CN112235385B - Offshore edge computing low-overhead cloud edge intelligent cooperative ally member discovery method - Google Patents
Offshore edge computing low-overhead cloud edge intelligent cooperative ally member discovery method Download PDFInfo
- Publication number
- CN112235385B CN112235385B CN202011075338.5A CN202011075338A CN112235385B CN 112235385 B CN112235385 B CN 112235385B CN 202011075338 A CN202011075338 A CN 202011075338A CN 112235385 B CN112235385 B CN 112235385B
- Authority
- CN
- China
- Prior art keywords
- node
- service
- cooperative
- evaluation
- queue
- 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.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/51—Discovery or management thereof, e.g. service location protocol [SLP] or web services
-
- 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)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
- Computer And Data Communications (AREA)
Abstract
The invention discloses a method for discovering intelligent cooperative alligators at a low-overhead cloud edge of offshore edge computing, which comprises the steps of constructing an intelligent cooperative service network framework at the offshore cloud edge based on edge computing, abstracting behavior characteristics of nodes at the offshore edge computing, establishing a node trust and recommendation quantitative comprehensive evaluation model for inhibiting joint cheating, fusing and clustering the nodes of the alligators to different cooperative service pools according to comprehensive attribute evaluation, and realizing graded nearby service; based on the priority and load balance theory of the collaborative service request, a collaborative service pool building rule and a segment-page type self-adaptive lightweight heavy-load evasion member finding algorithm are designed to find a credible collaborative service member. The simulation analysis is carried out on the MCECS-MEC model performance by using the Router View data set, and the result shows that compared with the AODV and SR algorithm, the MCECS-MEC model performance reduces redundant transmission flow by 57.7 percent, 55.04 percent and link searching by 93.47 percent, the load rate is stabilized at 65 percent, the influence of overload, hot zone and cavity effects on the network performance can be effectively reduced, and the efficiency and the quality of the intelligent cooperative service at the edge of the offshore computing cloud are improved.
Description
Technical Field
The invention relates to the technical field of edge computing, in particular to a low-overhead cloud-edge intelligent collaborative ally member discovering method for offshore edge computing.
Background
The marine communication network has low efficiency and service quality due to the fact that the marine climate environment is complex and variable, unmanned island and reef are not uniformly distributed, and the base station is limited by resources, energy, meteorological environment and the like, and the development of the communication network is obviously lagged behind that of a land network.
In order to solve the problem, a summer-Ming-Hua research team provides a novel heaven-earth-sea integrated marine communication network architecture, the network architecture mainly comprises a bandwidth high-speed link formed by relay nodes such as a satellite, a stratosphere, a land data center, a base station, an island base station, a sea floating platform and a ship, wherein the land data center, the base station, the island base station and the like form a backbone communication network, and the ship, intelligent terminal equipment and the like are used as service bearing and task execution nodes to realize interactive application service. The Weisong research team is applied to indicate that edge computing is that node computing and storage capacity is given to the edge of the network, task processing is cooperatively executed near the data source side in a computing migration mode, computing results are uploaded to a cloud computing center, or tasks of the cloud computing center are issued to edge end nodes to be executed, so that network delay and data interaction overhead are reduced, task computing efficiency, network capacity and bandwidth are improved, and data safety and privacy are guaranteed. Research and analysis find that the idea of edge computing cooperative service is applied to a maritime wireless data communication network, and a reliable cooperative service system is constructed to be one of effective mechanisms for improving QoS (quality of service), network capacity and efficiency of a maritime wireless network, so that a key problem is how to find trusted cooperative member execution service quickly, efficiently, with light weight and low consumption.
Aiming at the problem, the key idea of the method is that task demand data drive is used as an engine, characteristics of edge computing service nodes are automatically and quickly obtained, nodes adaptive to demands are selected, edge computing resources are adaptively fused in a sharing mode of computing, storing, network communication resources and the like, an edge computing reliable collaborative service system taking an alliance as a center is constructed, computing tasks of nodes with limited capacity are migrated into a reliable and reliable collaborative service pool to be executed, processing, analysis and computing are executed on the data and the tasks locally and nearby as far as possible at the edge side, the energy consumption of the nodes is effectively reduced, and the QoS, the capacity, the efficiency and the life cycle of the nodes are improved.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, designs a low-overhead cloud-edge intelligent collaborative member discovery method for offshore edge computing, utilizes an information piggybacking technology to realize the sharing of characteristic attribute information and recommended service evaluation information of local network collaborative nodes, constructs a trusted collaborative filtering mechanism recommended collaborative service node and constructs a collaborative service characteristic information base; the method comprises the steps of constructing excellent, good, medium and general four-level queues according to node performance attributes in a classified mode, designing sparse, adaptive, balanced and efficient search algorithms by taking overhead as constraint, accessing the excellent and general queues by using a self-adaptive collaborative member discovery algorithm, accessing the good queues by using a heavy-load avoidance collaborative member discovery algorithm, and accessing the medium queues in the search by using a priority heavy-load avoidance collaborative member discovery algorithm, so that the low-overhead, reliable, balanced, rapid and efficient cloud-edge intelligent collaborative service of marine edge computing is realized.
