-time monitoring of edge nodes in the community. When an edge node applies for a scheduling task, the edge server will collect information such as the number of resources, clock frequency, and energy consumption of the node in real time. In addition, the edge server evaluates the Hr of the edge node based on its historical scheduling behavior. Hr is the highest weight indicator for judging the degree of deception of edge nodes.

[0165] Assuming x is represented as an edge node, its Hr in the corresponding edge community is defined as formula (31):

[0166]

[0167] In the formula, Hr org is the initial value of the new edge node Hr of the super community; Complete the first The Hr value of edge node x after each task; The weight of the last task evaluation, used to evaluate the Importance of each task and balanced historical evaluation; For edge servers in scheduling The behavior evaluation function of the edge node x when performing a task is further defined as formula (32):

[0168]

[0169] In the formula, R x Indicates the computing resources actually owned by node x, Indicates completion of The number of resources required for each task; Indicates that the available resources of node x are less than The resource requirements of edge tasks, therefore, due to the deception of edge node x, the edge server sets the task execution evaluation value of node x to 0; Indicates that the available resources of node x are sufficient to schedule A marginal task; and Indicates the actual computing time and expected computing time of edge node x for the scheduled task; and Indicates The actual transmission time and expected transmission time of each task; It is The evaluation weight of the task completion time is used to balance the importance of the current task and the historical task completion time. Obviously, the greater the gap between the task calculation time and the transmission time and the expected value, the smaller the node's Hr.

[0170] (2) The hyper-bond coefficient Hb is an evaluation indicator used to evaluate the strength and closeness of the social connection between users and edge nodes in the hyper-community. Hb represents the satisfaction of edge users with the services provided by the assigned nodes. Hr objectively calculates the credibility of edge nodes based on the historical scheduling behavior of edge nodes. Hb represents the subjective evaluation of edge users in terms of service differences and execution. Obviously, the higher the Hb value, the more satisfied the edge users are with the services of a certain edge node. In addition, due to the large randomness of subjective factors, many constraints are set to ensure the effectiveness of Hb.

[0171] After each task is completed, the user must evaluate the completion of the task as an evaluation of the edge node, for edge node x, using the expected total completion time of the task CT(x) Exp As a measure of service quality, the edge node is evaluated. During the evaluation process, the edge task is assumed to be completed successfully. If the task fails or is migrated by the edge server, the Hb value of the node is directly evaluated as 0. The service difference function is defined as shown in formula (33):

[0172]

[0173] in For the The actual completion time of the task, if the service difference If the threshold λ is exceeded, the service of node x is considered unreliable. In summary, the first The task evaluation value As shown in formula (34):

[0174]

[0175] The edge node Hb value evaluated by edge user y is shown in formula (35):

[0176]

[0177] In the formula, Hb org is the initial value of the new edge node Hr of the super community; is the evaluation weight of y to x on the previous task, which is used to balance the importance of the evaluation of y to x on the previous task and the evaluation of y to x on the previous task; dist x,y The weight of the distance between edge user y and edge node x. The closer the user is to the edge node, the shorter the task transmission time and the higher the evaluation value. Therefore, the distance weight dist is set. x,y (0≤dist x,y ≤1) to balance the relationship between Hb and the distance factor. The farther the distance between the user and the node, the greater the distance between the user and the node.x,y The smaller;

[0178] In addition, the subjective factors of user evaluation are also considered. If users maliciously slander or overestimate edge nodes in the super community, it will directly affect the Hb value of edge nodes. Therefore, formula (36) is used to test the accuracy of node user evaluation:

[0179]

[0180] If the difference between x and y exceeds the threshold σ, user y's evaluation of edge node x Invalid, which can avoid the subjective factors of human evaluation to a certain extent and prevent slander or exaggeration;

[0181] (4) The super selfishness value Hs is a measure used to quantify the degree of individualistic behavior in the super community. This behavior tends to pursue the maximization of one's own interests while ignoring the needs and interests of other edge nodes and users. The comprehensive reputation Cr value of the edge node is calculated by weighting Hr and Hb, as shown in formula (37):

[0182]

[0183] in Execute the first Cr for each task; τ (0.5≤τ≤1) is the weight coefficient between Hr and Hb;

[0184] According to The value of Hs can be obtained as shown in formula (38):

[0185]

[0186] In the formula is the Hs of edge node x, is the threshold. If n tasks are executed continuously, the Cr of edge node x is less than the threshold. but If 1 is selected, this situation indicates that edge node x often cheats, or cannot meet the needs of edge users in terms of task completion time and data transmission rate. Such nodes are abnormal nodes, that is, they cannot normally complete edge tasks generated by edge users;

