CN112132202A - Edge computing collaborative member discovery method based on comprehensive trust evaluation - Google Patents

Edge computing collaborative member discovery method based on comprehensive trust evaluation Download PDF

Info

Publication number
CN112132202A
CN112132202A CN202010985519.5A CN202010985519A CN112132202A CN 112132202 A CN112132202 A CN 112132202A CN 202010985519 A CN202010985519 A CN 202010985519A CN 112132202 A CN112132202 A CN 112132202A
Authority
CN
China
Prior art keywords
node
nodes
cooperative
trust
service
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010985519.5A
Other languages
Chinese (zh)
Other versions
CN112132202B (en
Inventor
乐光学
杨晓慧
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiaxing University
Original Assignee
Jiaxing University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jiaxing University filed Critical Jiaxing University
Priority to CN202010985519.5A priority Critical patent/CN112132202B/en
Publication of CN112132202A publication Critical patent/CN112132202A/en
Application granted granted Critical
Publication of CN112132202B publication Critical patent/CN112132202B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44—Arrangements for executing specific programs
    • G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
    • G06F9/45533—Hypervisors; Virtual machine monitors
    • G06F9/45558—Hypervisor-specific management and integration aspects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44—Arrangements for executing specific programs
    • G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
    • G06F9/45533—Hypervisors; Virtual machine monitors
    • G06F9/45558—Hypervisor-specific management and integration aspects
    • G06F2009/4557—Distribution of virtual machine instances; Migration and load balancing
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Data Mining & Analysis (AREA)
  • Human Resources & Organizations (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Economics (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Development Economics (AREA)
  • Educational Administration (AREA)
  • Evolutionary Computation (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Strategic Management (AREA)
  • Game Theory and Decision Science (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Probability & Statistics with Applications (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The invention discloses a marginal computing cooperative member discovery method based on comprehensive trust evaluation, which adaptively fuses computing power, storage, bandwidth and behavior difference marginal nodes in a mode of constructing a task driving type cooperative virtual service pool, evaluates the reliability of resources by the trust degree of the nodes, constructs a cooperative service member cluster by the availability, reliability, robustness and resource sharing of the nodes, measures the performance of the constructed cooperative member cluster by taking load balancing capacity, packet loss rate, delay, task completion rate and the like as indexes, and proves that a node cooperative game has Nash balancing steady state, realizes load balancing smoothness, inhibits hot zone effect and improves the performance of member selection and the efficiency of cooperative service by establishing a node cooperative service utility model.