The technical scheme for realizing the purpose of the invention is as follows:
a marine edge computing low-overhead cloud edge intelligent collaborative member discovery method comprises the following steps:
1) suppose the scale of the offshore edge computing network is n, i<n，j<N, edge node connection matrix N ═ a ij }，a ij E {0,1}, and if the node j is the collaborative service recommendation node of the node i, a ij 1, otherwise a ij 0; edge computation service relation matrix B ═ { B ═ B ij }，b ij E {0,1}, if node i can request edge computing service from node j, then b ij 1, otherwise b ij 0; the node trust relationship matrix T ═ T ij (s) }，t ij (s) E is {0,1}, and s is the current observation period number; edge computing service node comprehensive performance evaluation relation matrix G ═ G ij }，g ij ∈{0，1}；
Assuming that the observation period is T, the trust degree T of the node i to the node j in the s observation period ij (s) Satisfying the following formula (1):
wherein f is ij,succ (s) Representing the successful interaction times of the nodes i and j in the s-th observation period, f ij,fail (s) Representing the number of failed interaction times of the node i and the node j in the s-th observation period; epsilon (s) A dishonest behavior penalty factor for the s-th cycle; f. of j,succ (s) Representing the successful interaction times of the node j in the s-th period in the local area network, f j,fail (s) Representing the number of failed interaction times of the node j in the s-th period in the local area network; alpha represents the reference effective interaction times of the network node in the observation period time; f. of j The average interaction times of the node j in the period time are obtained; tau. j Representing the average online time of the node j in the observation period time; lev j Representing the average offline times of the node j in the observation period time; jac (i, j) is a jaccard coefficient of the node i and the node j, represents the contact ratio of the cooperative service node sets of the two nodes, and has more significance for recommendation when the contact ratio is higher; rtt (i, j) represents the network delay of node i and node j;
represents a periodic decay constant factor, u ∈ (0, 1);
constructing comprehensive evaluation g of service capability of node i to node j ij The following formula (2):
wherein h is j 、w j And m j Respectively representing the computing capacity, bandwidth and memory size which can be shared by the edge computing service node j;
2) according to the principle of autonomous fusion and near service, sensing the computing power, storage space and network performance characteristics of the cooperative service node by using the edge network, taking comprehensive evaluation of the service quality of the cooperative node as a basis, filtering an evaluation result based on a trust model, recommending the trusted cooperative service node, and constructing a cooperative service characteristic information base based on a distributed index; when a collaborative service task scheduling requirement exists, quickly indexing and extracting trusted service nodes to construct a trusted collaborative service set;
3) according to the service capability of the edge computing service node and the service quality evaluation and trust relationship of the neighbor nodes, the priority of the computing cooperative service node is constructed, and the priority evaluation enters a high-level, good-level, medium-level and general four-level queue according to the node service performance priority;
4) when the node requests the edge computing cooperative service, a proper edge computing service node is selected from the four-level queue to initiate a task cooperative computing migration request; the self-adaptive collaborative member discovery algorithm is used for accessing the excellent queue and the general queue, the heavy-load avoidance collaborative member discovery algorithm is used for accessing the good queue, the priority heavy-load avoidance collaborative member discovery algorithm is used for searching the medium queue, and the low-overhead credible and reliable, service balance, rapidness and high efficiency collaborative member discovery are realized;
5) after the cooperative service node responds to the computation task migration request, the computation task, the execution rule and the related dependent resources are migrated to the cooperative service node, rapid coupling, mapping and efficient cooperation are realized, and an execution result is returned;
6) after the cooperative service is completed, the node makes comprehensive evaluation according to the cooperative service quality, updates the cooperative service characteristic information base and periodically publishes the information to the neighbor nodes according to rules.
The step 2) specifically comprises the following steps:
2-1) adopting a trust filtering mechanism to obtain the recommendation information of the cooperative node, and setting an s-th observation period to obtain the evaluation r of the service node k ik ＝{r 1k ，r 2k ，…，r nk }, node j recommends a collaborative service evaluation r of node k to node i ik As shown in the following formula (3);
r ik ＝t ij t jk (3)
calculating the node score according to the mean square error to obtain the recommendation trust degree R ik As shown in the following formula (4);
wherein θ is a recommended anomaly threshold;
2-2) the fusion node i and the service node k are evaluated in the observation period, and the comprehensive evaluation of the service node k by the computing node i is shown in the following formula (5):
u ik ＝β m R ik +(1-β m )t ik (5)
wherein, beta is an evaluation sparsity coefficient, beta belongs to (0,1), and m is the evaluation frequency of the node i to the service node k; if node i has enough evaluation times for serving node k, i.e., (1-1/beta) m )<And when 0.1, ignoring the recommendation evaluation of the neighbor node to the service node k.
The step 3) specifically comprises the following steps:
3-1) calculating service priority evaluation eta of node i to node j j As shown in the following equation (6);
wherein Q is j Computing a load evaluation function, p, for a service node k for an edge j For memory occupancy, ρ j To calculate the capacity, d j For the bandwidth occupancy rate, p, rho and d are respectively a memory load threshold, a calculation load threshold and a bandwidth load threshold;
3-2) evaluating eta according to priority of node service performance j And (3) entering a high-level, good-level, medium-level and general four-level queue in a quartile method in a grading way, wherein the enqueuing rule is shown in the following formula (7):
the four-level bit method for entering the high-quality, good-quality, medium-quality and general four-level queues in a grading way specifically comprises the following steps:
3-2-1) setting the priority evaluation eta of the service performance of the node i to the node j in the s observation period j ＝{η 1 ，η 2 ，…，η n }，b ij Not equal to 0; evaluation η of priority j Correcting, rejecting invalid or redundant eta j 0;
3-2-2) setting quartile points according to a formula (7);
3-2-3) obtaining the recommended evaluation u of the service node k by the node i through the weighted average of the confidence k,i And if the nodes i and k already perform the cooperative service and the node i evaluates the cooperative service quality of the node k.
The step 4) specifically comprises the following steps:
4-1) designing a self-adaptive lightweight collaborative member discovery algorithm to scan excellent and general queues, selecting queue nodes in a segmented mode according to the exponential power of 2 to ensure access sparsity, carrying predecessor and successor nodes when selecting collaborative service nodes to realize local balance, and reselecting if the nodes are occupied, wherein a selected collaborative service node queue pointer function D is shown in the following formula (8):
wherein n is an amplification factor and l is a queue length; the MCECS-MEC adaptive collaborative ally-allied member discovery adaptation model is shown as the following formula (9):
the self-adaptive lightweight collaborative member discovery algorithm specifically comprises the following steps:
4-1-1) for a non-empty service queue, let k equal to 0, when k is<k max Turn 4-1-2);
4-1-2) generating the mapping values f in turn according to the formula (7) i J; if node P ij C ij If the value is 0, scheduling is executed, otherwise, regeneration is carried out according to the formula (7); scheduling times Req of corresponding node ij +, if Req ij Not less than M, then C ij ＝1；
4-1-3) the node initiates a cooperative service request, if receiving the cooperative service request, the step 4-1-4) is switched to, P ij 1 is ═ 1; otherwise k + +, go to step 4-1-1), if k is not less than k max And ending;
4-1-4) executing the collaborative service; when the node finishes the cooperative service, marking the service state as idle, P ij 0, and the synergy is evaluated;
4-1-5) updating the node state information table and the cooperative characteristic information table;
4-1-6) judging whether the test period is finished or not, and if not, turning to the step 4-1-1); if so, reset Req ij ＝0，C ij Finish when equal to 0;
4-2) constructing an MCECS-MEC heavy load evasion cooperative member discovery algorithm, scanning a good queue and a medium queue based on the priority MCECS-MEC heavy load evasion cooperative member discovery algorithm, and selecting a cooperative service node queue pointer function D as shown in the following formula (10):
head (L) is the head node pointer function of the return queue L; the heavy-load avoidance cooperative member discovery algorithm adaptation model is shown in the following formula (11):
setting the maximum value of the collaboration times as M, and specifically, the heavy-load avoidance collaboration member discovering algorithm and the priority-based heavy-load avoidance collaboration member discovering algorithm comprise the following steps:
4-2-1) sorting the non-empty queues L in descending order of priority, and enabling the pointer f to be Head;
4-2-2) if the node j mapped by f is occupied, fully loading the queue and ending; otherwise, initiating a cooperation request;
4-2-3) scanning the queue in the order of formula (10), if there is f that meets the requirements of the adapted model (11) i Mapped node j, then Head f i ；
4-2-4) executing the cooperative service;
when the node finishes the cooperative service, marking the service state as idle, P ij ＝0，And evaluating the synergy; if the current queue executes the priority order, and η Head >η j If yes, then Head ═ f, end;
4-2-5) updating the node state information table and the cooperation characteristic information table;
judging whether the test period is finished or not, and if not, turning to the step 4-2-1); if so, reset Req ij ＝0，C ij And end when 0.