[0187] Since computing resources and the number of nodes in the edge community are limited, it is unrealistic to permanently prohibit the scheduling behavior of edge nodes. Therefore, in order to reduce the frequency of abnormal behavior of edge nodes and improve the resource utilization efficiency of nodes, a suspension mechanism that allows confession is proposed. Formula (39) represents the time when the node scheduling capability is suspended:

[0188]

[0189] In the formula Indicates the time to suspend the scheduling capability of node x, and to activate its task scheduling function again after the suspension ends. For the CT of edge tasks, is the unit time that node x is suspended;

[0190] The edge server in the super community will schedule the contract node according to the end-user task allocation and calculate the Hr, Hb and Hs of the node; in the contract system, the edge server classifies the nodes according to the real-time social attribute value of the node, and the classification of the contract node set is based on formula (40):

[0191]

[0192] Where Ψ is the set of contract nodes; is the threshold value. The value of the edge node x after tasks is greater than , it is defined as a contract node;

[0193] The classification of abnormal node sets is based on formula (41):

[0194]

[0195] Where ψ is the set of abnormal nodes. After the task is completed, the edge node When the value of is equal to 1, it is defined as an abnormal node;

[0196] In the edge community, some edge nodes are not assigned tasks, which means that the contract nodes at this time are sufficient to meet the needs of current users. Other nodes that are not assigned tasks by the edge server are called dormant nodes, waiting to be awakened by the edge server. The classification of the dormant node set is shown in formula (42):

[0197]

[0198] In the formula is a set of dormant nodes;

[0199] After completing the above steps, the detailed steps of building a MEC contract system based on a time-varying hypergraph are as follows: First, obtain the vertex information of end users, edge nodes, and edge servers from the edge community of TH at time t1, and calculate the vertex information of end users, edge nodes, and edge servers according to the association matrix Calculate the cardinality of a node in a specific supercommunity Secondly, according to the various indicators and service difference functions of edge nodes when executing tasks Compute edge nodes and Third, by integrating comprehensive reputation With threshold and By comparison, the contract node set Ψ and the abnormal node set ψ are obtained; in addition, the edge nodes that are not currently assigned tasks by the edge server are added to the dormant node set, and finally the real-time node information after node classification is returned to the association matrix and updates TH at time t2; it is worth noting that each iteration of the contract system is composed of or t) Operation initiated.

[0200] S5: Establishing an EEH strategy based on a contract system; The present invention proposes an edge energy harvesting mechanism based on a contract system, such as Figure 3 As shown in the figure, the specific steps to establish the EEH strategy based on the contract system are:

[0201] First, the vertex information in TH at time t2 is traversed. In the contract node set Ψ, EEH will always power the contract node, and ies will always reverse charge the contract node until its resource allocation phase ends. This process is called contract energy harvesting CEH. Secondly, in the abnormal node set ψ, since the scheduling function of the abnormal node is prohibited by the edge server, its EEH is always in a suspended state until the suspended state of the node is released. This process is called abnormal energy harvesting AEH. In addition, in the dormant node set In the process, the dormant nodes will always receive energy from ees, but only when the energy is lower than a certain threshold or a new user joins the community to replenish energy in time, the dormant nodes will start wireless reverse charging. This process is called dormant energy harvesting DEH; finally, the total energy of all nodes obtained from EEH (the sum of ees and ies) is returned to table L.

[0202] S6: Use the EEH strategy based on the contract system to formulate the demand response logic, minimize the edge efficiency model, and prevent edge node deception, ultimately achieving the goal of green MEC; The specific steps of using the EEH strategy based on the contract system to formulate the demand response logic are:

[0203] exist In the example, suppose an edge user joins the super edge {e5(Work Community)} and applies for an offload task At this time, the edge server first calculates the Calculate the cardinality of nodes in a superedge To ensure that there are available nodes in the super community to meet user needs; secondly, user tasks are aggregated and assigned to the contract node with the nearest physical location that meets the offloading resource requirements; Figure 1As shown in (c), the edge server s5 traverses the hyperedge association matrix and detects that the edge node n4 closest to the user's physical location has abnormal behavior such as deception, and assigns the task to the contract node u with the second-best physical location. x If no new users join or apply for new edge tasks, the idle edge nodes will be put to sleep to reduce energy consumption.

[0204] Selection of benchmark method and simulation parameters:

[0205] For the constructed ECN scenario, the following five algorithms are implemented as benchmark algorithms compared with the GEC-CS algorithm:

[0206] (1) GEC-OPT: This method generates an optimal solution based on task scheduling and optimally solves the integer programming model for a single time slot. As a frontier algorithm for comparison, this method performs well compared to other baseline models, but is only applicable to small-scale edge scenarios.

[0207] (2)GEC-RF: This method randomly assigns computing tasks to the edge nodes closest to the physical location of the edge community without considering the overall energy consumption of the system.

[0208] (3) GEC-RRSA: This method is a traditional polling algorithm. In the edge scenario, the edge server will poll all edge nodes in its edge community and assign tasks.