Description

Edge computing collaborative member discovery method based on comprehensive trust evaluation
Technical Field
The invention relates to the technical field of edge computing, in particular to an edge computing collaborative member discovering method based on comprehensive trust evaluation in edge computing.
Background
With the rapid development of the internet and the wide use of mobile intelligent internet of things equipment, emerging network applications and service demands such as internet of things, ultra-high definition videos, real-time video streaming, unmanned driving, industrial internet of things, air-space-ground-sea integrated communication and the like are greatly increased. These emerging applications all have the characteristics of large data volume, high bandwidth and sensitivity to time delay. Due to rapid development and mutual fusion of edge computing and artificial intelligence technologies, the trend of three-transformation and one-fusion of network intellectualization, digitalization, ubiquitous integration and heterogeneous integration in the future is increasingly obvious. IDC reports indicate that more than 500 billion terminals and devices will access the network in 2020 and up to 1000 billion in 2030. The development of the network combining three types and one thing leads the number and the types of the equipment carried by the network, the service and the calculation mode, the protocol cluster and the like to become more and more numerous, and finally the network can be evolved into a stable, open and complex huge chaotic information and data streaming service system. The application of live network, short video and the like enables the network traffic to grow exponentially, the video traffic occupies 73% of the internet traffic, and the video traffic is expected to reach 82% by 2021. At present, the bandwidth, capacity, delay, resources, computing power and the like of the cloud computing center become bottlenecks which restrict the development of the cloud computing center.
Heterogeneous convergence and mass data traffic processing are one of the main challenges faced by the next generation network development, and high bandwidth and low delay are key indexes thereof. The edge computing is one of effective solutions, and the core idea of the method is to push down computing tasks, services and the like from cloud computing and a network core to a novel mode of a network edge, implement computing and analysis processing on a network edge side generated by data, reduce bandwidth consumption and delay, and realize local and near services so as to meet real-time service requirements. The industry and the academic community successively provide new network computing modes, and the new network computing modes can provide computing resources on the intelligent terminal side and can process data nearby according to application requirements. The edge computing promotes the evolution of cloud computing to a layered architecture, and the three-dimensional space, time and distance is realized by fusing resources such as a network edge server and a distributed intelligent terminal to provide real-time reliable high-quality service for users.
With the increase of resources such as computing power and storage capacity of edge intelligent terminals and the revolution of computing modes, network edge resources begin to evolve towards AI computing, servers, micro-server clusters and micro-data centers. The resource requirements of application scenes with different complexity are different, for example, the smart phone applied to social contact, shopping, entertainment and the like has more and idle resources such as computing power, storage, bandwidth and the like, but for the application fields such as ocean monitoring, unmanned driving, video monitoring, industrial control and the like, a large amount of data processing, analysis and calculation need to be completed at a terminal, and the smart phone has the characteristics of isomerism, delay sensitivity, real-time response and the like, and the resources such as computing power, storage, bandwidth and the like of the smart terminal can not meet the bearing functional service requirements. Edge calculation can effectively solve this problem.
In summary, the following problems mainly exist in the prior art: 1) because the edge computing tasks are usually non-uniform and the capability of the edge nodes is limited, in the case of surge, hot area and high load, problems of network congestion, jitter, failure, downtime and the like are generated due to the overload of the tasks of the edge nodes. Through interaction and cooperation of edges, the advantages of nodes at each layer of a cloud, an edge and an end can be fully exerted, resource use is balanced, and comprehensive optimization of resource utilization rate, energy consumption, bandwidth, storage and the like is realized; 2) due to the differences of computing power, storage, bandwidth and behavior of the edge nodes, the QoS of edge computing is directly influenced by the selection of the edge nodes.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide an edge computing collaborative member discovering method based on comprehensive trust evaluation, which adaptively fuses the different edge nodes such as computing power, storage, bandwidth and behavior in a mode of constructing a task driving type collaborative virtual service pool, evaluates the reliability of resources by the trust degree of the nodes, constructs a collaborative service member cluster by the availability, reliability, robustness, resource sharing and the like of the nodes, measures the performance of the constructed collaborative member cluster by taking the load balancing capacity, packet loss rate, delay, task completion rate and the like as indexes, establishes a node bad behavior punishment mechanism, inhibits the cheating behaviors of the nodes such as selfish, rationality and strategy, effectively realizes load balancing smoothness, inhibits the hot zone effect, improves the performance of member selection and the efficiency of collaborative service, and realizes high efficiency, low cost and the like, Trusted resource coordination services.
The technical scheme for realizing the purpose of the invention is as follows:
an edge computing collaborative member discovery method based on comprehensive trust evaluation is characterized in that n nodes m edges are assumed in an edge computing system at a certain time, the formed network topology formalized description is G (V, E), E is an adjacent matrix of a network, and if a node i is connected with a node j, E is an adjacent matrix of the network ij1, otherwise eijThe method comprises the steps that 0, V, represents a set of all server nodes, calculation power and storage resource information of the nodes are automatically broadcasted after the nodes are added into a network, and a main node automatically flushes cache resource related information in an information life cycle, stores the cache resource related information in a local database and continuously updates the cache resource related information; the method specifically comprises the following steps:
1) modeling the service probability among nodes through the node connectivity, analyzing the stability of a network topological structure according to the average service duration of the nodes, and constructing a cooperative connectivity probability model among the nodes; evaluating the direct trust of the nodes according to the inter-node interaction times and the interaction success rate characteristics, utilizing neighbor recommendation trust, adopting improved mean square error to filter cheating nodes, and fusing the direct trust and the indirect trust to construct an aging node comprehensive trust quantification model;
2) evaluating the comprehensive performance of the nodes by using the computational power, storage, bandwidth and comprehensive trust of the nodes, selecting a collaborative service node meeting the requirement according to a collaborative member discovery strategy, and constructing a trusted collaborative service member set;
3) establishing a node income utility model by using the selfish and rational behavior characteristics of the nodes and the trust and available service resources of the nodes, verifying that the found cooperative service set has a Nash equilibrium steady state through an evolutionary repeated game, and establishing a credible cooperative service cluster;