The invention has the beneficial effects that: the invention provides a marine edge computing low-overhead cloud edge intelligent collaborative ally member discovering method which comprises the steps of constructing a marine cloud edge intelligent collaborative service network framework based on edge computing, abstracting marine edge computing node behavior characteristics, establishing a node trust and recommendation quantitative comprehensive evaluation model for inhibiting joint cheating, fusing and clustering member-authorized nodes to different collaborative service pools according to comprehensive attribute evaluation of the members, and realizing graded nearby services; based on the priority and load balance theory of the collaborative service request, a collaborative service pool building rule and a segment-page type self-adaptive lightweight heavy-load avoidance member discovering algorithm are designed to find a credible collaborative service member. Simulation analysis is carried out on the MCECS-MEC model performance based on the Router View public data set, and simulation experiments show that compared with AODV and SR algorithms, the MCECS-MEC model reduces redundant transmission flow of 57.7% and 55.04% and link searching of 93.47%, the load rate is stabilized at 65%, the MCECS-MEC model can effectively reduce the influence of overload, hot area, cavity effect and the like on the network performance, and the cloud-edge intelligent cooperative service efficiency and quality of offshore edge computing are improved.
Drawings
FIG. 1 is a diagram of a marine intelligent wireless edge computing network architecture;
FIG. 2 is a schematic diagram of an adaptive lightweight collaborative alliance discovery algorithm;
FIG. 3 is a schematic diagram of a heavy-duty evasion collaborative member discovery algorithm;
FIG. 4 is a graph comparing the calculated mean flow at the edge of the sea;
FIG. 5 is a chart comparing the receiving rates of the edge computing collaborative tasks at sea;
FIG. 6 is a chart comparing success rates of edge computing collaborative services at sea;
FIG. 7 is a chart of average load comparison for a marine edge computing collaborative services pool;
FIG. 8 is a chart of comparing reputation of a marine edge computing collaborative service node;
FIG. 9 is a comparison graph of offshore edge computing collaborative service access delays;
FIG. 10 is a graph comparing the response delay of the offshore edge computing task execution;
fig. 11 is a comparison graph of the link re-search rate of the offshore edge computing cooperative service.
Detailed Description
The invention is further illustrated but not limited by the following figures and examples.
The embodiment is as follows:
the offshore intelligent wireless edge computing network is shown in figure 1 and comprises a land cloud computing data center, a shore-based base station, an island-reef relay base station, an intelligent relay floating platform, an unmanned ship, a shipborne intelligent device, a mobile intelligent device and the like, wherein the shore-based base station, the island-reef relay base station and the intelligent relay floating platform form an offshore mobile communication backbone network, the ship, the unmanned ship and the unmanned ship form a secondary wireless mobile route, the shipborne intelligent device, the mobile intelligent device and the like are network end nodes, the intelligent wireless edge computing network is accessed in a wireless Mesh autonomous fusion mode, and the base stations are accessed into the land data center to upload offshore sensing data and acquire resources and services.
The offshore intelligent equipment such as unmanned ships, unmanned planes, ships and the like can generate a large amount of calculation tasks and source data, if the offshore intelligent equipment is completely uploaded to a shore-based data center for processing, a large amount of network resources, energy, bandwidth and the like are consumed, so that the cavity avalanche effect caused by network jitter, overload, hot areas and the like is caused, the network performance and capacity are rapidly reduced, even network collapse occurs, the offshore intelligent equipment is endowed with a calculation processing mission, data analysis and calculation are carried out near the data source side, so that the data uploading expense and the calculation efficiency are reduced, the response speed and the response performance are improved, the ocean high-speed and high-capacity streaming data service and non-blind area coverage are realized, an effective solution for providing robust QoS network service for users is provided, in order to improve the service quality and the efficiency, the shore-based land cloud calculation data center is taken as a core, an island reef base station, an intelligent relay floating platform and the like construct basic edge cloud calculation service access convergence, and an edge computing cooperative service system is constructed, and computing tasks are migrated to a cooperative service pool to be executed, so that flexible, robust, efficient and reliable service is realized.
The scale N of the network nodes is less than or equal to 1500, 1 network center, 1 shore-based center base station, 3 island and reef relay base stations, 3 intelligent floating platforms, 175 super nodes and a plurality of general nodes. The base station cluster size N is less than or equal to 250, the network topology is a distributed random topology structure, and nodes can dynamically and unrestrictedly join and leave the network. The initial network adopts a flooding routing mechanism to obtain characteristic information of a computing service node, an initial offshore edge computing network cluster is built according to the service capability and the credibility of the computing service node, the network reaches a stable state in 15min, the network topology structure is shown in figure 2, and the network characteristic parameters are shown in table 1.
In order to test the performance of the constructed offshore edge computing collaborative service system, flow, load, credit, task receiving rate, collaborative success rate, throughput rate, load balance and the like are used as evaluation indexes, the offshore edge computing low-overhead cloud edge intelligent collaborative member discovery method is adopted to perform contrastive analysis with random nomadic Routing (SR), On-demand Routing (AODV), Local Cooperative clustering of storage clustering, LCR-SR) based On neighbor collaborative Recommendation, Local Cooperative Routing of Ad On-demand clustering, LCR-DV based On neighbor collaborative Recommendation, and an experimental control group is set:
·SR
·AODV
·LCR-SR
·LCR-AODV
·MCECS-MEC
setting that 100 computing tasks to be migrated exist in a data source server, wherein the quantity of each computing task to be migrated is not less than 500MB and not more than 1024MB, and carrying out surge type service request test on the data source server where a target computing task is located, setting:
1) if the cooperative service request of any node is more than or equal to 6 times/s, the edge network with the scale of 1500 nodes sends more than or equal to 9000 cooperative service requests to the target server at the same time;
2) the network is constrained as follows: the link re-searching times is less than or equal to 3, and 5 continuous stream data files are executed each time;
3) the sampling period of the access amount and the flow is 30s, the sampling period of the credit degree, the task receiving rate, the cooperative success rate, the calculation load, the delay, the link searching rate and the like is 5s, and the experimental test time is 6 h.