[0209] (4) GEC-GA: Genetic algorithm is a heuristic algorithm that allocates computing tasks to the most powerful edge nodes without considering the energy consumption of the entire system (the present invention gives priority to contract nodes and then compares the remaining resources).

[0210] (5) GEC-EDR: In the present invention, this method can power the edge nodes that consume the most energy and try to assign computing tasks to edge servers, thereby deactivating the idle edge servers as a whole.

[0211] The present invention uses the EdgeCloudSim simulation platform to test the performance of the scheme, in which JAVA is used to encode the algorithm. In the experiment, the test environment is carried out on a Windows machine (Intel i7-13700 CPU, 16GB RAM), and the extended EdgeCloudSim platform is used to simulate the ECN under the community structure based on the hypergraph. Several underlying communities are established in the ECN (simulating different edge scenarios in the real world), and each community has an edge server and several edge nodes adjacent to the physical location to provide computing services for edge users. The number of mobile devices is set to 1000 (the generation rule conforms to the Poisson distribution). Before the simulation starts, the node bandwidth is between 300 and 500 mbps, and the computing power of the node server is set to 1000 MIPS and the mobile server is set to 2000 MIPS. The storage space of the edge node and edge server is set to 3-4g and 10GB, and the number of subtask groups is randomly set to a value between 2 and 4. Table 3 shows some basic experimental parameters.

[0212] Table 3. Simulation parameters

[0213]

[0214]

[0215] EEH related parameters are: c = 3 × 108 m / s, T ees =1800s, In addition, the parameters based on ies are defined as μ = 4π × 10-7 H / m, R0=0.01Ω / m, T ies =10s,

[0216] In general, the GEC-CS strategy aims to solve user computing tasks with minimal edge system energy consumption while improving user QoS with the lowest possible task failure rate and task completion time. The performance of GEC-CS may vary in different edge scenarios, depending on many factors, such as the number of edge computing task groups, the number of edge nodes, and the amount of resources currently available for task scheduling on these edge nodes. Therefore, in order to fully evaluate the performance of GEC-CS, we simulated various GEC-CS edge scenarios: (1) the number of task groups; (2) the number of edge nodes; (3) the size of computing resources of edge nodes. Each edge node is concentrated in the edge server, and the amount of computing resources of the edge node is automatically generated according to the normal distribution. The computing resource (R) is the average value of the normal distribution. It should be noted that in ECN, an edge community is allocated 3-5 edge nodes on average, while an edge server will centrally manage 1-3 edge communities, so as the number of edge nodes (n) increases, the scale of the edge scenario will also increase. The experimental settings are summarized in Table 4:

[0217] Table 4. Experimental parameters

[0218]

[0219] Description of demand response effectiveness and simulation results:

[0220] (1) EEH case study

[0221] In the present invention, a specific edge community is used as a case to verify the effectiveness of EEH. Assume that there are two TV towers and two base stations as environmental RF sources within the ees, marked as TV 1, TV 2, BS1 and BS2, with coordinates in the meter coordinate system of (1600, 1600), (-1600, 1600), (140, -140) and (-140, -140), respectively. The edge nodes are randomly distributed in multiple edge communities. For illustration, within a single super community, the vertex coordinates are (10, 10), (-10, 10), (10, -10), (-10, -10) and (0, 0). Table IV gives the energy collected by different edge nodes from nearby environmental RF sources, and wireless reverse charging is performed based on the ies within the edge community. The detailed energy distribution is shown in Table IV. Figure 4 shown.

[0222] In this simulation, the number of edge nodes is set to 5, of which nodes 2 and 4 are designated as contract nodes, nodes 1 and 3 are designated as dormant nodes, and node 5 is designated as an abnormal node. The analysis of energy data shows that due to the stronger transmission capacity of TV towers compared to base stations, more energy can be provided to edge nodes. In addition, the wireless reverse charging technology of IES has a higher receiving power, which enables nodes to quickly accumulate a large amount of energy. It can be clearly seen from Table 4 that contract nodes 2 and 4 obtain more energy from Ethereum because they are always powered within the EEH framework. In contrast, dormant edge nodes continuously receive energy from EES, but only start IES wireless reverse charging when their energy level is below a certain threshold or when a new user joins the community. In addition, the table also shows that abnormal node 5 does not receive energy supply from the EEH system because its abnormal state limits its scheduling function.