4) designing an improved greedy algorithm to optimize aggregation clustering, constructing a virtual collaborative service pool, collaboratively executing a calculation task, measuring the performance of the constructed collaborative member cluster by taking load balancing capacity, packet loss rate, delay and task completion rate as indexes, and verifying and discovering collaborative service member strategy efficiency;
5) after the cooperative service is completed, evaluating the quality of the cooperative service, and updating trust information of the push cooperative service node in a local network; and updating the characteristic database of the cooperative service node periodically.
The step 1) specifically comprises the following steps:
1-1) the probability p (i) of any node k to reach node i is given by:
Figure BDA0002689107090000031
where ζ (i) is the connectivity of the node i, n (i) is a set of neighbor nodes of the node i, and j is a neighbor node of the node i;
stability probability p of node i in any time period ti onlineThe formula of (1) is:
Figure BDA0002689107090000032
wherein T represents an observation period, Tk,out、tk,inRespectively representing the node down and on-line time in the observation period T;
1-2) constructing a cooperative connection probability P (i) of any node k of the edge network to reach a alliance main node i by formulas (1) and (2):
Figure BDA0002689107090000033
1-3) direct Trust T of node j to node ii：
Figure BDA0002689107090000034
Figure BDA0002689107090000041
s.t.
Figure BDA0002689107090000042
k1＜n,f(0)＝0
Wherein, Ti,j (t)Representing the direct trust level, s, of node j to node i during the period tijRepresenting the number of times that node j successfully services node i during time t, fijRepresenting the number of times node j failed to serve i to the node, f: (i) As a penalty function, k1Representing the number of nodes directly trusting the node i;
1-4) constructing the recommendation trust degree r of the node i according to different recommendation situations and the behavior trust degrees of recommendersi：
Figure BDA0002689107090000043
Wherein r isi,jRepresents the recommendation degree, T, of the node j to the node ii,mRepresenting the direct trust level, T, of node i with the recommended node mm,jRepresenting the direct trust level, k, of node m and the recommended node j2The number of recommenders;
1-5) judging whether the node is a malicious node for collaborative cheating or not by calculating the root mean square of the recommended trust degree, if so, judging whether the node is the malicious node for collaborative cheating
Figure BDA0002689107090000048
If the value is less than the lower bound theta, the node is considered to be a malicious node, the recommendation is received with a small probability p, otherwise, the node is received with the probabilities of 1-p, and r of the recommendation confidence degree formula (5) is obtainediCorrecting the recommended trust level R after correctioniComprises the following steps:
Figure BDA0002689107090000044
s.t.
Figure BDA0002689107090000045
abandon this recommendation
Wherein,
Figure BDA0002689107090000049
expressing root mean square, P (i) the cooperative connection probability of any node k to reach the ally owner node i;
1-6) the dynamic trust value of node i is expressed as follows:
Figure BDA0002689107090000046
ωlfor an observation period TlThe corresponding weight is:
Figure BDA0002689107090000047
where μ is the time attenuation coefficient, t0Is the initial time.
The step 2) specifically comprises the following steps:
2-1) assuming mutual independence between the attributes, each node feature vector is represented as Xi＝(ci,si,Trustnew(i) ) overall performance of the node
Figure BDA0002689107090000058
The expression is as follows:
Figure BDA0002689107090000059
wherein, ci *、si *Trust (i) respectively represents the computing power and the shared storage after the normalization processing, and lambda is a weight factor;
2-2) comprehensive evaluation of current node
Figure BDA00026891070900000510
If the number of the nodes is larger than or equal to the threshold psi, the nodes are added into the trusted cooperative service set; when-node comprehensive evaluation
Figure BDA00026891070900000511
Below the threshold psi, the trusted collaborative service set is not involved in the trusted collaborative service set construction, but still remains in the network, maintaining the dynamic stability of the network.
The step 3) specifically comprises the following steps:
3-1) in any game, i represents the node itself, i represents other adjacent game nodes without the node i, and the income obtained at the end of one stage is BsThe resource consumed at this stage is CsThus, the node is in each time slot tnUtility function of
Figure BDA0002689107090000051
The expression is as follows:
Figure BDA0002689107090000052
then the utility function U of the node during the observation period TiThe expression is as follows:
Figure BDA0002689107090000053
wherein, mu1Discount rate for node revenue at each time period;
3-2) assuming that the node starts the game by a cooperation strategy, and adopting a 'fashion round trip' strategy in the subsequent game stage, wherein the node i simulates the behavior of the opponent-i in the previous stage; setting the 1 st stage, maintaining the probability of all node cooperation as 1, and thus maintaining the cooperation all the time; stage 2, the node i changes its own strategy due to its attribute characteristics and stability, i.e. the cooperation probability changes
Figure BDA0002689107090000054
Then its cooperative probability of opponent-i in stage 3 is changed to
Figure BDA0002689107090000055
Node i mimics in stage 3 the cooperative probability of an adversary in stage 2, i.e.
Figure BDA0002689107090000056
Through repeated game, the cooperative probability sequence of the finally generated nodes is as follows:
Figure BDA0002689107090000057
wherein the node is in time slot tnThe cooperation probability rho of the strategy of the node is changed is related to the attribute and stability of the node, the higher the comprehensive evaluation is, the higher the changed cooperation probability is, and the expression is as follows;
Figure BDA0002689107090000061
3-3) substituting the expression (12) into the expression (11) to obtain the final benefit expression of the node i as follows:
Figure BDA0002689107090000062
only UiThe strategy is adopted only when the node is more than or equal to 0, so that the strategy is adopted when the node is not less than
Figure BDA0002689107090000063
And the node takes the strategy to reach a Nash equilibrium steady state, and a trusted cooperative service cluster is constructed.
The step 4) specifically comprises the following steps:
4-1) constructing priority function g of member node jjThe expression is as follows:
Figure BDA0002689107090000064
wherein,
Figure BDA0002689107090000065
rtt (i, j) is the network delay from the alliance owner node i to the alliance member node j;
4-2) constructing a load balancing function f (i) of the alliance main node according to the load and the average connection probability of the alliance main node, wherein the expression is as follows:
Figure BDA0002689107090000066
wherein Q isiIs the load threshold of the alliance main node, xi receives the task request concurrency control coefficient of the node i, delta QiDynamically adjusting the increments for load adaptation;
4-3) constructing an adaptive function eta (i) expression for evaluating a cluster formed by an alliance owner node i according to the alliance owner node load balance and the alliance node priority, wherein the adaptive function eta (i) expression is as follows:
Figure BDA0002689107090000067
solving through an improved greedy algorithm to optimize aggregation clustering;
4-4) clustering completed resource pool X ═ S1,S2,…,Sk|j∈SiJ ═ 1,2,.. multidot.n }, in order to guarantee the cooperative service quality, dynamically aggregating the optimal resources according to the performance of the resource pool, and constructing an expression of an evaluation function η (X) of the performance of the resource pool as follows:
Figure BDA0002689107090000071
wherein S isiAnd k is the number of constructed clusters for the service cluster formed by the alliance master node i.
In step 4-3), the improved greedy algorithm is described as follows:
4-3-1) setting the number of expected clusters as k, and taking the first k nodes with optimal comprehensive performance evaluation as initial aggregation centers;