A marine edge computing low-overhead cloud edge intelligent collaborative member discovery method includes the steps of adopting neighbor sensing collaborative service characteristic attributes, obtaining recommendation information through a network piggybacking technology, constructing a collaborative service set through individual network requirements and credible collaborative fusion filtering, designing a low-overhead four-level queue intelligent Hash search algorithm, and achieving stable, balanced, efficient and reliable computing task migration, and specifically includes the following steps:
1) suppose the scale of the offshore edge computing network is n, i<n，j<N, edge node connection matrix N ═ a ij }，a ij E {0,1}, and if the node j is the collaborative service recommendation node of the node i, a ij 1, otherwise a ij 0; edge computation service relation matrix B ═ { B ═ B ij }，b ij E {0,1}, and b if node i can request edge computing service from node j ij 1, otherwise b ij 0. node trust relationship matrix T ═ T ij (s) }，t ij (s) E (0,1), wherein s is the current observation period number; edge computing service node comprehensive performance evaluation relation matrix G ═ G ij }，g ij ∈(0,1)；
Assuming that the observation period is T, the trust degree T of the node i to the node j in the s observation period ij (s) Satisfies formula (1):
wherein:
f ij,succ (s) and f ij,fail (s) Respectively representing the successful and failed interactive times of the node i and the node j in the s-th observation period; epsilon (s) A dishonest behavior penalty factor for the s-th cycle; f. of j,succ (s) And f j,fail (s) Respectively representing the successful and failed interactive times of the node j in the s-th period in the local area network; alpha represents the reference effective interaction times of the network node in the observation period time, f j The average interaction times of the node j in the period time are obtained; tau is j And lev j Respectively representing the average online time and the offline times of the node j in the observation period time; jac (i, j) is a jaccard coefficient of the node i and the node j, represents the contact ratio of the cooperative service node sets of the two nodes, and has more significance for recommendation when the contact ratio is higher; rtt (i, j) represents the network delay of node i and node j;
represents a periodic decay constant factor, u ∈ (0, 1);
constructing comprehensive evaluation g of service capability of node i to node j ij As in equation (2):
wherein h is j ，w j ，m j Respectively representing the computing capacity, bandwidth and memory size which can be shared by the edge computing service node j;
2) according to the principle of autonomous fusion and near service, sensing characteristics of cooperative service nodes such as computing capacity, storage space and network performance by using an edge network, taking comprehensive evaluation of service quality of the cooperative nodes as a basis, filtering evaluation results based on a trust model, recommending the trusted cooperative service nodes, and constructing a cooperative service characteristic information base based on distributed indexes; when a collaborative service task scheduling requirement exists, quickly indexing and extracting trusted service nodes to construct a trusted collaborative service set;
3) according to the service capability of the edge computing service node and the service quality evaluation and trust relationship of the neighbor nodes, the priority of the computing cooperative service node is constructed: entering a high-quality, good, medium and general four-level queue according to the priority evaluation of the node service performance;
4) when the node requests the edge computing cooperative service, selecting a proper edge computing service node from the four-level queue to initiate a task cooperative computing migration request; the self-adaptive collaborative member discovery algorithm is used for accessing the excellent queue and the general queue, the heavy-load avoidance collaborative member discovery algorithm is used for accessing the good queue, the priority heavy-load avoidance collaborative member discovery algorithm is used for searching the medium queue, and the low-overhead credible and reliable, service balance, rapidness and high efficiency collaborative member discovery are realized;
5) after the cooperative service node responds to the computation task migration request, the computation task, the execution rule and the related dependent resources thereof are migrated to the cooperative service node, so that quick coupling, mapping and efficient cooperation are realized, and an execution result is returned;
6) after the cooperative service is completed, the node makes comprehensive evaluation according to the cooperative service quality, updates the cooperative service characteristic information base and periodically publishes the information to the neighbor nodes according to rules.
The step 2) specifically comprises the following steps:
2-1) adopting a trust filtering mechanism to obtain the recommendation information of the cooperative node, and setting an s-th observation period to obtain the evaluation r of the service node k ik ＝{r 1k ，r 2k ，…，r nk }, node j recommends a collaborative service evaluation r of node k to node i ik The following formula (3):
r ik ＝t ij t jk (3)
calculating the node score according to the mean square error to obtain the recommendation trust degree R ik As shown in equation (4);
wherein θ is a recommended anomaly threshold;
2-2) the fusion node i and the service node k are evaluated in the observation period, and the comprehensive evaluation of the service node k by the computing node i is shown in the following formula (5):
u ik ＝β m R ik +(1-β m )t ik (5)
wherein beta is an evaluation sparsity coefficient, beta belongs to (0,1), and m is the evaluation times of the service node k by the node i; if node i has enough evaluation times for serving node k, i.e., (1-1/beta) m )<When 0.1, neglecting the recommendation evaluation of the neighbor node to the service node k;
the step 3) specifically comprises the following steps:
3-1) calculating service priority evaluation eta of node i to node j j As shown in the following equation (6):
wherein Q is j Computing a load evaluation function, p, for a service node k for an edge j For memory occupancy, ρ j To calculate the capacity, d j For the bandwidth occupancy rate, p, rho and d are respectively a memory load threshold, a calculation load threshold and a bandwidth load threshold;
3-2) evaluating eta according to priority of node service performance j And (3) entering a high-level, good-level, medium-level and general four-level queue in a quartile method in a grading way, wherein the enqueuing rule is shown in the following formula (7):
the step 3-2), the four-decimal place method is classified into a high-quality, good, medium and general four-level queue, and the method specifically comprises the following steps:
3-2-1) setting the priority evaluation eta of the service performance of the node i to the node j in the s observation period j ＝{η 1 ，η 2 ，…，η n }，b ij Not equal to 0; evaluation η of priority j Correcting, rejecting invalid or redundant eta j 0;
3-2-2) setting quartile points according to a formula (7);
3-2-3) obtaining the recommended evaluation u of the service node k by the node i through the weighted average of the confidence k,i If the nodes i and k already perform the cooperative service and the node i evaluates the cooperative service quality of the node k,