[0223] (2) Comparison of algorithm performance in different edge scenarios

[0224] Figure 5-7 The performance of the six methods in different edge scenarios is compared, and the impact of u, n, and R in ECN is demonstrated. Overall, GEC-OPT successfully schedules more tasks with the lowest total energy consumption and fastest processing time, and achieves the highest overall data rate. GEC-CS achieves the second highest performance. Figure 5 As can be seen in (a), the increase of u in group 1 leads to an increasing trend in the total system energy consumption generated by each method. As the EEH continues to supply power to the nodes, the total energy consumption increases slowly from the initial rapid growth. Figure 5 As shown in (b) and 5(c), the overall data rate and task failure rate obtained by these methods will increase similarly with the linear increase in the number of edge tasks. As long as the cumulative resource demand of mobile users does not exceed the total capacity of all edge nodes in the super community, it can be expected that Figure 8 Increase in task completion time.

[0225] from Figure 6 It can be seen from (a) that when n increases in Group 2, more edge nodes can process computing tasks in parallel when powered by EEH, thereby reducing the load on a single node, so the overall energy consumption of these methods is slightly reduced. It is worth noting that when the scale of the edge scenario increases, the contract system can better distinguish the social attribute categories of each node, and the wide adaptability of EEH makes the gap between the algorithm and GEC-OPT smaller and smaller. As n increases, the average task completion of each method remains stable. Therefore, its average task failure rate also remains stable. From Figure 6As can be seen from (a), only the method of the present invention can significantly improve the total data rate of GEC-OPT, while the other four methods - GEC-RF, GEC-RRSA, GEC-GA and GEC-EDR - do not significantly improve the total data rate with the increase of n. This difference can be attributed to the fact that GEC-RF frequently assigns tasks to abnormal nodes, while GEC-RRSA underutilizes the available resources on edge nodes. In addition, GEC-GA and GEC-EDR lack adaptability to the dynamic characteristics of the ECN proposed by the present invention. Figure 6 As shown in (b), the task completion time decreases as the number of edge nodes increases until all edge tasks are completed, i.e., n ≥ 70.

[0226] like Figure 6 As shown in (c), increasing the resources of a single edge node can promote more efficient task processing and reduce the frequency of task allocation and migration between devices, thereby reducing the communication and coordination energy consumption of the entire ECN. Therefore, as R increases, the system energy consumption produced by these methods is significantly reduced. Figure 7 (a) with Figure 6 (c) Similarly, since increasing the resources of a single node can significantly reduce the number of migrations in the ECN, the task failure rate will be slightly reduced. Figure 7 As shown in (b), the increase of R will increase the user's data transmission rate. When all assigned end users reach the maximum data rate, the overall data rate tends to be stable, that is, R = 4. Figure 7 As shown in (c), using more resources on a single physical machine can usually serve edge users more efficiently, so the task completion time of these methods is significantly reduced.

[0227] (3) Efficiency evaluation in different edge scenarios

[0228] 1) Small-scale edge scenarios

[0229] In order to fully demonstrate the efficiency of the algorithm proposed in this paper, we give the number of iterations required for the demand response strategy at each time interval. In addition to the experimental parameters in Table 4, we also specially designed three small-scale edge scenarios for comparison, as shown in the following table. In the experiment of Table 5, we considered a larger-scale edge scenario, which requires a larger number of edge servers to handle the computing tasks brought by end users. Therefore, compared with the experiments in Table 4, our algorithm requires more iterations to identify contracts and abnormal nodes to help edge servers complete the resource allocation process. In general, in the experiments of Table 4 and Table 5, the effects of variables u, n, and R on the number of iterations required for our algorithm are similar, so we compare and analyze the results.

[0230] Table 5. Small scale experimental parameters

[0231]

[0232]

[0233] like Figure 8 As shown in (b), in group 1.1, as u increases from 1 to 10, the time it takes for GEC-OPT to find a demand response strategy increases exponentially; similarly, Figure 8 (d) and Figure 8 As shown in (f), in groups 2.1 and 3.1, GEC-OPT also increases exponentially with the increase of n and R. Compared with GEC-OPT, the other five methods maintain a small increase, but the solution of demand response in ECN takes almost no time. This also verifies our previous theory that it is impractical to use GEC-OPT to solve large-scale demand response problems. In group 3.1, when R increases from 0.1 to 0.8, a single edge node has more computing resources to serve more mobile users, which increases the average number of options for assigning a single edge task, so GEC-CS needs to spend more time to find the optimal demand response solution. As R continues to grow, the contract nodes in the super community have enough computing resources to accommodate all nearby mobile users, so the total time consumed to make demand response decisions becomes stable.

[0234] 2) Large-scale edge scenarios

[0235] In addition to the above small-scale edge scenarios, we further discuss the performance of GEC-CS in large-scale edge scenarios. Figure 9-10 It is the total time consumption of different methods to find demand response strategy solutions in the large-scale edge scenario constructed in Table 6, and is compared with the total time consumption in the small-scale edge scenario. Among them, the MEC demand response problem is usually regarded as an NP-hard problem, that is, in order to meet the needs of edge users in real time, the demand response algorithm usually provides a suboptimal solution that is close to the optimal solution, which makes GEC-OPT unable to find the optimal demand response solution within a reasonable time. As the scale of the problem increases, the performance of GEC-OPT cannot be used as a control group. Therefore, we will not consider the GEC-OPT method in the large-scale edge scenario in Table 6. In addition, it is worth noting that the EdgeCloudSim platform randomly generates terminal tasks through virtual machines. Although the number of tasks generated in the large-scale edge scenario is the same, there are still differences in the specific parameters of the edge tasks (such as resource requirements and the physical location of the tasks).