4-3-2) for each cluster center, calculating the priority and the load increment of the adjacent nodes according to a formula (15) and a formula (16), and calculating f (i) according to the formula (16), wherein if f (i) is more than 0, greedy acquisition is carried out to belong to the same collaborative service cluster; if no more nodes are added, selecting the node with the highest priority in the cooperative service cluster as a new aggregation center to repeat 4-3-2), otherwise, stopping adding;
4-3-3) if the system still has nodes which do not join the cooperative service cluster, thenAccording to greedy principle, selecting
Figure BDA0002689107090000072
The optimal node becomes a new aggregation center, and 4-3-2) is repeated until all the nodes are added into the cooperative service cluster;
4-3-4) noting the current solution X ═ S1,S2,...,SkThe value eta (X) is adapted according to the formula (18);
4-3-5) making t equal to 0, t < tmaxAnd executing:
randomly selecting a node in the system, exiting all adjacent nodes of the node from the current cooperative service cluster, and repeating 4-3-2) to form a new cooperative service cluster; recalculating current solution X' ═ S1′,S2′,...,Sk'}, calculating a system adaptation value eta (X');
if η (X ') > η (X), X' is recorded as { S ═ S1′,S2′,...,Sk' } is the optimal solution; otherwise t + +;
4-3-6) constructing an edge computing cooperative service cluster according to the optimal solution, mapping cooperative service resources and initializing a cooperative service pool.
The invention provides an edge computing cooperative member discovering method based on comprehensive trust evaluation, which adaptively fuses different edge nodes such as computing power, storage, bandwidth and behavior in a mode of constructing a task driving type cooperative virtual service pool, evaluates the reliability of resources by the trust degree of the nodes, constructs a cooperative service member cluster by the availability, reliability, robustness, resource sharing and the like of the nodes, measures the performance of the constructed cooperative member cluster by taking the load balancing capacity, the packet loss rate, the delay, the task completion rate and the like as indexes, and proves that a node cooperative game has a Nash balancing steady state, realizes load balancing smoothness, inhibits a hotspot effect and improves the performance of member selection and the efficiency of cooperative service by establishing a node cooperative service utility model.
Drawings
FIG. 1 is a trusted alliance discovery policy framework diagram;
FIG. 2 is a diagram of a recommended trust scenario for different scenarios;
FIG. 3 is a diagram of a collaborative services topology;
FIG. 4 is a graph of the average load rate of the base station and the alliance cooperative cluster;
FIG. 5 is a collaborative services system flow diagram;
FIG. 6 is a concurrency graph of a collaborative services system;
FIG. 7 is a comparison graph of the base station and member average confidence levels;
fig. 8 is a diagram of task receiving rate, cooperative service success rate, packet loss rate, and cooperative service response delay;
FIG. 9 is a graph of base station loading rates;
FIG. 10 is a collaborative services pool flow diagram;
FIG. 11 is a collaboration services pool concurrency graph;
FIG. 12 is a collaborative services affiliate reputation graph;
fig. 13 is a graph illustrating cooperative task receiving rate, cooperative service success rate, cooperative service packet loss rate, and cooperative service response delay.
Detailed Description
The invention will be further elucidated with reference to the drawings and examples, without however being limited thereto.
An edge computing collaborative member discovery method based on comprehensive trust evaluation comprises the following steps:
assuming that m edges exist in n nodes in the edge computing system at a certain time, the formed network topology is formally described as G (V, E). E is the adjacency matrix of the network, if node i is connected with node j, then E ij1, otherwise e ij0. V ═ 1,2, 3.., n } represents the set of all server nodes. Assuming that the node autonomously broadcasts resource information such as computing power, storage and the like after joining the network, the main node automatically flushes and caches the resource related information in the information life cycle, stores the resource related information in a local database and continuously updates the resource related information.
1) Modeling the service probability among nodes through the node connectivity, analyzing the stability of a network topological structure according to the average service duration of the nodes, and constructing a cooperative connectivity probability model among the nodes; evaluating the direct trust of the nodes according to the characteristics of the number of times of interaction between the nodes, the success rate of interaction and the like, utilizing neighbor recommendation trust, adopting improved mean square error to filter cheating nodes, and fusing the direct trust and the indirect trust to construct an aging node comprehensive trust quantification model;
2) evaluating the comprehensive performance of the nodes by using the computational power, storage, bandwidth, comprehensive trust degree and the like of the nodes, selecting a collaborative service node meeting the requirement according to a collaborative member discovery strategy, and constructing a trusted collaborative service member set;
3) establishing a node income utility model by using the selfish and rational behavior characteristics of the nodes and the trust and available service resources of the nodes, verifying that the found cooperative service set has a Nash equilibrium steady state through an evolutionary repeated game, and establishing a credible cooperative service cluster;
4) designing an improved greedy algorithm to optimize aggregation clustering, constructing a virtual collaborative service pool, collaboratively executing a calculation task, measuring the performance of the constructed collaborative member cluster by taking load balancing capacity, packet loss rate, delay, task completion rate and the like as indexes, and verifying and discovering collaborative service member strategy efficiency;
5) after the cooperative service is completed, evaluating the quality of the cooperative service, and updating trust information of the push cooperative service node in a local network; and updating the characteristic database of the cooperative service node periodically.
The step 1) specifically comprises the following steps:
1-1) the probability p (i) of any node k to reach node i is given by:
Figure BDA0002689107090000091
where ζ (i) is the connectivity of the node i, n (i) is a set of neighbor nodes of the node i, and j is a neighbor node of the node i;
stability probability p of node i in any time period tionlineThe formula of (1) is:
Figure BDA0002689107090000092
wherein T represents an observationPeriod, tk,out、tk,inRespectively representing the down and up time of the node in the observation period T.
1-2) the formula (1) and (2) can construct the cooperative connection probability P (i) of any node k of the edge network to reach the alliance main node i, wherein the expression is as follows:
Figure BDA0002689107090000093
1-3) direct Trust T of node j to node iiThe expression is as follows:
Figure BDA0002689107090000094
Figure BDA0002689107090000095
s.t.
Figure BDA0002689107090000096
k1＜n,f(0)＝0
wherein, Ti,j (t)Representing the direct trust level, s, of node j to node i during the period tijRepresenting the number of times that node j successfully services node i during time t, fijRepresenting the number of times node j failed to serve i to the node, f: (i) As a penalty function, k1Indicating the number of nodes having direct trust with node i.
1-4) constructing the recommendation trust degree r of the node i according to different recommendation situations and recommender behavior credibility shown in FIG. 2iThe expression is as follows:
Figure BDA0002689107090000101
wherein r isi,jRepresents the recommendation degree, T, of the node j to the node ii,mRepresenting the direct trust level, T, of node i with the recommended node mm,jRepresenting nodes m and recommendationsDirect trust of node j, k2Is the number of recommenders.
1-5) judging whether the node is a malicious node for collaborative cheating or not by calculating the root mean square of the recommended trust degree, if so, judging whether the node is the malicious node for collaborative cheating