the step 4) specifically comprises the following steps:
4-1) designing a self-adaptive lightweight collaborative member discovery algorithm to scan excellent and general queues, selecting queue nodes in a segmented manner according to the exponential power of 2 to ensure the access sparsity, carrying predecessor nodes and successor nodes when selecting collaborative service nodes to realize local equilibrium, and if the nodes are occupied, reselecting the nodes.A queue pointer function D of the selected collaborative service nodes is shown in the following formula (8):
wherein n is an amplification factor, l is a queue length, and the MCECS-MEC adaptive collaborative ally-member discovery adaptation model is shown in the following formula (9):
the adaptive lightweight collaborative member discovery algorithm, as shown in fig. 2, specifically includes the following steps:
4-1-1) for a non-empty service queue, let k equal to 0, when k is<k max Turning to 4-1-2);
4-1-2) generating the mapping values f in turn according to the formula (7) i J; if node P ij C ij If the value is 0, scheduling is executed, otherwise, regeneration is carried out according to the formula (7); scheduling times Req of corresponding node ij +, if Req ij Not less than M, then C ij ＝1；
4-1-3) node initiates a cooperative service request, and if the cooperative service request is received, the Step4 is turned to ij 1; otherwise k + +, go to Step1, if k ≧ k max And ending;
4-1-4) executing the collaborative service; when the node completes the cooperationAfter servicing, marking its service status as idle, P ij 0 and evaluating the synergy;
4-1-5) updating the node state information table and the cooperative characteristic information table;
4-1-6) if the test period is not finished, if not, turning to the step 4-1-1); if so, reset Req ij ＝0，C ij Ending when the value is 0;
4-2) constructing an MCECS-MEC heavy load evasion cooperative member discovery algorithm, scanning a good queue and a medium queue based on the priority MCECS-MEC heavy load evasion cooperative member discovery algorithm, and selecting a cooperative service node queue pointer function D as shown in the following formula (10):
the heavy load avoidance collaborative ally allied member discovery algorithm adaptation model is shown in the following formula (11):
setting the maximum value of the collaboration times as M, and adopting a heavy-load avoidance collaboration member discovering algorithm and a priority-based heavy-load avoidance collaboration member discovering algorithm, as shown in FIG. 3, the method specifically comprises the following steps:
4-2-1) sorting the non-empty queues L in descending order of priority, and enabling the pointer f to be Head;
4-2-2) if the node j mapped by f is occupied, fully loading the queue and ending; otherwise, initiating a cooperation request;
4-2-3) scanning the queue according to the formulas (10) and (11) in sequence, and if f meeting the requirement exists i Mapped node j, then Head ═ f i ；
4-2-4) executing the cooperative service;
when the node finishes the cooperative service, marking the service state as idle, P ij 0 and evaluating the synergy; if the current queue executes the priority order, and η Head >η j If yes, then Head ═ f, end;
4-2-5) updating the node state information table and the cooperation characteristic information table;
the step 6) specifically comprises the following steps:
6-1) comparative analysis of system performance, as can be seen from FIGS. 4-6 and Table 1, the following characteristics are obtained:
a) the surge access test comprises two stages, wherein each stage comprises 4 surge impacts; in the surge access test of the first stage, the occurrence moments of 4 times of surge impact are respectively 30min, 60min, 90min and 120 min; the AODV and SR algorithm reaches the peak flow values of 10299.35MB/min and 9789.11MB/min at the 1 st surge impact time of 30min, the Martian effect is generated due to overload of the service node in subsequent surge access, and the flow is continuously reduced; the LCR-AODV and LCR-SR algorithms reach peak flow values of 17068.95MB/min and 17737.83MB/min when the 3 rd surge impacts for 90min, a Martian effect is generated due to overload of a service node in subsequent surge access, and the flow is continuously reduced; the MCECS-MEC algorithm is kept stable in the surge access of the first stage, and the flow peak value of 14091.35MB/min is reached at 150min when the 4 th surge impact occurs;
b) the second stage executes the same surge access test as the first stage, wherein the peak values of the 1 st, 3 rd and 4 th surge impacts of AODV, SR, LCR-AODV, LCR-SR and MCECS-MEC reach 6073.16MB/min, 7670.77MB/min, 12036.8MB/min, 12389.4MB/min and 10446.61MB/min respectively; because the cooperative service cluster constructed by the super nodes in the second stage is basically completed and part of tasks are migrated to the cooperative service cluster constructed by the super nodes for processing, the total service flow of the system is averagely reduced by 28.5 percent compared with that of the system in the first stage;
c) the task receiving rate and the service success rate of the AODV, the SR, the LCR-AODV and the LCR-SR reach 99% in 15min, when the surge test starts, the task receiving rate and the service success rate start to decrease and reach stability in 180min, the task receiving rate is respectively 68%, 68%, 82% and 82%, and the service success rate is respectively 68%, 67%, 82% and 82%; the task acceptance rate and service success rate of the MCECS-MEC algorithm were stable at 99% during the test.
The experimental results show that:
a) when the surge impact occurs for the first time in the first stage, the average flow of the LCR-AODV and the LCR-SR is reduced by 37.1 percent and 29.1 percent compared with the average flow of the AODV and the average flow of the SR, which shows that the characteristic attribute information and the service evaluation information of the cooperative service node are extracted through a neighbor cooperative recommendation algorithm, so that the node can quickly construct the cooperative service to implement task migration, and redundant access is reduced; compared with LCR-AODV and LCR-SR algorithms, the MCECS-MEC has the advantages that the average flow is reduced by 20.6 percent and 25.94 percent, and the fact that sparse task scheduling access is scheduled through a four-level queue is shown, so that the receiving rate and the success rate of access service are effectively improved, and redundant access is reduced; compared with AODV, SR, the MCECS-MEC scheduling algorithm reduces redundant flow of 57.7% and 55.04% of the system as a whole;
b) compared with AODV and SR, the LCR-SR has the advantages that the maximum task concurrency is improved by 15.07 percent and 16.33 percent, and the average throughput is improved by 21.30 percent and 23.66 percent; compared with LCR-AODV, LCR-SR, the MCECS-MEC algorithm has the advantages that the maximum task concurrency is improved by 16.11% and 16.98%, and the average throughput is improved by 25.79% and 23.86%; compared with AODV, SR, the maximum task concurrency number of the MCECS-MEC algorithm is respectively improved by 33.61% and 36.08%, and the average throughput is respectively improved by 52.58% and 53.17%;
c) the MCECS-MEC scheduling algorithm realizes efficient and sparse cooperative node feature extraction and task allocation, prevents the task overload of the optimal queue node, reduces the generation of network hot areas and realizes relatively stable calculation task migration.