[0236] Table 6. Large-scale experimental parameters

[0237]

[0238] like Fig. 9 As shown in (a), in group 1.2, as u increases from 1 to 10, the time it takes for GEC-OPT to find a demand response strategy increases exponentially; similarly, Fig. 9 (b) and Fig. 9 As shown in (c), in Group 2.1 and Group 3.1, GEC-OPT also increases exponentially with the increase of n and R. Compared with GEC-OPT, the other five methods maintain a small increase, but the solution of demand response in ECN takes almost no time. This also verifies our previous theory that it is impractical to use GEC-OPT to solve large-scale demand response problems. In Group 3.1, when R increases from 0.1 to 0.8, a single edge node has more computing resources to serve more mobile users, which increases the average number of options for assigning a single edge task, so GEC-CS needs to spend more time to find the optimal demand response solution, such as Fig. 9 As shown in (c), as R continues to grow, the contract nodes in the super community have enough computing resources to accommodate all nearby mobile users, so the total time consumed in making demand response decisions becomes stable.

[0239] exist Fig.10 In order to better evaluate the efficiency of GEC-CS, we exclude the GEC-OPT method. In groups 1.2-3.2, the time consumption of GEC-CS increases linearly with u, n, and r. Fig.10 The reason for the increase in (a) is simple: the increase in the number of edge users leads to an increase in the algorithm execution time required to make allocation decisions for all edge tasks. Fig.10 In (b), the increase in the number of edge nodes in the super community directly leads to an increase in the number of options available for edge nodes to be assigned, and indirectly leads to an increase in the number of abnormal node deceptions and sleep node startups or shutdowns detected by the edge server. Therefore, compared with other algorithms, the GEC-CS algorithm takes more time to find a demand response strategy. Fig.10 In (c), the total time consumption of the GEC-CS method increases linearly and tends to remain unchanged after R = 60. The reason is similar to Figure 7 (c) is the same. When the amount of resources of a single node reaches 60, the overall data rate reaches a peak, indicating that all tasks are being or will be scheduled by edge nodes. The node differences in the super community remain unchanged, and the total time consumed to generate demand response strategies also remains unchanged. It is worth noting that in Fig.10 In the experiments, GEC-RF and GEC-RRSA usually do not perform complex analysis and prediction of the node status or historical behavior, but make decisions through random selection or polling allocation. This simple and direct method can make the algorithm maintain a certain robustness when facing different user groups, number of machines or resource conditions, thereby maintaining a stable time consumption.

[0240] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A green multi-access edge computing demand response method with EEH based on a contract system, characterized in that: The following steps are involved: S1: Formalize the MEC scenario based on time-varying hypergraph; S2: Design the edge efficiency model, namely EU model, and design the edge energy collection model; S3: Establish a joint optimization model; S4: With the support of social contract theory, we build a MEC contract system based on time-varying hypergraph; S5: Establish EEH strategy based on contract system; S6: Use the EEH strategy based on the contract system to formulate demand response logic, minimize the edge efficiency model, and prevent edge node deception, ultimately achieving the goal of green MEC.

2. The green multi-access edge computing demand response method with EEH based on the contract system according to claim 1, characterized in that: The steps for constructing the MEC scenario of the time-varying hypergraph described in step S1 are: (1) Using triple TH t = <V t ,B t ,F t > to describe the time-varying hypergraph, of which the three subset components are as follows: -V t = {U t ,S t ,N t } is a finite set of edge servers, edge nodes, and edge users at t; - and m={1,2,...,M} is V t The non-empty subsets at time t correspond to different edge communities in real edge scenarios. In addition, they also meet the following restrictions: -F t Represents the changes in real edge scenarios and demand response strategies in the future. t ={wake_node,off_node,add_user,delete_user} and contains V t Operations; (2) In order to perform real-time demand response analysis on the edge system, the association matrix of the edge node is set to Describing TH in terms of triples of time-varying hypergraphs t = <V t ,B t ,F t >, incidence matrix The following conditions are met: - The line with EH t The vertices of are related, that is, edge nodes; - The columns and EH t The hyperedge correlation in ; (3) In order to better manage the node relationships in the edge community, the concept of cardinality is introduced, namely where the cardinality in the hyperedge is the number of edge nodes in the edge community m at time t in the time-varying hypergraph, Calculate as formula (1): In the formula is the incidence matrix at time t The number of columns, Y(*) is the indicator function, when Y(*) is true, its value is 1, otherwise it is 0; Will Represented as the cardinality set of all edge node rows in the hypergraph.