Figure BDA0002689107090000106
If the value is less than the lower bound theta, the node is considered to be a malicious node, the recommendation is received with a small probability p, otherwise, the node is received with the probabilities of 1-p, and r of the recommendation confidence degree formula (5) is obtainediCorrecting the recommended trust level R after correctioniComprises the following steps:
Figure BDA0002689107090000102
s.t.
Figure BDA0002689107090000103
abandon this recommendation
Wherein,
Figure BDA0002689107090000107
represents the root mean square, P (i) the connection probability that any node k can cooperate to reach the ally owner node i.
1-6) the dynamic trust value of node i is expressed as follows:
Figure BDA0002689107090000104
ωlfor an observation period TlThe corresponding weight expression is:
Figure BDA0002689107090000105
where μ is the time attenuation coefficient, t0Is the initial time;
the step 2) specifically comprises the following steps:
2-1) each node feature vector table is assumed to be independent of each otherShown as Xi＝(ci,si,Trustnew(i) ) overall performance of the node
Figure BDA0002689107090000108
The expression is as follows:
Figure BDA0002689107090000116
wherein, ci *，si *Trust (i) represents the computing power and shared storage after normalization, and λ is a weight factor.
2-2) comprehensive evaluation of current node
Figure BDA0002689107090000117
If the number of the nodes is larger than or equal to the threshold psi, the nodes are added into the trusted cooperative service set; when-node comprehensive evaluation
Figure BDA0002689107090000118
Below the threshold psi, the trusted collaborative service set is not involved in the trusted collaborative service set construction, but still remains in the network, maintaining the dynamic stability of the network.
The step 3) specifically comprises the following steps:
3-1) assume that in any game, i represents the node itself and i refers to other neighboring game nodes that do not contain node i. The gain obtained at the end of a phase is BsThe resource consumed at this stage is CsThus, the node is in each time slot tnUtility function of
Figure BDA0002689107090000111
The expression is as follows:
Figure BDA0002689107090000112
then the utility function U of the node during the observation period TiThe expression is as follows:
Figure BDA0002689107090000113
wherein, mu1A discount rate for the node benefit at each time period.
3-2) assuming that the nodes start the game with the cooperation strategy, in the later game stage, adopting a 'fashion round trip' strategy, and simulating the behaviors of the opponent i in the previous stage by the node i. And in the stage 1, the probability of all node cooperation is maintained to be 1, so that the cooperation is maintained all the time. Stage 2, the node i changes its own strategy due to its attribute characteristics and stability, i.e. the cooperation probability changes
Figure BDA0002689107090000119
Then its cooperative probability of opponent-i in stage 3 is changed to
Figure BDA00026891070900001110
Node i mimics in stage 3 the cooperative probability of an adversary in stage 2, i.e.
Figure BDA00026891070900001111
Through repeated game, the expression of the cooperative probability sequence of the finally generated node is as follows:
Figure BDA0002689107090000114
wherein the node is in time slot tnThe cooperation probability rho of the strategy of the node is changed to be related to the attribute and stability of the node, the higher the comprehensive evaluation is, the larger the changed cooperation probability is, and the cooperation probability rho expression is as follows:
Figure BDA0002689107090000115
3-3) substituting equation (12) into equation (11) to obtain the final profit expression of the node i:
Figure BDA0002689107090000121
only UiThe strategy is adopted only when the node is more than or equal to 0. Therefore, when
Figure BDA0002689107090000122
And the node takes the strategy to reach a Nash equilibrium steady state, and a trusted cooperative service cluster is constructed.
The step 4) specifically comprises the following steps:
4-1) constructing priority function g of member node jjThe expression is as follows:
Figure BDA0002689107090000123
wherein,
Figure BDA0002689107090000127
rtt (i, j) is the network delay from the alliance master node i to the alliance node j, which is the overall capability of the node j.
4-2) constructing a load balancing function f (i) of the alliance main node according to the load and the average connection probability of the alliance main node, wherein the expression is as follows:
Figure BDA0002689107090000124
wherein Q isiIs the load threshold of the alliance main node, xi receives the task request concurrency control coefficient of the node i, delta QiThe increments are dynamically adjusted for load adaptation.
4-3) constructing an adaptive function eta (i) expression for evaluating a cluster formed by an alliance owner node i according to the alliance owner node load balance and the alliance node priority, wherein the adaptive function eta (i) expression is as follows:
Figure BDA0002689107090000125
and optimizing the aggregation clustering by solving through an improved greedy algorithm.
4-4) Cluster formationFormed resource pool X ═ S1,S2,…,Sk|j∈SiJ ═ 1,2,.. multidot.n }, in order to guarantee the cooperative service quality, the optimal resource is dynamically aggregated according to the performance of the resource pool.
Figure BDA0002689107090000126
Wherein S isiAnd k is the number of constructed clusters for the service cluster formed by the alliance master node i.
The specific steps of the improved greedy algorithm in the step 4-3) are described as follows:
4-3-1) setting the number of expected clusters as k, and taking the first k nodes with optimal comprehensive performance evaluation as initial aggregation centers;
4-3-2) for each cluster center, calculating the priority and the load increment of the adjacent nodes according to a formula (15) and a formula (16), and calculating f (i) according to the formula (16), wherein if f (i) is more than 0, greedy acquisition is carried out to belong to the same collaborative service cluster; if no more nodes are added, selecting the node with the highest priority in the cooperative service cluster as a new aggregation center to repeat 4-3-2), otherwise, stopping adding;
4-3-3) if the system still has nodes which are not added into the cooperative service cluster, selecting the nodes according to the greedy principle
Figure BDA0002689107090000131
The optimal node becomes a new aggregation center, and 4-3-2) is repeated until all the nodes are added into the cooperative service cluster;
4-3-4) noting the current solution X ═ S1,S2,...,SkCalculating a system adaptation value eta (X) according to a formula (18);
4-3-5) making t equal to 0, t < tmaxAnd executing:
randomly selecting a node in the system, exiting all adjacent nodes of the node from the current cooperative service cluster, and repeating 4-3-2) to form a new cooperative service cluster; recalculating current solution X' ═ S1′,S2′,...,Sk'} calculating the system adaptation value eta (X')；
If η (X ') > η (X), X' is recorded as { S ═ S1′,S2′,...,Sk' } is the optimal solution; otherwise t + +;
4-3-6) constructing an edge computing cooperative service cluster according to the optimal solution, mapping cooperative service resources and initializing a cooperative service pool.
Example (b):
in order to verify and analyze the performance of the algorithm model, an offshore area edge computing simulation system is designed and realized, and cooperative service, computing migration, task cooperation, interaction behavior, credit evaluation and the like in edge computing are simulated. And constructing a collaborative service system based on the Route Views public data set, and simulating the uploading and downloading of the offshore sea area edge computing Web browser service to simulate the Web browsing application scene through the simulation experiment. The edge server autonomously integrates and constructs a collaborative service pool based on trust evaluation, a load balancing trigger is set in a task driving mode, when the load of the edge server reaches a threshold value, collaborative operation is triggered, and part of tasks are migrated to the collaborative service pool to be executed. The trusted collaborative member discovery strategy is shown in figure 1.