d) Compared with the AODV and the SR, the LCR-SR has the advantages that the task receiving rate and the success rate are improved by 20.6 percent; compared with LCR-AODV, LCR-SR, the MCECS-MEC algorithm has the advantages that the task receiving rate and the success rate are improved by 20.7%; compared with AODV, SR, the MCECS-MEC algorithm improves the overall service efficiency of the system by 45.6%;
6-2) comparative analysis of service balance, as can be seen from FIGS. 7-8, the following features are provided:
a) in the first stage surge access test, the AODV and SR algorithm reaches full load when the 1 st surge impact lasts for 30min, the trust degree reaches peak values of 0.89 and 0.93, and then the peak values are rapidly reduced; when the 1 st surge impact of the LCR-AODV and LCR-SR algorithm is carried out for 30min, the trust reaches peak values of 0.85 and 0.95, the load rate is 65 percent and 62 percent, the full load is reached when the 2 nd surge impact is carried out for 60min, the average trust is reduced to 0.5, and when the 3 rd surge impact is carried out, the average trust is reduced to 0.1; the MCECS-MEC algorithm load is kept stable, the average load rate is 71%, and the trust degree is stable at 0.93;
b) in the second stage surge access test, the full load is reached by the AODV and SR algorithms when the 1 st surge impact lasts for 210 min; the LCR-AODV, LCR-SR algorithm reaches full load when the 2 nd surge impact lasts for 240 min; the MCECS-MEC algorithm load is kept stable, and the average load rate is 65%;
the experimental results show that:
a) the MCECS-MEC scheduling algorithm can effectively solve the problems of high congestion, single-point failure and low data distribution efficiency in edge computing, improves service quality, realizes load balancing and provides guarantee for large-scale trusted edge computing service.
6-3) analysis of QoS and overhead, as can be seen from FIGS. 9-11, the following characteristics are obtained:
a) in the surge access test, the AODV and the SR directly discard overload requests, the cooperative service access delay is 30.18ms and 33.22ms, and the task execution delay is 30.21ms and 33.33 ms; the LCR-AODV and LCR-SR algorithm migrates the overload request to other cooperative nodes for execution, the average access delay of the cooperative service is 56.45ms and 53.46ms, and the task execution delay is 56.26ms and 53.40 ms; the MCECS-MEC algorithm realizes sparse task allocation, the average access delay of the cooperative service is 30.35ms, and the task execution delay is 31.57 ms;
b) the link re-searching rates of AODV and SR algorithms are 0.48 and 0.44 respectively; the link re-searching rates of the LCR-AODV and the LCR-SR algorithm are 0.20 and 0.19 respectively; the link re-searching rate of the MCECS-MEC algorithm is 0.03;
the experimental results show that:
a) compared with LCR-AODV, the MCECS-MEC algorithm has the advantages that the LCR-SR access delay is reduced by 46.23% and 43.22%, and the task execution delay is reduced by 43.89% and 40.88%;
b) compared with the AODV, the LCR-SR link has the advantages that the bandwidth utilization rate of the SR algorithm is improved by 58.33 percent and 56.82 percent; compared with LCR-AODV, the LCR-SR algorithm has the advantages that the link bandwidth utilization rate is improved by 85.00% and 84.21% by the MCECS-MEC algorithm; compared with the AODV and SR algorithm, the MCECS-MEC algorithm has the advantages that the system bandwidth utilization rate is improved by 93.47%;
to summarize:
the offshore edge calculation is restricted by a plurality of factors such as marine environment, limited resources, communication distance, communication system and the like, so that the problems of jitter, vehicle carrying, hot area, strategy and the like exist when the offshore edge calculation task is implemented; a method for discovering intelligent cooperative allied personnel at the edge of a sea by computing low-overhead cloud edges is provided. The method is based on an information piggybacking technology to achieve sharing of characteristic information and evaluation information of cooperative service nodes, trusted cooperative service nodes are rapidly extracted through trust filtering and service quality feedback, a cooperative service characteristic information base is built, based on task driving, four-level Hash ring queues of superior, good, medium and general are built in a classifying mode according to service priorities of the superior, good, medium and general queues, a self-adaptive cooperative member discovery algorithm is designed to access the cooperative service nodes, heavy load avoidance cooperative member discovery algorithms and priority heavy load avoidance cooperative member discovery algorithms are designed for the superior and medium queues to achieve sparse search and selection of the cooperative service nodes, and trusted, reliable and service-balanced cooperative service is achieved. In a constructed experimental environment, compared with AODV and SR methods, the algorithm reduces redundant transmission flow of 57.7% and 55.04% and link re-search of 93.47%, the calculation load rate of the cooperative service node is stabilized at 65%, the problems of overload and hot zone effect of the service node are effectively solved, the purposes of cooperative service, resource sharing and load balancing are achieved, and the service quality of the offshore edge calculation cooperative service is improved.
TABLE 1 characteristic parameter table (I) of intelligent cooperative service algorithm for offshore edge computing
TABLE 1 characteristic parameter Table of intelligent cooperative service algorithm for offshore edge computing (SUZUYI)
Claims (1)
1. A method for discovering intelligent cooperative alligators at an offshore edge computing low-overhead cloud edge is characterized by comprising the following steps:
1) suppose the scale of the offshore edge computing network is n, i<n，j<N, edge node connection matrix N ═ a ij }，a ij E {0,1}, and if the node j is the collaborative service recommendation node of the node i, a ij 1, otherwise a ij 0; edge computation service relation matrix B ═ { B ═ B ij }，b ij E {0,1}, if node i can request edge computing service from node j, then b ij 1, otherwise b ij 0; node trust relationship matrix T ═ T ij (s) }，t ij (s) E is {0,1}, and s is the current observation period number; edge computing service node comprehensive performance evaluation relation matrix G ═ G ij }，g ij ∈{0，1}；
Assuming that the observation period is T, the trust degree T of the node i to the node j in the s observation period ij (s) Satisfying the following formula (1):
wherein f is ij,succ (s) Representing the successful interaction times of the nodes i and j in the s-th observation period, f ij,fail (s) Representing the number of failed interaction times of the node i and the node j in the s-th observation period; epsilon (s) A dishonest behavior penalty factor for the s-th cycle; f. of j,succ (s) Representing the successful interaction times of the node j in the s-th period in the local area network, f j,fail (s) Representing the number of failed interaction times of the node j in the s-th period in the local area network; alpha represents the reference effective interaction times of the network node in the observation period time; f. of j The average interaction times of the node j in the period time are obtained; tau is j Representing the average online time length of the node j in the observation period time; lev j Indicating node j is in observation periodAverage offline times over time; jac (i, j) is a jaccard coefficient of the node i and the node j, represents the contact ratio of the cooperative service node sets of the two nodes, and has more significance for recommendation when the contact ratio is higher; rtt (i, j) represents the network delay of node i and node j;