3. The green multi-access edge computing demand response method with EEH based on the contract system according to claim 1, characterized in that: The steps for establishing the edge effectiveness model in step S2 are: The parameters of the EU model include task completion time CT, task failure rate FR, total data rate DR and system energy consumption EC. The formal expressions of each parameter are as follows: (1) CT consists of three parts: task transmission time, computation time, and waiting time. The transmission time Γ of task i from the end user device to the edge node is calculated. tran (t) is shown in formula (2): In the formula is the wireless channel bandwidth from the terminal device to the edge node, θ ik is the channel gain of task i arriving at node k for the first time, θ ki is the channel gain of the last return of task i to node k, ρ is the noise power spectral density; The total computation time Γ of task i com (t) is shown in formula (3): In summary, the total CT of all computing tasks is shown in formula (4): (2) FR represents the proportion of edge tasks that fail to be successfully scheduled within a specific time frame of MEC or under specific operating conditions and a measure of execution efficiency, which can be calculated by formula (5): Where K i and Represent the number of resources required and remaining resources of MEC respectively. If the edge task The number of required resources is greater than the number of remaining resources, or the waiting time is greater than the computing time, indicating that the user is on the edge of the task Scheduling failed; (3) DR is a measure of the total amount of data transmission between edge nodes and edge servers within a certain period of time. It represents the total data transmission rate and takes into account the upload and download speeds. DR can be calculated by formula (6), where Indicates DR, P i t represents the amount of data transmitted by edge node k in time period t; (4) In the present invention, the EC of the edge device consists of two parts: the energy consumption when the local CPU is awakened and the energy consumption during operation. The power consumption of the local CPU awakening is mainly consumed by the device CPU. k is the effective switching capacitance of the CPU of edge device k, which is determined by the chip structure of the edge device. To represent the energy consumed by the local CPU execution, as shown in formula (7): According to the above task transmission time, let η(t) be the transmission power of edge node k in time slot t, then calculate the operating energy consumption As shown in formula (8), and defined is the total EC of the system; 4. The green multi-access edge computing demand response method with EEH based on the contract system according to claim 1, characterized in that: The method for establishing the edge energy collection model in step S2 is: Define the energy obtained by edge node k from ees and ies as and Will define is the power received by edge node i from the nearby ees RF source. According to the Friis equation, It can be defined by formula (9): In the formula, and represents the first node close to the kth edge node The transmit power of the ambient RF source and the transmit antenna gain, represents the receiving antenna gain of edge node k, c represents the propagation speed of electromagnetic waves in free space, d k,l represents the distance between the kth edge node and the lth ambient RF source nearby, f l t is the emission frequency of the lth environmental RF source; Calculate the energy obtained by edge node k from ees in time slot t As shown in formula (10): Among them, the parameters is the energy collection efficiency of edge node k, T ees represents the time during which k continuously obtains energy from ees; Introducing the energy obtained by edge node k from ies through wireless reverse charging Edge nodes k and ies are both equipped with magnetic induction coils. Therefore, there is mutual inductance between the receiving coil at edge node k and the transmitting coil at edge node ies. The magnitude of the mutual inductance can be calculated by formula (11): where μ represents the magnetic permeability of free space, represents the number of turns of the transmitting coil in ies. Similarly, represents the number of turns of the receiving coil of edge node k, and Represent the radii of the transmitting and receiving coils respectively, and the denominator Indicates the distance between the receiving coil and the transmitting coil; and The size of is shown in formula (12) and formula (13): Where R0 is the fixed resistance per unit length of the induction coil. In addition, since the self-inductance is affected by many factors, the self-inductance of the transmitting coil of ies is and the self-inductance of the receiving coil at edge node k They are respectively expressed as formula (14) and formula (15): In the unit time slot t, we try to match the impedance to maximize the received power of edge node k, so we set the load impedance value to It is defined as the complex conjugate of the output impedance of the secondary circuit, as shown in equation (16): Where ω t is the angular frequency, which can be expressed as express, represents the operating frequency of time slot t, and the letter j represents the imaginary part of the complex number; In addition, the induced voltage of the ies receiving coil As shown in formula (17): in is the power supply voltage. According to the equivalent circuit principle, the received power at time slot t is Equal to the load impedance The power consumption is shown in formula (18): Assume that the total time for edge node k to be wirelessly charged by ies is T ies , then the total energy obtained by edge node k through ies As shown in formula (19): Finally, according to and The total energy provided by the described EEH technology to the edge node k is shown in formula (20): At the same time, in order to prevent EEH from over-discharging, its battery discharge is constrained, as shown in formula (21): in and are the minimum and maximum values ​​of the power supply energy per unit time respectively; Based on the existing research, the energy consumption formula is optimized according to EEH, and the total energy consumption of edge node k in time slot t is As shown in formula (22): Therefore, in the overall edge scenario architecture, the total energy consumption of all edge nodes in time slot t is As shown in formula (23):