Experiment one: SCECT model collaborative fusion performance simulation analysis
The data center server cluster is provided with 200 computing tasks to be migrated, the quantity of each migration computing task is more than or equal to 256 and less than or equal to 1024, and the computing complexity is O (n)2) Each time, 5 continuous calculation tasks are requested, the experimental test time is 4h, the flow sampling period is 30s, and the sampling periods of other indexes are 5 s.
1) And comparing and analyzing the SCECT with k-means and KNN by taking the similarity coupling degree, the credit degree, the service capability and the like as evaluation indexes. The topology structure of the collaborative service is shown in fig. 3, and the specific performance characteristic table of the service system constructed by different algorithms is shown in table 1. The polymerization results show that:
a) compared with the KNN, k-means algorithm, the SCECT algorithm has the advantages that the base station cluster member dispersion is respectively reduced by 34.39% and 5.54%, the member dispersion of the member cluster is respectively reduced by 34.93% and 39.31%, the system similarity coupling is improved by 64.40% and 51.56%, and the average aggregation capability of the system is improved by 34.5% and 15.67%;
b) more nodes are aggregated to form a cooperative service cluster by the nodes with better performance, and the Martian effect is achieved.
2) Starting from the 5 th min of the experiment, when a surge service request occurs every 30min, the load rate, the flow chart, the concurrency chart, the comparison chart of the base station cluster and the alliance owner cooperative service cluster average trust degree curve and the comparison chart of the base station and the alliance member average trust degree curve are respectively shown in fig. 4-7, and the task receiving rate, the cooperative service success rate, the packet loss rate and the cooperative service response delay are respectively shown in fig. 8. The experimental results show that:
a) compared with KNN, K-means, the load of the base station cluster is reduced by 14.1 percent by the SCECT, the load balance rate of the alliance collaborative service cluster is improved by 58.65 percent, and the balance capability of the constructed collaborative service system is improved by 36.38 percent;
b) compared with KNN, k-means, the service flow of the base station cluster is improved by 25.53% and 44.06%, the service flow of the allied cooperative cluster is improved by 19.80% and 28.48%, the overall service capacity of the system is improved by 21.23% and 31.26%, and the concurrent performance is improved by 22.00% and 32.61%;
c) the SCECT trust mechanism stimulates cooperative service alliers to participate in cooperative enthusiasm, and effectively selects a credible cooperative service node to cooperatively complete a task;
d) the cooperative service cluster constructed by the SCECT is superior to KNN, k-means in response delay, packet loss rate, cooperative success rate and task receiving rate; the overall service efficiency of the SCECT system is improved by 36.14% and 33.26% compared with KNN and k-means;
3) the comprehensive performance of the collaborative service cluster constructed by the SCECT is superior to that of the KNN and k-means algorithm, and the trust mechanism constructed by the SCECT effectively selects the credible collaborative service nodes to efficiently complete the collaborative service, so that the service quality is improved, the load balance is realized, and the guarantee is provided for the large-scale credible edge computing service.
Experiment two: SCECT model collaborative service pool performance analysis
The cloud server is provided with 100 computing tasks to be migrated, the number of M of each migration computing task is more than or equal to 500 and less than or equal to 1024, surge type service request testing is carried out on the cloud server where the target computing task is located, a user requests 5 continuous stream data files each time, the link re-searching frequency is not more than 3, and the experimental testing time is 4 h.
4) And comparing and analyzing the performance of the cooperative service pool with a random nomadic (SR) and an on-demand cooperative routing (AODV) by taking the flow, the load, the credit, the task receiving rate, the cooperative success rate and the like as evaluation indexes. Base station load rate, cooperative service pool flow, cooperative service pool concurrency number, and cooperative service member credit are shown in fig. 9 to 12, respectively, and cooperative task receiving rate, cooperative service success rate, cooperative service packet loss rate, and cooperative service response delay are shown in fig. 13. The experimental results show that.
a) Compared with AODV, SR, the SCECT improves the load unloading rate of the base station by 30.8%; the SCECT migrates part of tasks to the cooperative service pool for execution, so that local load balance is realized, and the avalanche effect is effectively avoided;
b) compared with AODV and SR, SCECT improves the flow of the cooperative service pool by 2.4 times and 2.5 times respectively, and improves the concurrency by 1.3 times and 1.5 times; the maximum service capacity of the system is improved by 185% and 200%;
c) the SCECT selects a trusted cooperative service member to participate in cooperation through a trust mechanism, so that the cooperative service efficiency can be effectively improved;
d) the SCECT constructed cooperative service cluster is superior to AODV and SR in response delay, packet loss rate, cooperative success rate and task receiving rate;
e) the SCECT system overall service efficiency is improved 64.41% and 40.62% compared with AODV and SR.
5) The SCECT selects more effective allied members to participate in cooperation, and compared with AODV and SR system, the balance capacity is improved by 185% and 200%, and the service efficiency is improved by 64.41% and 40.62%; the trust mechanism constructed by the SCECT effectively selects the trusted cooperative service node to efficiently complete the cooperative service, improves the service quality, realizes load balance and provides guarantee for large-scale trusted edge computing service.
6) The SCECT algorithm can effectively improve the performance of selecting allied members, improve the efficiency of cooperative service and realize efficient and credible resource cooperation.
To summarize:
due to the limited load and capacity of edge devices, close cooperation is required when processing a large amount of edge computing data. However, the edge devices have the characteristics of inconsistent performance, dynamic flexibility, autonomy and the like, and the cooperative service fails due to the behaviors of selfish, rationality and the like when the cooperative service is executed. An edge computing collaborative member discovery method based on comprehensive trust evaluation is provided. The method evaluates the reliability of resources according to the trust degree of the nodes, constructs the collaborative service member cluster according to the availability, reliability, robustness, resource sharing and the like of the nodes, and measures the performance of the constructed collaborative member cluster by taking the load balancing capability, the packet loss rate, the delay, the task completion rate and the like as indexes. And a node bad behavior punishment mechanism is established, and cheating behaviors of nodes such as selfish, rationality and strategy are inhibited. The node cooperative game is proved to have Nash equilibrium steady state by establishing a node cooperative service utility model. In order to verify and analyze the performance of the algorithm model, an offshore sea area edge computing simulation system is designed based on a Route Views public data set, the system simulates cooperative service, computing migration, task cooperation, interaction behavior, credit evaluation and the like in edge computing in a task-driven mode, and an SCMECT model is optimized according to a simulation experiment result, so that the SCECT model is closer to practical application. Compared with simulation experiment results, the SCECT can effectively realize load balancing smoothness, inhibit hot zone effect and improve the performance of member selection and the efficiency of cooperative service.
TABLE 1 System Performance characteristics Table
Figure BDA0002689107090000161