represents a periodic decay constant factor, u ∈ (0, 1);
building comprehensive evaluation g of service capability of node i to node j ij The following formula (2):
wherein h is j 、w j And m j Respectively representing the computing capacity, bandwidth and memory size which can be shared by the edge computing service node j;
2) according to the principle of autonomous fusion and near service, sensing the computing power, storage space and network performance characteristics of the cooperative service node by using the edge network, taking comprehensive evaluation of the service quality of the cooperative node as a basis, filtering an evaluation result based on a trust model, recommending the trusted cooperative service node, and constructing a cooperative service characteristic information base based on a distributed index; when a collaborative service task scheduling requirement exists, quickly indexing and extracting trusted service nodes to construct a trusted collaborative service set;
3) according to the service capability of the edge computing service node and the service quality evaluation and trust relationship of the neighbor nodes, the priority of the computing cooperative service node is constructed, and the priority evaluation enters a high-level, good-level, medium-level and general four-level queue according to the node service performance priority;
4) when the node requests the edge computing cooperative service, selecting a proper edge computing service node from the four-level queue to initiate a task cooperative computing migration request; the self-adaptive collaborative member discovery algorithm is used for accessing the excellent queue and the general queue, the heavy-load avoidance collaborative member discovery algorithm is used for accessing the good queue, the priority heavy-load avoidance collaborative member discovery algorithm is used for searching the medium queue, and the low-overhead credible and reliable, service balance, rapidness and high efficiency collaborative member discovery are realized;
5) after the cooperative service node responds to the computation task migration request, the computation task, the execution rule and the related dependent resources are migrated to the cooperative service node, rapid coupling, mapping and efficient cooperation are realized, and an execution result is returned;
6) after the cooperative service is finished, the node makes comprehensive evaluation according to the cooperative service quality, updates a cooperative service characteristic information base and periodically publishes the cooperative service characteristic information base to the neighbor nodes according to rules;
the step 2) specifically comprises the following steps:
2-1) adopting a trust filtering mechanism to obtain the recommendation information of the cooperative node, and setting an s-th observation period to obtain the evaluation r of the service node k ik ＝{r 1k ，r 2k ，…，r nk }, node j recommends a collaborative service evaluation r of node k to node i ik As shown in the following formula (3);
r ik ＝t ij t jk (3)
calculating the node score according to the mean square error to obtain the recommended trust degree R ik As shown in the following formula (4);
wherein
Is a recommended anomaly threshold;
2-2) the fusion node i and the service node k are evaluated in the observation period, and the comprehensive evaluation of the service node k by the computing node i is shown in the following formula (5):
u ik ＝β m R ik +(1-β m )t ik (5)
wherein, beta is an evaluation sparsity coefficient, beta belongs to (0,1), and m is the evaluation frequency of the node i to the service node k; if node i has enough evaluation times for serving node k, i.e., (1-1/beta) m )<At 0.1, the neighbor node is ignoredPerforming recommendation evaluation on a service node k;
the step 3) specifically comprises the following steps:
3-1) calculating service priority evaluation eta of node i to node j j As shown in the following equation (6);
wherein Q is j Computing a load evaluation function, p, for a service node k for an edge j For memory occupancy, ρ j For computing power, d j For the bandwidth occupancy rate, p, rho and d are respectively a memory load threshold, a calculation load threshold and a bandwidth load threshold;
3-2) evaluating eta according to priority of node service performance j And (3) entering a high-level, good-level, medium-level and general four-level queue in a quartile method in a grading way, wherein the enqueuing rule is shown in the following formula (7):
the four-level bit method for entering the high-quality, good-quality, medium-quality and general four-level queues in a grading way specifically comprises the following steps:
3-2-1) setting the priority evaluation eta of the service performance of the node i to the node j in the s observation period j ＝{η 1 ，η 2 ，…，η n }，b ij Not equal to 0; evaluation η of priority j Correcting, rejecting invalid or redundant eta j 0;
3-2-2) setting quartile points according to a formula (7);
3-2-3) obtaining the recommended evaluation u of the service node k by the node i through the weighted average of the confidence k,i If the nodes i and k already carry out the cooperative service and the node i evaluates the cooperative service quality of the node k;
the step 4) specifically comprises the following steps:
4-1) designing a self-adaptive lightweight collaborative member discovery algorithm to scan excellent and general queues, selecting queue nodes in a segmented mode according to the exponential power of 2 to ensure access sparsity, carrying predecessor and successor nodes when selecting collaborative service nodes to realize local balance, and reselecting if the nodes are occupied, wherein a selected collaborative service node queue pointer function D is shown in the following formula (8):
wherein n is an amplification factor and l is a queue length; the MCECS-MEC adaptive collaborative ally-allied member discovery adaptation model is shown as the following formula (9):
the self-adaptive lightweight collaborative member discovery algorithm specifically comprises the following steps:
4-1-1) for a non-empty service queue, let k equal to 0, when k is<k max Turn 4-1-2);
4-1-2) generating the mapping values f in turn according to the formula (7) i J is; if node P ij C ij If the value is 0, scheduling is executed, otherwise, regeneration is carried out according to the formula (7); scheduling times Req of corresponding node ij +, if Req ij Not less than M, then C ij ＝1；
4-1-3) the node initiates a cooperative service request, if receiving the cooperative service request, the step 4-1-4) is switched to, P ij 1 is ═ 1; otherwise k + +, go to step 4-1-1), if k is not less than k max And ending;
4-1-4) executing the collaborative service; when the node finishes the cooperative service, marking the service state as idle, P ij 0, and the synergy is evaluated;
4-1-5) updating the node state information table and the cooperative characteristic information table;
4-1-6) judging whether the test period is finished or not, and if not, turning to the step 4-1-1); if so, reset Req ij ＝0，C ij Ending when the value is 0;
4-2) constructing an MCECS-MEC heavy load evasion cooperative member discovery algorithm, scanning a good queue and a medium queue based on the priority MCECS-MEC heavy load evasion cooperative member discovery algorithm, and selecting a cooperative service node queue pointer function D as shown in the following formula (10):
head (L) is the head node pointer function of the return queue L; the heavy-load avoidance cooperative member discovery algorithm adaptation model is shown in the following formula (11):