5. The green multi-access edge computing demand response method with EEH based on the contract system according to claim 1, characterized in that: The steps for establishing the joint optimization model in step S3 are: Assume TH t is a hypergraph about ECN, TH t There are M super communities, randomly distributed in the super communities edge nodes and n edge users; in addition, the edge servers in the super community assign the edge tasks initiated by the end users to the edge nodes near the physical location for offloading. At the same time, the constrained optimization problem COP model of the EU minimization problem is constructed based on the ECN hypergraph. In the offloading process, the EU model jointly gives four measurable indicators, which are CTΓ total (t), FRγ, DR and EC In addition, weight coefficients α, β, γ, and δ are introduced to balance the interests between each other. The constraint relationship of the weight coefficients is α+β+γ+δ=1 and {α, β, γ, δ}>

0. In summary, formula (24) is the optimization goal of this study:

6. The green multi-access edge computing demand response method with EEH based on the contract system according to claim 1, characterized in that: The steps for establishing the MEC contract system based on the time-varying hypergraph described in step S4 are: (1) Super reputation Hr refers to the reputation level of edge nodes in ECN based on their past behaviors and interactions. In each edge community, the corresponding edge server is responsible for real-time monitoring of edge nodes in the community. When an edge node applies for a scheduling task, the edge server will collect information such as the number of resources, clock frequency, and energy consumption of the node in real time; In addition, the edge server evaluates the Hr of the edge node based on its historical scheduling behavior. Hr is the highest weight indicator for judging the degree of deception of the edge node; Assuming x is represented as an edge node, its Hr in the corresponding edge community is defined as formula (31): In the formula, Hr org is the initial value of the new edge node Hr of the super community; Complete the first The Hr value of edge node x after each task; The weight of the last task evaluation, used to evaluate the Importance of each task and balanced historical evaluation; For edge servers in scheduling The behavior evaluation function of the edge node x when performing a task is further defined as formula (32): In the formula, R x Indicates the computing resources actually owned by node x, Indicates completion of The number of resources required for each task; Indicates that the available resources of node x are less than The resource requirements of edge tasks, therefore, due to the deception of edge node x, the edge server sets the task execution evaluation value of node x to 0; Indicates that the available resources of node x are sufficient to schedule A marginal task; and Indicates the actual computing time and expected computing time of edge node x for the scheduled task; and Indicates The actual transmission time and expected transmission time of each task; It is The evaluation weight of the task completion time is used to balance the importance of the current task and the historical task completion time. Obviously, the greater the gap between the task calculation time and the transmission time and the expected value, the smaller the node's Hr. (2) The hyper-tie coefficient Hb is an evaluation index used to evaluate the strength and closeness of the social connection between users and edge nodes in the hyper-community. Hb represents the satisfaction of edge users with the services provided by the assigned nodes. Hr objectively calculates the credibility of edge nodes based on the historical scheduling behavior of edge nodes. Hb represents the subjective evaluation of edge users in terms of service differences and execution. Obviously, the higher the Hb value, the more satisfied the edge users are with the services of a certain edge node. After each task is completed, the user must evaluate the completion of the task as an evaluation of the edge node, for edge node x, using the expected total completion time of the task CT(x) Exp As a measure of service quality, the edge nodes are evaluated. During the evaluation process, it is assumed that the edge tasks are completed successfully; If the task fails or is migrated by the edge server, the Hb value of the node is directly evaluated as 0, and the service difference function is defined as shown in formula (33): in For the The actual completion time of the task, if the service difference If the threshold λ is exceeded, the service of node x is considered unreliable. In summary, the first The task evaluation value As shown in formula (34): The edge node Hb value evaluated by edge user y is shown in formula (35): In the formula, Hb org is the initial value of the new edge node Hr of the super community; is the evaluation weight of y to x on the previous task, which is used to balance the importance of the evaluation of y to x on the previous task and the evaluation of y to x on the previous task; dist x,y The weight of the distance between edge user y and edge node x. The closer the user is to the edge node, the shorter the task transmission time and the higher the evaluation value. Therefore, the distance weight dist is set. x,y (0≤dist x,y ≤1) to balance the relationship between Hb and the distance factor. The farther the distance between the user and the node, the greater the distance between the user and the node. x,y The smaller; In addition, the subjective factors of user evaluation are also considered. If users maliciously slander or overestimate edge nodes in the super community, it will directly affect the Hb value of edge nodes. Therefore, formula (36) is used to test the accuracy of node user evaluation: If the difference between x and y exceeds the threshold σ, user y's evaluation of edge node x Invalid, which can avoid the subjective factors of human evaluation to a certain extent and prevent slander or exaggeration; (3) The super selfishness value Hs is a measure used to quantify the degree of individualistic behavior in the super community. This behavior tends to pursue the maximization of one's own interests while ignoring the needs and interests of other edge nodes and users. The comprehensive reputation Cr value of the edge node is calculated by weighting Hr and Hb, as shown in formula (37): in Execute the first Cr for each task; τ (0.5≤τ≤1) is the weight coefficient between Hr and Hb; According to The value of Hs can be obtained as shown in formula (38): In the formula is the Hs of edge node x, is the threshold. If n tasks are executed continuously, the Cr of edge node x is less than the threshold. but If 1 is selected, this situation indicates that edge node x often cheats, or cannot meet the needs of edge users in terms of task completion time and data transmission rate. Such nodes are abnormal nodes, that is, they cannot normally complete edge tasks generated by edge users; Since computing resources and the number of nodes in the edge community are limited, it is unrealistic to permanently prohibit the scheduling behavior of edge nodes. Therefore, in order to reduce the frequency of abnormal behavior of edge nodes and improve the resource utilization efficiency of nodes, a suspension mechanism that allows confession is proposed. Formula (39) represents the time when the node scheduling capability is suspended: In the formula Indicates the time to suspend the scheduling capability of node x, and to activate its task scheduling function again after the suspension ends. For the CT of edge tasks, is the unit time that node x is suspended; The edge server in the super community will schedule the contract node according to the end-user task allocation and calculate the Hr, Hb and Hs of the node; in the contract system, the edge server classifies the nodes according to the real-time social attribute value of the node, and the classification of the contract node set is based on formula (40): Where Ψ is the set of contract nodes; is the threshold value. The value of the edge node x after tasks is greater than , it is defined as a contract node; The classification of abnormal node sets is based on formula (41): Where ψ is the set of abnormal nodes. After the task is completed, the edge node When the value of is equal to 1, it is defined as an abnormal node; In the edge community, some edge nodes are not assigned tasks, which means that the contract nodes at this time are sufficient to meet the needs of current users. Other nodes that are not assigned tasks by the edge server are called dormant nodes, waiting to be awakened by the edge server. The classification of the dormant node set is shown in formula (42): In the formula is a set of dormant nodes; After completing the above steps, the detailed steps of building a MEC contract system based on a time-varying hypergraph are as follows: First, obtain the vertex information of end users, edge nodes, and edge servers from the edge community of TH at time t1, and calculate the vertex information of end users, edge nodes, and edge servers according to the association matrix Calculate the cardinality of a node in a specific supercommunity Secondly, according to the various indicators and service difference functions of edge nodes when executing tasks Compute edge nodes and Third, by integrating comprehensive reputation With threshold and By comparison, the contract node set Ψ and the abnormal node set ψ are obtained; in addition, the edge nodes that are not currently assigned tasks by the edge server are added to the dormant node set, and finally the real-time node information after node classification is returned to the association matrix and updates TH at time t2; it is worth noting that each iteration of the contract system is composed of or Operation initiated.