Claims (6)

1. A method for discovering edge computing cooperative members based on comprehensive trust evaluation is characterized in that n nodes m edges in an edge computing system at a certain moment are assumed, the formed network topology formalized description is G (V, E), E is an adjacent matrix of a network, and if a node i is connected with a node j, E is a node jij1, otherwise eij0, V ═ 1,2, 3.., n } denotes the set of all server nodes, assuming that the nodes autonomously broadcast their computing power and storage resource information after joining the network,the main node stores the relevant information of the cache resource in a local database by automatically flushing the cache resource in the information life cycle and continuously updates the cache resource; the method specifically comprises the following steps:
1) modeling the service probability among nodes through the node connectivity, analyzing the stability of a network topological structure according to the average service duration of the nodes, and constructing a cooperative connectivity probability model among the nodes; evaluating the direct trust of the nodes according to the inter-node interaction times and the interaction success rate characteristics, utilizing neighbor recommendation trust, adopting improved mean square error to filter cheating nodes, and fusing the direct trust and the indirect trust to construct an aging node comprehensive trust quantification model;
2) evaluating the comprehensive performance of the nodes by using the computational power, storage, bandwidth and comprehensive trust of the nodes, selecting a collaborative service node meeting the requirement according to a collaborative member discovery strategy, and constructing a trusted collaborative service member set;
3) establishing a node income utility model by using the selfish and rational behavior characteristics of the nodes and the trust and available service resources of the nodes, verifying that the found cooperative service set has a Nash equilibrium steady state through an evolutionary repeated game, and establishing a credible cooperative service cluster;
4) designing an improved greedy algorithm to optimize aggregation clustering, constructing a virtual collaborative service pool, collaboratively executing a calculation task, measuring the performance of the constructed collaborative member cluster by taking load balancing capacity, packet loss rate, delay and task completion rate as indexes, and verifying and discovering collaborative service member strategy efficiency;
5) after the cooperative service is completed, evaluating the quality of the cooperative service, and updating trust information of the push cooperative service node in a local network; and updating the characteristic database of the cooperative service node periodically.
2. The edge computing collaborative member discovery method based on comprehensive trust evaluation according to claim 1, wherein the step 1) specifically comprises the following steps:
1-1) the probability p (i) of any node k to reach node i is given by:
Figure FDA0002689107080000011
where ζ (i) is the connectivity of the node i, n (i) is a set of neighbor nodes of the node i, and j is a neighbor node of the node i;
stability probability p of node i in any time period ti onlineThe formula of (1) is:
Figure FDA0002689107080000021
wherein T represents an observation period, Tk,out、tk,inRespectively representing the node down and on-line time in the observation period T;
1-2) constructing a cooperative connection probability P (i) of any node k of the edge network to reach a alliance main node i by formulas (1) and (2):
Figure FDA0002689107080000022
1-3) direct Trust T of node j to node ii：
Figure FDA0002689107080000023
Figure FDA0002689107080000024
s.t.
Figure FDA0002689107080000025
k1＜n,f(0)＝0
Wherein, Ti,j (t)Representing the direct trust level, s, of node j to node i during the period tijRepresenting the number of times that node j successfully services node i during time t, fijRepresenting the number of times node j failed to serve i to the node, f: (i) As a penalty function, k1Representing the number of nodes directly trusting the node i;
1-4) constructing the recommendation trust degree r of the node i according to different recommendation situations and the behavior trust degrees of recommendersi：
Figure FDA0002689107080000026
Wherein r isi,jRepresents the recommendation degree, T, of the node j to the node ii,mRepresenting the direct trust level, T, of node i with the recommended node mm,jRepresenting the direct trust level, k, of node m and the recommended node j2The number of recommenders;
1-5) judging whether the node is a malicious node for collaborative cheating or not by calculating the root mean square of the recommended trust degree, if so, judging whether the node is the malicious node for collaborative cheating
Figure FDA0002689107080000027
If the value is less than the lower bound theta, the node is considered to be a malicious node, the recommendation is received with a small probability p, otherwise, the node is received with the probabilities of 1-p, and r of the recommendation confidence degree formula (5) is obtainediCorrecting the recommended trust level R after correctioniComprises the following steps:
Figure FDA0002689107080000031
s.t.
Figure FDA00026891070800000311
abandon this recommendation
Wherein,
Figure FDA0002689107080000033
expressing root mean square, P (i) the cooperative connection probability of any node k to reach the ally owner node i;
1-6) the dynamic trust value of node i is expressed as follows:
Figure FDA0002689107080000034
ωlfor an observation period TlThe corresponding weight is:
Figure FDA0002689107080000035
where μ is the time attenuation coefficient, t0Is the initial time.
3. The edge computing collaborative member discovery method based on comprehensive trust evaluation according to claim 1, wherein the step 2) specifically comprises the following steps:
2-1) assuming mutual independence between the attributes, each node feature vector is represented as Xi＝(ci,si,Trustnew(i) ) overall performance of the node
Figure FDA0002689107080000036
The expression is as follows:
Figure FDA0002689107080000037
wherein, ci *、si *Trust (i) respectively represents the computing power and the shared storage after the normalization processing, and lambda is a weight factor;
2-2) comprehensive evaluation of current node
Figure FDA0002689107080000039
If the number of the nodes is larger than or equal to the threshold psi, the nodes are added into the trusted cooperative service set; when-node comprehensive evaluation
Figure FDA00026891070800000310
Below the threshold psi, it does not participate in the trusted collaborative service set construction, but remains in the network, keeping the networkDynamic stability of (3).
4. The edge computing collaborative member discovery method based on comprehensive trust evaluation according to claim 1, wherein the step 3) specifically comprises the following steps:
3-1) in any game, i represents the node itself, i represents other adjacent game nodes without the node i, and the income obtained at the end of one stage is BsThe resource consumed at this stage is CsThus, the node is in each time slot tnUtility function of
Figure FDA0002689107080000041
The expression is as follows:
Figure FDA0002689107080000042
then the utility function U of the node during the observation period TiThe expression is as follows:
Figure FDA0002689107080000043
wherein, mu1Discount rate for node revenue at each time period;
3-2) assuming that the node starts the game by a cooperation strategy, and adopting a 'fashion round trip' strategy in the subsequent game stage, wherein the node i simulates the behavior of the opponent-i in the previous stage; setting the 1 st stage, maintaining the probability of all node cooperation as 1, and thus maintaining the cooperation all the time; stage 2, the node i changes its own strategy due to its attribute characteristics and stability, i.e. the cooperation probability changes
Figure FDA0002689107080000044
Then its cooperative probability of opponent-i in stage 3 is changed to
Figure FDA0002689107080000045
Node i mimics in stage 3 the cooperative probability of an adversary in stage 2, i.e.
Figure FDA0002689107080000046
Through repeated game, the cooperative probability sequence of the finally generated nodes is as follows:
Figure FDA0002689107080000047
wherein the node is in time slot tnThe cooperation probability rho of the strategy of the node is changed is related to the attribute and stability of the node, the higher the comprehensive evaluation is, the higher the changed cooperation probability is, and the expression is as follows;
Figure FDA0002689107080000048
3-3) substituting the expression (12) into the expression (11) to obtain the final benefit expression of the node i as follows:
Figure FDA0002689107080000049
only UiThe strategy is adopted only when the node is more than or equal to 0, so that the strategy is adopted when the node is not less than
Figure FDA00026891070800000410
And the node takes the strategy to reach a Nash equilibrium steady state, and a trusted cooperative service cluster is constructed.
5. The edge computing collaborative member discovery method based on comprehensive trust evaluation according to claim 1, wherein the step 4) specifically comprises the following steps:
4-1) constructing priority function g of member node jjThe expression is as follows:
Figure FDA0002689107080000051
wherein,
Figure FDA0002689107080000052
rtt (i, j) is the network delay from the alliance owner node i to the alliance member node j;
4-2) constructing a load balancing function f (i) of the alliance main node according to the load and the average connection probability of the alliance main node, wherein the expression is as follows:
Figure FDA0002689107080000053
wherein Q isiIs the load threshold of the alliance main node, xi receives the task request concurrency control coefficient of the node i, delta QiDynamically adjusting the increments for load adaptation;
4-3) constructing an adaptive function eta (i) expression for evaluating a cluster formed by an alliance owner node i according to the alliance owner node load balance and the alliance node priority, wherein the adaptive function eta (i) expression is as follows:
Figure FDA0002689107080000054
solving through an improved greedy algorithm to optimize aggregation clustering;
4-4) clustering completed resource pool X ═ S1,S2,…,Sk|j∈SiJ ═ 1,2,.. multidot.n }, in order to guarantee the cooperative service quality, dynamically aggregating the optimal resources according to the performance of the resource pool, and constructing an expression of an evaluation function η (X) of the performance of the resource pool as follows:
Figure FDA0002689107080000055
wherein S isiAnd k is the number of constructed clusters for the service cluster formed by the alliance master node i.
6. The edge computing collaborative member discovery method based on comprehensive trust evaluation according to claim 5, wherein in the step 4-3), the improved greedy algorithm is described as follows:
4-3-1) setting the number of expected clusters as k, and taking the first k nodes with optimal comprehensive performance evaluation as initial aggregation centers;
4-3-2) for each cluster center, calculating the priority and the load increment of the adjacent nodes according to a formula (15) and a formula (16), and calculating f (i) according to the formula (16), wherein if f (i) is more than 0, greedy acquisition is carried out to belong to the same collaborative service cluster; if no more nodes are added, selecting the node with the highest priority in the cooperative service cluster as a new aggregation center to repeat 4-3-2), otherwise, stopping adding;
4-3-3) if the system still has nodes which are not added into the cooperative service cluster, selecting the nodes according to the greedy principle
Figure FDA0002689107080000056
The optimal node becomes a new aggregation center, and 4-3-2) is repeated until all the nodes are added into the cooperative service cluster;
4-3-4) noting the current solution X ═ S1,S2,...,SkThe value eta (X) is adapted according to the formula (18);
4-3-5) making t equal to 0, t < tmaxAnd executing:
randomly selecting a node in the system, exiting all adjacent nodes of the node from the current cooperative service cluster, and repeating 4-3-2) to form a new cooperative service cluster; recalculating current solution X' ═ S1′,S2′,...,Sk'}, calculating a system adaptation value eta (X');
if η (X ') > η (X), X' is recorded as { S ═ S1′,S2′,...,Sk' } is the optimal solution; otherwise t + +;
4-3-6) constructing an edge computing cooperative service cluster according to the optimal solution, mapping cooperative service resources and initializing a cooperative service pool.
CN202010985519.5A 2020-09-18 2020-09-18 Edge computing collaboration alliance discovery method based on comprehensive trust evaluation Active CN112132202B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010985519.5A CN112132202B (en) 2020-09-18 2020-09-18 Edge computing collaboration alliance discovery method based on comprehensive trust evaluation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010985519.5A CN112132202B (en) 2020-09-18 2020-09-18 Edge computing collaboration alliance discovery method based on comprehensive trust evaluation