setting the maximum value of the collaboration times as M, and specifically, the heavy-load avoidance collaboration member discovering algorithm and the priority-based heavy-load avoidance collaboration member discovering algorithm comprise the following steps:
4-2-1) sorting the non-empty queues L in descending order of priority, and enabling the pointer f to be Head;
4-2-2) if the node j mapped by f is occupied, fully loading the queue and ending; otherwise, initiating a cooperation request;
4-2-3) scanning the queue in the order of formula (10), if there is f that meets the requirements of the adapted model (11) i Mapped node j, then Head ═ f i ；
4-2-4) executing the cooperative service;
when the node finishes the cooperative service, marking the service state as idle, P ij 0, and the synergy is evaluated; if the current queue executes the priority order, and η Head >η j If yes, then Head ═ f, end;
4-2-5) updating the node state information table and the cooperative characteristic information table;
judging whether the test period is finished, if not, turning to the step 4-2-1); if so, reset Req ij ＝0，C ij Finish when 0.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011075338.5A CN112235385B (en) | 2020-10-09 | 2020-10-09 | Offshore edge computing low-overhead cloud edge intelligent cooperative ally member discovery method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011075338.5A CN112235385B (en) | 2020-10-09 | 2020-10-09 | Offshore edge computing low-overhead cloud edge intelligent cooperative ally member discovery method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112235385A CN112235385A (en) | 2021-01-15 |
| CN112235385B true CN112235385B (en) | 2022-08-30 |
Family
ID=74120104
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011075338.5A Active CN112235385B (en) | 2020-10-09 | 2020-10-09 | Offshore edge computing low-overhead cloud edge intelligent cooperative ally member discovery method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112235385B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113985475B (en) * | 2021-10-28 | 2023-09-05 | 北京石油化工学院 | Microseism monitoring data transmission method based on Internet of things terminal Bian Yun cooperation |
| CN115174595A (en) * | 2022-07-07 | 2022-10-11 | 哈尔滨理工大学 | Edge cloud resource collaborative optimization scheme based on edge cloud system |
| CN115333611B (en) * | 2022-08-31 | 2023-10-24 | 长沙理工大学 | Wireless communication method, unmanned aerial vehicle and mobile edge computing system |
| CN116132964B (en) * | 2022-12-16 | 2024-04-16 | 西安电子科技大学 | Parameter setting method for reducing multi-node cooperative discovery delay |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109379739A (en) * | 2018-09-28 | 2019-02-22 | 嘉兴学院 | A kind of credible cooperating service method of sea wireless Mesh netword |
| CN109587735A (en) * | 2018-11-12 | 2019-04-05 | 电子科技大学 | A kind of cooperation with service caching method for measuring mobile edge calculations based on time delay |
| CN110502328A (en) * | 2019-08-28 | 2019-11-26 | 嘉兴学院 | A kind of credible cotasking moving method of sea edge calculations |
| CN110851429A (en) * | 2019-11-21 | 2020-02-28 | 嘉兴学院 | Edge computing credible cooperative service method based on influence self-adaptive aggregation |
| CN111459670A (en) * | 2020-03-30 | 2020-07-28 | 中科边缘智慧信息科技（苏州）有限公司 | Method for performing cooperative processing at different levels of edge calculation |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7730186B2 (en) * | 2006-05-25 | 2010-06-01 | Fuji Xerox Co., Ltd. | Networked queuing system and method for distributed collborative clusters of services |
| CN111447512B (en) * | 2020-03-09 | 2021-10-26 | 重庆邮电大学 | Energy-saving method for edge cloud unloading |
-
2020
- 2020-10-09 CN CN202011075338.5A patent/CN112235385B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109379739A (en) * | 2018-09-28 | 2019-02-22 | 嘉兴学院 | A kind of credible cooperating service method of sea wireless Mesh netword |
| CN109587735A (en) * | 2018-11-12 | 2019-04-05 | 电子科技大学 | A kind of cooperation with service caching method for measuring mobile edge calculations based on time delay |
| CN110502328A (en) * | 2019-08-28 | 2019-11-26 | 嘉兴学院 | A kind of credible cotasking moving method of sea edge calculations |
| CN110851429A (en) * | 2019-11-21 | 2020-02-28 | 嘉兴学院 | Edge computing credible cooperative service method based on influence self-adaptive aggregation |
| CN111459670A (en) * | 2020-03-30 | 2020-07-28 | 中科边缘智慧信息科技（苏州）有限公司 | Method for performing cooperative processing at different levels of edge calculation |
Non-Patent Citations (3)
| Title |
|---|
| 边缘计算可信协同服务机制研究;戴亚盛;《《中国优秀硕士学位论文全文数据库（电子期刊）信息科技辑》》;20191215;全文 * |
| 边缘计算可信协同服务策略建模;乐光学等;《计算机研究与发展》;20200515;全文 * |
| 边缘计算多约束可信协同任务迁移策略;乐光学;《电信科学》;20191120;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112235385A (en) | 2021-01-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112235385B (en) | Offshore edge computing low-overhead cloud edge intelligent cooperative ally member discovery method | |
| CN109818786B (en) | Method for optimally selecting distributed multi-resource combined path capable of sensing application of cloud data center | |
| CN110851429B (en) | Edge computing credible cooperative service method based on influence self-adaptive aggregation | |
| CN110502328B (en) | Offshore edge computing trusted cooperative task migration method | |
| Wu et al. | Computation offloading method using stochastic games for software-defined-network-based multiagent mobile edge computing | |
| WO2023040022A1 (en) | Computing and network collaboration-based distributed computation offloading method in random network | |
| CN102299854B (en) | Opportunistic network environment-oriented multi-object routing decision making system | |
| Liu | Intelligent routing based on deep reinforcement learning in software-defined data-center networks | |
| CN101616074B (en) | Multicast routing optimization method based on quantum evolution | |
| Zhang et al. | Deploying SDN control in Internet of UAVs: Q-learning-based edge scheduling | |
| CN104244356A (en) | Orientation ant colony route optimization method based on evolution graph full route forecasting | |
| Qi et al. | Deep reinforcement learning based task scheduling in edge computing networks | |
| Yan et al. | Data offloading enabled by heterogeneous UAVs for IoT applications under uncertain environments | |
| Liu et al. | A distributed dependency-aware offloading scheme for vehicular edge computing based on policy gradient | |
| Xiao et al. | A novel task allocation for maximizing reliability considering fault-tolerant in VANET real time systems | |
| Luan et al. | A hierarchical hybrid subtask scheduling algorithm in UAV-assisted MEC emergency network | |
| Yu et al. | Resources sharing in 5G networks: Learning-enabled incentives and coalitional games | |
| CN110661566B (en) | Unmanned aerial vehicle cluster networking method and system adopting depth map embedding | |
| WO2024036909A1 (en) | Fair load unloading and migration method for edge service network | |
| Tang et al. | Optimization search strategy for task offloading from collaborative edge computing | |
| CN114666339B (en) | Edge unloading method and system based on noose set and storage medium | |
| Zhao et al. | Adaptive multi-UAV trajectory planning leveraging digital twin technology for urban IIoT applications | |
| CN114827933A (en) | Multipath routing method for wireless sensor network | |
| Lu et al. | Resource-efficient distributed deep neural networks empowered by intelligent software-defined networking | |
| CN103619047A (en) | Opportunistic routing method in multiple-concurrent-flow wireless mesh network |
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 |