7. The green multi-access edge computing demand response method with EEH based on the contract system according to claim 6, characterized in that: The specific steps of establishing the EEH strategy based on the contract system in step S5 are: First, the vertex information in TH at time t2 is traversed. In the contract node set Ψ, EEH will always power the contract node, and ies will always reverse charge the contract node until its resource allocation phase ends. This process is called contract energy harvesting CEH. Secondly, in the abnormal node set ψ, since the scheduling function of the abnormal node is prohibited by the edge server, its EEH is always in a suspended state until the suspended state of the node is released. This process is called abnormal energy harvesting AEH. In addition, in the dormant node set In the process, the dormant nodes will always receive energy from ees, but only when the energy is lower than a certain threshold or a new user joins the community to replenish energy in time, the dormant nodes will start wireless reverse charging. This process is called dormant energy harvesting DEH; finally, the total energy of all nodes obtained from EEH is returned to table L.

8. The green multi-access edge computing demand response method with EEH based on the contract system according to claim 7, characterized in that: The specific steps of formulating the demand response logic using the EEH strategy based on the contract system in step S6 are: exist In the example, suppose an edge user joins the super edge {e5(Work Community)} and applies for an offload task At this time, the edge server first calculates the Calculate the cardinality of nodes in a superedge To ensure that there are available nodes in the super community to meet user needs; secondly, the user task is aggregated and assigned to the contract node with the nearest physical location and meeting the unloading resource requirements; the edge server s5 traverses the super edge association matrix and detects that the edge node n4 closest to the user's physical location has abnormal behaviors such as deception, and assigns the task to the contract node u with the second best physical location. x If no new users join or apply for new edge tasks, the idle edge nodes will be put to sleep to reduce energy consumption.