Publications (2)

Publication Number Publication Date
CN112132202A true CN112132202A (en) 2020-12-25
CN112132202B CN112132202B (en) 2023-11-17

Family

ID=73842938

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010985519.5A Active CN112132202B (en) 2020-09-18 2020-09-18 Edge computing collaboration alliance discovery method based on comprehensive trust evaluation

Country Status (1)

Country Link
CN (1) CN112132202B (en)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112732442A (en) * 2021-01-11 2021-04-30 重庆大学 Distributed model for edge computing load balancing and solving method thereof
CN113225385A (en) * 2021-04-16 2021-08-06 天津大学 Micro-service request distribution method facing edge computing environment and based on game theory
CN113422803A (en) * 2021-05-19 2021-09-21 武汉工程大学 Seamless migration method for intelligent unmanned aerial vehicle inspection task based on end edge cloud cooperation
CN114172849A (en) * 2021-11-26 2022-03-11 华中科技大学 Deterministic traffic shaping method based on game theory
CN114301987A (en) * 2022-03-07 2022-04-08 天津市城市规划设计研究总院有限公司 Dynamic scheduling method and system for virtualized network resources
CN114326822A (en) * 2022-03-09 2022-04-12 中国人民解放军66136部队 Unmanned aerial vehicle cluster information sharing method based on evolutionary game
CN114401134A (en) * 2022-01-14 2022-04-26 重庆邮电大学 Edge-side-cooperative Internet of things distributed trusted management method
CN116932228A (en) * 2023-09-14 2023-10-24 湖南希赛网络科技有限公司 Edge AI task scheduling and resource management system based on volunteer calculation

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090328148A1 (en) * 2008-06-30 2009-12-31 Industry-Academic Cooperation Foundation Of Kyung Hee University Method of trust management in wireless sensor networks
US20130166712A1 (en) * 2011-12-27 2013-06-27 Microsoft Corporation Cloud-edge topologies
CN110740194A (en) * 2019-11-18 2020-01-31 南京航空航天大学 Micro-service combination method based on cloud edge fusion and application
CN110851429A (en) * 2019-11-21 2020-02-28 嘉兴学院 Edge computing credible cooperative service method based on influence self-adaptive aggregation
CN111031102A (en) * 2019-11-25 2020-04-17 哈尔滨工业大学 Multi-user, multi-task mobile edge computing system cacheable task migration method

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090328148A1 (en) * 2008-06-30 2009-12-31 Industry-Academic Cooperation Foundation Of Kyung Hee University Method of trust management in wireless sensor networks
US20130166712A1 (en) * 2011-12-27 2013-06-27 Microsoft Corporation Cloud-edge topologies
CN110740194A (en) * 2019-11-18 2020-01-31 南京航空航天大学 Micro-service combination method based on cloud edge fusion and application
CN110851429A (en) * 2019-11-21 2020-02-28 嘉兴学院 Edge computing credible cooperative service method based on influence self-adaptive aggregation
CN111031102A (en) * 2019-11-25 2020-04-17 哈尔滨工业大学 Multi-user, multi-task mobile edge computing system cacheable task migration method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
乐光学等: "边缘计算可信协同服务策略建模", 计算机研究与发展, no. 05, pages 186 - 208 *

Cited By (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112732442B (en) * 2021-01-11 2023-08-25 重庆大学 Distributed model for edge computing load balancing and solving method thereof
CN112732442A (en) * 2021-01-11 2021-04-30 重庆大学 Distributed model for edge computing load balancing and solving method thereof
CN113225385B (en) * 2021-04-16 2022-05-06 天津大学 Micro-service request distribution method facing edge computing environment and based on game theory
CN113225385A (en) * 2021-04-16 2021-08-06 天津大学 Micro-service request distribution method facing edge computing environment and based on game theory
CN113422803A (en) * 2021-05-19 2021-09-21 武汉工程大学 Seamless migration method for intelligent unmanned aerial vehicle inspection task based on end edge cloud cooperation
CN114172849A (en) * 2021-11-26 2022-03-11 华中科技大学 Deterministic traffic shaping method based on game theory
CN114172849B (en) * 2021-11-26 2024-07-12 华中科技大学 Deterministic traffic shaping method based on game theory
CN114401134A (en) * 2022-01-14 2022-04-26 重庆邮电大学 Edge-side-cooperative Internet of things distributed trusted management method
CN114401134B (en) * 2022-01-14 2023-12-15 深圳市兴海物联科技有限公司 Internet of things distributed trusted management method with end-side cooperation
CN114301987A (en) * 2022-03-07 2022-04-08 天津市城市规划设计研究总院有限公司 Dynamic scheduling method and system for virtualized network resources
CN114326822A (en) * 2022-03-09 2022-04-12 中国人民解放军66136部队 Unmanned aerial vehicle cluster information sharing method based on evolutionary game
CN116932228A (en) * 2023-09-14 2023-10-24 湖南希赛网络科技有限公司 Edge AI task scheduling and resource management system based on volunteer calculation
CN116932228B (en) * 2023-09-14 2023-12-12 湖南希赛网络科技有限公司 Edge AI task scheduling and resource management system based on volunteer calculation

Also Published As

Publication number Publication date
CN112132202B (en) 2023-11-17

Similar Documents

Publication Publication Date Title
CN112132202B (en) Edge computing collaboration alliance discovery method based on comprehensive trust evaluation
CN112020103A (en) Content cache deployment method in mobile edge cloud
US20230362095A1 (en) Method for intelligent traffic scheduling based on deep reinforcement learning
D’Oro et al. Auction-based resource allocation in OpenFlow multi-tenant networks
Wu et al. A reputation value‐based task‐sharing strategy in opportunistic complex social networks
Chen et al. RLMR: Reinforcement learning based multipath routing for SDN
Bakshi et al. Energy-efficient cluster head selection algorithm for IoT using modified glow-worm swarm optimization
Sreedevi et al. An Efficient Intra‐Cluster Data Aggregation and finding the Best Sink location in WSN using EEC‐MA‐PSOGA approach
CN115118728A (en) Ant colony algorithm-based edge load balancing task scheduling method
CN112867092A (en) Intelligent data routing method for mobile edge computing network
CN107948070B (en) A kind of mobile P 2 P network virtual link choosing method based on QoS
CN115426327A (en) Calculation force scheduling method and device, electronic equipment and storage medium
Macedo et al. Fuzzy-based load self-configuration in mobile P2P services
Kosunalp et al. SARL: A reinforcement learning based QoS-aware IoT service discovery model
CN114599043A (en) Air-space-ground integrated network resource allocation method based on deep reinforcement learning
Fathy et al. Impact of mobility on performance of P2P content discovery protocols over MANET
Jamali et al. Adaptive peer to peer resource discovery in grid computing based on reinforcement learning
Li et al. Cooperative computation offloading combined with data compression in mobile edge computing system
Du et al. Multiattribute evaluation model based on the KSP algorithm for edge computing
Lv et al. Mobile edge computing oriented multi-agent cooperative routing algorithm: A DRL-based approach
Amoretti et al. Performance evaluation of advanced routing algorithms for unstructured peer-to-peer networks
Ma Multi-Task Offloading via Graph Neural Networks in Heterogeneous Multi-access Edge Computing
Liu et al. Service placement considering robustness and dynamic in edge computing
CN112486669B (en) Self-organizing mobile edge computing platform and method
Wang et al. Interlinked Chain Method for Blockchain-Based Collaborative Learning in Vehicular Networks

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant