CN110351352A - Edge calculations or mist calculate micro- computing cluster forming method based on incentive mechanism under environment - Google Patents
Edge calculations or mist calculate micro- computing cluster forming method based on incentive mechanism under environment Download PDFInfo
- Publication number
- CN110351352A CN110351352A CN201910592703.0A CN201910592703A CN110351352A CN 110351352 A CN110351352 A CN 110351352A CN 201910592703 A CN201910592703 A CN 201910592703A CN 110351352 A CN110351352 A CN 110351352A
- Authority
- CN
- China
- Prior art keywords
- equipment
- task
- micro
- expense
- computing cluster
- 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
Links
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
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/104—Peer-to-peer [P2P] networks
- H04L67/1044—Group management mechanisms
- H04L67/1046—Joining mechanisms
-
- 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
- H04L67/104—Peer-to-peer [P2P] networks
- H04L67/1044—Group management mechanisms
- H04L67/1048—Departure or maintenance mechanisms
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/06—Selective distribution of broadcast services, e.g. multimedia broadcast multicast service [MBMS]; Services to user groups; One-way selective calling services
- H04W4/08—User group management
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/70—Services for machine-to-machine communication [M2M] or machine type communication [MTC]
-
- 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
Abstract
The invention discloses a kind of edge calculations or mist to calculate micro- computing cluster forming method based on incentive mechanism under environment, by large number of equipment of collecting and cooperate, realizes to calculate and share with the communication resource, to reach the mobile task execution of low latency and high energy efficiency.The invention is firstly introduced into an effective incentive mechanism to promote the resource-sharing between equipment；Then Game with Coalitions is utilized, to promote to be performed in unison with task to mutual reciprocity and mutual benefit between equipment.Wherein, the cooperation wish of resourceful equipment is damaged in order to prevent, device packets are multiple micro- computing clusters by the incentive mechanism of proposition, in each micro- computing cluster, equipment can exchange mutually beneficial action by helping other equipment calculating or transformation task, so that the benefit of all tasks be made to become higher.The present invention designs the solution that micro- computing cluster is formed by centralized and distributed algorithm, and the high efficiency for further proving algorithm and the significant performance in terms of reducing energy and time delay expense.
Description
Technical field
The present invention relates to deep learning, edge calculations and distributed computing technology fields, and in particular to a kind of edge meter
It calculates or mist calculates micro- computing cluster forming method based on incentive mechanism under environment.
Background technique
With the technology that develops on an unprecedented scale (such as 5G) of mobile communication, it is contemplated that billions of equipment will be connected to internet, and operation is each
The mobile applications of kind various kinds, this has resulted in generating huge data in equipment end.The such huge data of processing
Amount, needs huge computing capability and energy supply.Since physical size and electricity limit, the computing capability of mobile device and
Energy supply is limited.In order to mitigate excessive calculating and energy demand, task unloading has been widely regarded as a kind of effective
Business processing mode.The basic thought of task unloading is to move to task from the mobile device of resource and energy deficiency to have foot
The calculate node of enough resource and electricity.Traditional solution is mobile cloud computing, and task is uploaded to powerful cloud and taken by it
Business device is handled, such as Amazon EC2.To the mobile applications with moderate delay tolerance, such as individual Yun Cun
Storage service, mobile cloud computing are very effective computation migration mode.But for some emerging augmented realities and online
The application such as interactive game, they usually have stringent delay requirement (usually within several ms).But mobile cloud meter
The communication delay of calculation reaches hundreds of milliseconds, considerably beyond above-mentioned delay requirement.Therefore effective method is to unload task
It is downloaded to edge device or Edge Server close to data generation end to be handled, reaches and effectively reduce time delay and energy conservation.But
It is that the edge device that mist calculates in environment is numerous, the processing capacity of individual equipment is limited, wherein effective mode is that collaboration is multiple
Equipment collaboration process task.But first, consider the isomerism of edge device, task cooperation how is carried out between equipment
Calculating and the communication resource can be effectively shared, second, how to carry out the selection of cooperative equipment so that the task execution of each equipment reaches
To mutual reciprocity and mutual benefit, it is contemplated that it is to generate energy consumption expense that each equipment cooperation other equipment, which carry out task execution, it is therefore desirable to be had
The incentive mechanism of effect goes the task execution for promoting equipment to realize mutual reciprocity and mutual benefit.
The problems such as in order to solve cloud computing and equipment isomery, a better method be equipment is divided into it is multiple based on connection
Micro- computing cluster of alliance's game theory, and a kind of effective incentive mechanism is proposed to promote the edge device in micro- computing cluster mutual
Mutually unload and be performed in unison with task.
Summary of the invention
The purpose of the present invention is to solve drawbacks described above in the prior art, a kind of edge calculations are provided or mist calculates
Micro- computing cluster forming method under environment based on incentive mechanism, this method are based on incentive mechanism, the association that each equipment is chosen
The task execution expense obtained as equipment will not be higher than locally executing or uploading the expense of Edge Server execution, i.e., each
Equipment can be obtained while assisting other equipment execution task and be assisted from the profitable task of other equipment.Based on alliance
Gaming characteristics, equipment will not change its cooperative equipment selection result after forming micro- computing cluster again, reach stable micro- calculating collection
Group structure.Equipment progress task cooperation reaches shared communication and computing resource in cluster, realizes low time delay and low energy edge
The task of system is handled.
The purpose of the present invention can be reached by adopting the following technical scheme that:
A kind of edge calculations or mist calculate micro- computing cluster forming method based on incentive mechanism under environment, and described is micro-
Computing cluster forming method includes:
S1, network detection step, each mobile terminal device broadcast data packet detection can nearby establish D2D therewith
(Device-to-Device) cooperative equipment connected, and establish corresponding cooperative equipment preference list and task execution mode
List；
S2, equipment compete step, and the mobile terminal device competition that any micro- computing cluster is not added, which becomes, detects micro- calculating
Its identity is stored in data packet MESG by the initiating equipment of cluster, the initiating equipment for winning competition, and to its optimal association
Make equipment and sends data packet MESG；If all devices have all been added to any micro- computing cluster, step S5 is jumped to；
S3, micro- computing cluster detection steps each receive the mobile terminal device of data packet MESG for its identity
It is sequentially stored in data packet MESG, and sends data packet MESG to its optimal cooperative equipment, when a mobile terminal device weight
Multiple connection receives data packet MESG, then the mobile terminal device knows a micro- computing cluster, and by the information of micro- computing cluster
It is broadcast to all equipment；
S4, cooperative equipment preference list update step, detect by a micro- computing cluster of wheel, all mobile terminal devices
From cooperative equipment preference list, the equipment identities of the micro- computing cluster received are identified and are rejected, to update cooperative equipment
Preference list goes to step S2；
According to the cooperative equipment and task execution mode list being assigned to, matching is corresponding for S5, each mobile terminal device
Task execution mode starts to execute the task of equipment of itself.
Wherein, above-mentioned mobile terminal device mobile phone, laptop, tablet computer and palm equipment for surfing the net, multimedia
Equipment, stream media equipment, mobile internet device (MID, mobile internet device), wearable device or other classes
The terminal device of type.
Further, the network detection step includes:
S11, the detection of each equipment i broadcast data packet can nearby establish the potential cooperative equipment of D2D connection therewith,
And the calculating and communication resource parameter of cooperative equipment are obtained, foundation can connect cooperative equipment set, include equipment i sheet in the set
Machine；
S12, each equipment i are according to task execution mode, calculating and communication resource parameter based on each of which cooperative equipment
The task expense under the cooperative equipment Feasible Mode is calculated, according to task expense from small to large by cooperative equipment identity
Sequence forms cooperative equipment preference list, and the task execution mode is including locally executing, D2D unloading executes, task is straight
It connects to pass to Edge Server execution and task is forwarded to Edge Server by D2D and execute, when equipment i is as the machine
Cooperative equipment, task execution mode are to locally execute and task is uploaded directly into Edge Server to execute, then equipment i will
The minimum mode of executive overhead is matched as optimal execution mode for the machine, and the task of other cooperative equipments is held in addition to equipment i
Row mode is that D2D unloading executes and task is forwarded to Edge Server execution by D2D, then equipment i will set for each cooperation
The standby matching lower mode of executive overhead is as optimal execution mode, and then equipment i is by the optimal execution mould of each cooperative equipment
Formula records to form corresponding task execution mode list.
Further, equipment competition step includes:
S21, all devices open timer, and the equipment that random setting time, most fast timing terminate wins competition as spy
The initiating equipment of micrometer computing cluster, and broadcast BUSY data packet notice other equipment；
Its identity is stored in MESG data packet by S22, initiating equipment, and to its optimal cooperative equipment, that is, is in
First equipment of its cooperative equipment preference list sends MESG data packet.
Further, micro- computing cluster detection steps include:
S31, the equipment for each receiving MESG data packet judge whether to receive the data packet for the first time, if it is first
Secondary reception, then jump to step S32, otherwise jumps to step S33；
Its identity is sequentially added in data packet by S32, the equipment for receiving MESG data packet, and by data packet
It is sent to its optimal cooperative equipment, then goes to step S31；
S33, the equipment of MESG data packet is received by the equipment identities mark of the micro- computing cluster formed in MESG data packet
Knowledge is broadcast to all devices.
The method that micro- computing cluster is formed in micro- computing cluster detection steps is based on Game with Coalitions thought, a micro- calculating
Cluster is an alliance.According to the information of MESG data packet, each equipment in a micro- computing cluster is available
Its cooperative equipment, the cooperative equipment are the equipment that identity is in the latter position of current device in MESG data packet.So one
Equipment in a micro- computing cluster can reach effective shared resource by task cooperative to reduce the target of task expense, be somebody's turn to do
The opposite number of the task execution expense summation of micro- computing cluster all devices is the benefit of the alliance.Each equipment is according to it
Cooperative equipment preference list selects task cooperative equipment, therefore each equipment task that finally matched cooperative equipment obtains
The task expense that executive overhead all selects oneself to obtain as cooperative equipment for the equipment is less than or equal to, here it is the present invention
The incentive mechanism of method.
Each MESG data packet is all transmitted from an equipment to its optimal cooperative equipment, therefore the formation of this step is all
Micro- computing cluster has finally all reached the structure of stable equilibrium, i.e. any appliance will not all change the association oneself being finally matched to
More preferably cooperative equipment is looked for as equipment, here it is the equilibrium states of the method for the present invention Game with Coalitions.
Further, the task expense of each equipment i include locally execute expense, D2D unloading executive overhead,
Task is uploaded directly into Edge Server processing expense and task is forwarded to the expense of Edge Server execution by D2D,
Wherein,
Described locally executes expense, i.e. equipment i locally execute the expense of task, and calculation formula is as follows:
Φi=IiPi (2)
In formula (1), θiIndicate that equipment i locally execute the expense of task,Expression task for
The weight of energy consumption and time delay, ΦiIndicate total computing resource of required by task, ciIndicate equipment i free time computing capability,It indicates
The energy consumption of equipment i unit computing resource；
In formula (2), PiIt is the processing density and O of required by taskiIt is the output data size of task, wherein task institute
Cpu cycle number needed for the processing density, that is, every input data needed；
The D2D unloads executive overhead, i.e. the task of the machine is unloaded to neighbouring equipment by D2D link by equipment i
Executive overhead, calculation formula are as follows:
In formula (3),Indicate that the task of the machine is unloaded at neighbouring equipment j by equipment i by D2D link
Reason, i.e. D2D unload executive overhead,WithRespectively indicate D2D link transmission of the task in equipment i and equipment j of equipment i
Time and energy expense,WithThe respectively indicating equipment i of the task handles time and the energy of consumption on equipment j
Expense；
In formula (4), IiIndicate that the task of equipment i inputs size, DijIndicate the D2D data transmission of equipment i to equipment j
Rate, OiIt is the output data size for the indicating equipment i of the task；
In formula (5), cjIndicate equipment j free time computing capability；
In formula (6),WithEquipment i is respectively indicated in the transmission energy consumption of D2D network and receives energy consumption；
In formula (7),Indicate the energy consumption of equipment j unit computing resource；
Task is uploaded directly into Edge Server processing expense by the equipment i, and calculation formula is as follows:
Ti c2=Φi/Ki (11)
In formula (8), θi cIndicate that task is uploaded directly into the overhead of Edge Server processing, T by equipment ii c1With
Indicate that task is uploaded to the transmission time of Edge Server and the expense of transmission energy consumption, T by equipment ii c2Indicate the task of equipment i
The time overhead handled on Edge Server；
In formula (9),WithEquipment i is respectively indicated in the uploading rate and downloading data rate of cellular network；
In formula (10),WithEquipment i is respectively indicated in the transmission energy consumption of cellular network and receives energy consumption；
In formula (11), KiIndicate the computing capability of virtual machine in the Edge Server of the task of running equipment i；
Task is forwarded to the expense of Edge Server execution by the equipment i by D2D, and calculation formula is as follows:
Ti c2=Φi/Ki (15)
In formula (12),Indicate that task is forwarded to Edge Server by the equipment j of D2D connection and executed by equipment i
Overhead,WithIndicate the time of D2D transmission and energy expense between equipment i and equipment j,WithIt indicates to turn
Send out time and the energy expense of the cellular transmission of equipment j, Ti c2Indicate the time overhead of the virtual machine processing task i of equipment j.
In formula (13), DjiIndicate the D2D message transmission rate of equipment j to equipment i；
In formula (14),WithEquipment j is respectively indicated in the uploading rate and downloading data rate of cellular network；
In formula (16),WithEquipment j is respectively indicated in the transmission energy consumption of D2D network and receives energy consumption；
In formula (17),WithEquipment j is respectively indicated in the transmission energy consumption of cellular network and receives energy consumption.
The present invention has the following advantages and effects with respect to the prior art:
1, mist disclosed by the invention calculates the scheme that micro- computing cluster collaboration processing task is formed in environment, considers simultaneously
Computing resource and the communication resource it is shared, enable the task to complete in equipment end, cause too long without uploading to cloud
Communication delay.The equipment for transferring various isomeries simultaneously participates in cooperative cooperating, and it is sharp to effectively improve the resource that mist calculates in environment
With rate.
2, the scheme that micro- computing cluster collaboration processing task is formed in mist calculating environment disclosed by the invention is rich based on alliance
It draws incentive mechanism, equipment selects optimal teamworker when marginal end carries out task immigration as far as possible, so that its selection scheme
Will not than locally execute or directly upload Edge Server execution effect it is poor, it is rich to promote each equipment to participate in alliance
In playing chess, better task execution scheme is obtained.
3, comparison armamentarium locally executes task, armamentarium Edge Server executes task, any pairing, greed
With reciprocity scheme, method disclosed by the invention is significantly improved in terms of reducing time overhead and energy consumption.
4, algorithm disclosed by the invention has lower time complexity.
Detailed description of the invention
Fig. 1 is that mist disclosed in the present invention calculates micro- computing cluster forming method process based on incentive mechanism under environment
Figure；
Fig. 2 is micro- computing cluster (A, C, D, B) schematic diagram；
Fig. 3 is the embodiment of the present invention in all micro- computing cluster schematic diagrames to row in 7 equipment；
Fig. 4 is the time delay overhead performance gain diagram under the change of D2D join domain；
Fig. 5 is the energy expense performance gain figure under the change of D2D join domain；
Fig. 6 is the performance gain figure that time delay and energy combine under the change of D2D join domain；
Fig. 7 is the method for the present invention mean iterative number of time variation diagram in the case that number of devices increases；
Fig. 8 is compatible incentives collaboration ratio chart in the case of number of devices increases；
Fig. 9 is overhead figure.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
Member's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Embodiment
Micro- computing cluster based on incentive mechanism under environment is calculated present embodiment discloses a kind of edge calculations or mist to be formed
Method, process are as shown in Figure 1.This method is based on incentive mechanism and Game with Coalitions, forms micro- computing cluster by equipment, collects
Equipment is cooperateed in group, realizes the task processing of low time delay and low energy limbic system.The simulated experimental environments of the present embodiment
Parameter is as shown in table 1 below, the random generating device in one 500 × 500 region.
1. simulated experimental environments parameter list of table
To illustrate the forming process of micro- computing cluster for wherein 7 equipment A, B, C, D, E, F, G, the specific steps are as follows:
S1, network detection step, each device broadcasts Packet probing can nearby establish the cooperation of D2D connection therewith
Equipment and their calculating and communication resource parameter.For example the D2D of equipment A can connect cooperative equipment collection and be combined into { A, B, C }, B's
D2D can connect cooperative equipment collection and be combined into { A, B, D }, and the D2D of C can connect cooperative equipment collection and be combined into { C, D, E }, and the D2D of D can Lian Xiezuo
Cluster tool is { B, C, D }, and the D2D of E can connect cooperative equipment collection and be combined into { C, E, F }, and the D2D of F can connect cooperative equipment collection and be combined into
{ E, F, G }, the D2D of A can connect cooperative equipment collection and be combined into { E, F, G }.
Each equipment is according to four kinds of task execution modes, calculating and communication resource parameter based on each of which cooperative equipment
Calculate the task expense under the cooperative equipment Feasible Mode.Cooperative equipment identity is arranged from small to large according to task expense
Sequence forms task execution mode list, as shown in table 2.
2. cooperative equipment preference list of table
S2, equipment competition, the equipment competition that any micro- computing cluster is not added are set as the initiation for detecting micro- computing cluster
Standby, its identity is stored in data packet MESG by the initiating equipment for winning competition, and sends number to its optimal cooperative equipment
According to packet MESG, specific step is as follows；
All devices open timer, random setting time.Assuming that be that the equipment that most fast timing terminates is won competing by equipment A
It strives as the initiating equipment for detecting micro- computing cluster, it, which broadcasts BUSY data packet notice other equipment, will initiate the micro- calculating of wheel
Cluster detection.Equipment A generates MESG data packet, and its identity is added in data packet MESG, i.e. data packet MESG=
(A).As known from Table 2, the optimal cooperative equipment of equipment A is C, therefore equipment A sends data packet MESG to equipment C；
S3, micro- computing cluster detection steps may determine that micro- meter by the way that whether an equipment repeats received data packet MESG
The formation of cluster is calculated, specific step is as follows；
Equipment C receives data packet MESG for the first time, its identity is sequentially added in data packet by it, then at this time
Data packet MESG=(A, C).Then data packet MESG is sent to its optimal association according to its cooperative equipment preference list by equipment C
Make equipment D.Similarly, equipment D receives data packet MESG for the first time, after identity is sequentially added to data packet MESG,
Data packet MESG=(A, C, D), then data packet MESG is sent to its optimal cooperative equipment B by equipment D.Equipment B is also first
It is secondary to receive data packet MESG, after identity is sequentially added to data packet MESG, data packet MESG=(A, C, D, B),
Data packet MESG is sent to its optimal cooperative equipment A again by equipment B；
After equipment A receives data packet MESG, equipment A is second of reception at this time, and identity has been stored in
Data packet MESG illustrates that a micro- computing cluster=(A, C, D, B) has been formed, as shown in Figure 2.At (A, C, D, B), this is micro-
In computing cluster, the cooperative equipment of A is C, and the cooperative equipment of C is D, and the cooperative equipment of the cooperative equipment B, B of D are A.
Equipment A identifies the equipment identities of the micro- computing cluster formed in data packet MESG, i.e., (A, C, D, B) is broadcast to
All devices.Therefore the micro- computing cluster detection of a wheel terminates.
S4, cooperative equipment preference list update step are specific as follows:
By the micro- computing cluster detection of a wheel, all devices receive the micro- computing cluster (A, C, D, B) formed
Afterwards, A, C, D, B are rejected from cooperative equipment preference list.So, the updated cooperative equipment preference list of each equipment point
It is notC=(E),E=(F, E), F=(E, G, F), G=(E, F, G).
Return step S2, until all devices are all added to any micro- computing cluster.
Finally, all micro- computing clusters formed in the present embodiment are as shown in Figure 3.That is (A, C, D, B), (E, F) and (G),
Wherein the cooperative equipment of E is F, and the cooperative equipment of F is E, and the cooperative equipment of G is oneself.
S5, each equipment match corresponding task execution according to the cooperative equipment and task execution mode list that are assigned to
Mode starts to execute the task of oneself.
It can be obtained from embodiment result, be based on incentive mechanism, the task execution that the cooperative equipment that each equipment is chosen obtains
Expense will not be higher than locally executing or uploading the expense of Edge Server execution, i.e., each equipment is held in assistance other equipment
It can obtain while row task and be assisted from the profitable task of other equipment.Based on Game with Coalitions characteristic, equipment forms micro- meter
Its cooperative equipment selection result will not be changed again after calculating cluster, reach stable micro- computing cluster structure.
Micro- computing cluster based on incentive mechanism under environment is calculated to assess a kind of edge calculations disclosed by the invention or mist
Forming method locally executes task, whole using armamentarium in the performance for reducing time overhead and energy expense, comparison scheme
Device end server executes task, any pairing and greed pairing.In order to embody algorithm specifically lower time complexity, I
Compare the number of iterations in the method for the present invention and the upper bound.
Finally, equipment is only examined when selecting optimal teamworker in view of method disclosed by the invention is based on incentive mechanism
The expense for considering itself task cannot be higher than locally executing or uploading the expense of Edge Server execution, complete without considering
The reduction of office's expense.I.e. equipment, which only takes into account, is not lost number one, does not account for global benefit.Therefore we incite somebody to action this
Scheme is compared with a maximized figure matching scheme of consideration interests of the whole, both assessments cost of device performance.
Fig. 4 to Fig. 6 is illustrated constantly increase with D2D join domain in the case where, what the present invention disclosed that scheme obtains opens
Pin matches strategy, random pair strategy with greed, locally executes the case where strategy and Edge Server implementation strategy.It can see
It arrives, compares above four kinds of strategies, the scheme disclosed by the invention task expense that is averaged significantly is declined.And as D2D connects
Range increase is connect, the performance gain of task expense decline has slightly smaller increase.This is because as D2D range increases, more
Equipment participates in Game with Coalitions, so that each equipment has an opportunity to find better teamworker.
Fig. 7 illustrates the increase with number of devices, and the mean iterative number of time of the method for the present invention is well below upper bound N2, N
For number of devices, illustrate that the present invention discloses the high efficiency of the algorithm of scheme.
Fig. 8 and Fig. 9 introduces the figure matching algorithm for innings minimizing overhead of demanding perfection, Fig. 8 illustrate the present invention disclose scheme,
The compatible incentives of greed pairing, random pair and figure matching algorithm cooperate with ratio, i.e., the equipment being improved by Cooperation benefit
Percentage.Mentioning for equipment itself benefit is all only considered since the present invention discloses scheme, greed pairing and random pair
Height does not consider the maximization of global benefit, therefore the present invention discloses the total of scheme, greed pairing and random pair strategy in Fig. 9
Task expense is bigger than the general assignment expense of figure matching algorithm.But figure matching algorithm compromises the interests of equipment component, from Fig. 8
In as can be seen that the equipment ratio that is improved of itself benefit more than the present invention disclose scheme in figure matching algorithm, greed is matched
Want low with random pair.
The above embodiment is a preferred embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment
Limitation, it is other it is any without departing from the spirit and principles of the present invention made by change, modification, substitution, combination, letter
Change, should be equivalent substitute mode, be included within the scope of the present invention.
Claims (5)
1. a kind of edge calculations or mist calculate micro- computing cluster forming method based on incentive mechanism under environment, which is characterized in that
Micro- computing cluster forming method includes:
S1, network detection step, each mobile terminal device broadcast data packet detection can nearby establish the association of D2D connection therewith
Make equipment, and establishes corresponding cooperative equipment preference list and task execution mode list；
S2, equipment compete step, and the mobile terminal device competition that any micro- computing cluster is not added, which becomes, detects micro- computing cluster
Initiating equipment, its identity is stored in data packet MESG by the initiating equipment for winning competition, and sets to its optimal cooperation
Preparation send data packet MESG；If all mobile terminal devices have all been added to any micro- computing cluster, step S5 is jumped to；
S3, micro- computing cluster detection steps each receive the mobile terminal device of data packet MESG for its identity sequentially
It is stored in data packet MESG, and sends data packet MESG to its optimal cooperative equipment, when a mobile terminal device repeats to connect
Data packet MESG is received, then the mobile terminal device knows a micro- computing cluster, and the information of micro- computing cluster is broadcasted
To all equipment；
S4, cooperative equipment preference list update step, detect by a micro- computing cluster of wheel, and all mobile terminal devices are from cooperation
In device preference list, the equipment identities of the micro- computing cluster received are identified and are rejected, to update cooperative equipment preference column
Table goes to step S2；
S5, each mobile terminal device match corresponding task according to the cooperative equipment and task execution mode list that are assigned to
Execution pattern starts to execute the task of equipment of itself.
2. edge calculations according to claim 1 or mist calculate micro- computing cluster side of being formed based on incentive mechanism under environment
Method, which is characterized in that the network detection step includes:
S11, the detection of each equipment i broadcast data packet nearby can establish the potential cooperative equipment of D2D connection therewith, and obtain
The calculating and communication resource parameter of cooperative equipment, foundation can connect cooperative equipment set, include equipment i the machine in the set；
According to task execution mode, calculating and communication resource parameter based on each of which cooperative equipment are calculated by S12, each equipment i
Task expense under the cooperative equipment Feasible Mode, according to task expense from small to large by cooperative equipment identity sequence shape
At cooperative equipment preference list, the task execution mode is including locally executing, D2D unloading executes, directly uploads task
It is executed to Edge Server and task is forwarded to Edge Server by D2D and executed, when equipment i is set as the cooperation of the machine
Standby, task execution mode is to locally execute and task is uploaded directly into Edge Server to execute, then equipment i will be the machine
With the minimum mode of executive overhead as optimal execution mode, the task execution mode of other cooperative equipments is in addition to equipment i
D2D unloading executes and task is forwarded to Edge Server by D2D and executes, then equipment i will hold for the matching of each cooperative equipment
The lower mode of row expense is as optimal execution mode, and then equipment i is by the optimal execution mode record shape of each cooperative equipment
At corresponding task execution mode list.
3. according to claim a kind of edge calculations or mist calculate micro- computing cluster based on incentive mechanism under environment and are formed
Method, which is characterized in that the equipment competes step and includes:
S21, all devices open timer, and it is micro- as detecting that the equipment that random setting time, most fast timing terminate wins competition
The initiating equipment of computing cluster, and broadcast BUSY data packet notice other equipment；
Its identity is stored in MESG data packet by S22, initiating equipment, and to its optimal cooperative equipment, i.e., in its association
Make first equipment of device preference list, sends MESG data packet.
4. edge calculations according to claim 1 or mist calculate micro- computing cluster side of being formed based on incentive mechanism under environment
Method, which is characterized in that micro- computing cluster detection steps include:
S31, the equipment for each receiving MESG data packet judge whether to receive the data packet for the first time, if it is connecing for the first time
It receives, then jumps to step S32, otherwise jump to step S33；
Its identity is sequentially added in data packet, and delivers a packet to by S32, the equipment for receiving MESG data packet
Its optimal cooperative equipment, then goes to step S31；
S33, the equipment for receiving MESG data packet identify the equipment identities of the micro- computing cluster formed in MESG data packet wide
It broadcasts to all devices.
5. edge calculations according to claim 2 or mist calculate micro- computing cluster side of being formed based on incentive mechanism under environment
Method, which is characterized in that the task expense of each equipment i includes locally executing expense, D2D unloading executive overhead, appointing
Business is uploaded directly into Edge Server processing expense and task is forwarded to the expense of Edge Server execution by D2D, wherein
Described locally executes expense, i.e. equipment i locally execute the expense of task, and calculation formula is as follows:
Φi=IiPi (2)
In formula (1), θiIndicate that equipment i locally execute the expense of task,Expression task for energy consumption and
The weight of time delay, ΦiIndicate total computing resource of required by task, ciIndicate equipment i free time computing capability,Indicate that equipment i is mono-
The energy consumption of position computing resource；
In formula (2), PiIt is the processing density and O of required by taskiIt is the output data size of task, wherein the place of required by task
Cpu cycle number needed for managing density, that is, every input data；
The D2D unloads executive overhead, i.e. the task of the machine is unloaded to neighbouring equipment by D2D link and executed by equipment i
Expense, calculation formula are as follows:
In formula (3),Indicate that the task of the machine is unloaded to neighbouring equipment j by D2D link and handled by equipment i, i.e.,
D2D unloads executive overhead,WithRespectively indicate time of the task in equipment i and the D2D link transmission of equipment j of equipment i
And energy expense,WithThe respectively indicating equipment i of the task handles time and the energy expense of consumption on equipment j；
In formula (4), IiIndicate that the task of equipment i inputs size, DijIndicate the D2D message transmission rate of equipment i to equipment j, Oi
It is the output data size for the indicating equipment i of the task；
In formula (5), cjIndicate equipment j free time computing capability；
In formula (6),WithEquipment i is respectively indicated in the transmission energy consumption of D2D network and receives energy consumption；
In formula (7),Indicate the energy consumption of equipment j unit computing resource；
Task is uploaded directly into Edge Server processing expense by the equipment i, and calculation formula is as follows:
Ti c2=Φi/Ki (11)
In formula (8),Indicate that task is uploaded directly into the overhead of Edge Server processing, T by equipment ii c1WithIt indicates
Task is uploaded to the transmission time of Edge Server and the expense of transmission energy consumption, T by equipment ii c2Indicate the task of equipment i on side
The time overhead handled on edge server；
In formula (9),WithEquipment i is respectively indicated in the uploading rate and downloading data rate of cellular network；
In formula (10),WithEquipment i is respectively indicated in the transmission energy consumption of cellular network and receives energy consumption；
In formula (11), KiIndicate the computing capability of virtual machine in the Edge Server of the task of running equipment i；
Task is forwarded to the expense of Edge Server execution by the equipment i by D2D, and calculation formula is as follows:
Ti c2=Φi/Ki (15)
In formula (12),Indicate that task is forwarded to always opening for Edge Server execution by the equipment j of D2D connection by equipment i
Pin,WithIndicate the time of D2D transmission and energy expense between equipment i and equipment j,WithIndicate forwarding device j
Cellular transmission time and energy expense, Ti c2Indicate the time overhead of the virtual machine processing task i of equipment j；
In formula (13), DjiIndicate the D2D message transmission rate of equipment j to equipment i；
In formula (14),WithEquipment j is respectively indicated in the uploading rate and downloading data rate of cellular network；
In formula (16),WithEquipment j is respectively indicated in the transmission energy consumption of D2D network and receives energy consumption；
In formula (17),WithEquipment j is respectively indicated in the transmission energy consumption of cellular network and receives energy consumption.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910592703.0A CN110351352B (en) | 2019-07-03 | 2019-07-03 | Micro-computing cluster forming method based on incentive mechanism in edge computing or fog computing environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910592703.0A CN110351352B (en) | 2019-07-03 | 2019-07-03 | Micro-computing cluster forming method based on incentive mechanism in edge computing or fog computing environment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110351352A true CN110351352A (en) | 2019-10-18 |
| CN110351352B CN110351352B (en) | 2022-04-12 |
Family
ID=68177475
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910592703.0A Active CN110351352B (en) | 2019-07-03 | 2019-07-03 | Micro-computing cluster forming method based on incentive mechanism in edge computing or fog computing environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110351352B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111163178A (en) * | 2020-01-10 | 2020-05-15 | 中国地质大学（武汉） | Game theory-based service deployment and task unloading method in edge computing |
| CN111277669A (en) * | 2020-03-03 | 2020-06-12 | 重庆邮电大学 | Internet of things equipment collaborative intention intelligent sensing method |
| CN112235387A (en) * | 2020-10-10 | 2021-01-15 | 华北电力大学（保定） | Multi-node cooperative computing unloading method based on energy consumption minimization |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104899404A (en) * | 2015-07-06 | 2015-09-09 | 广州特种机电设备检测研究院 | Simulation cloud platform and implementation method |
| CN105610944A (en) * | 2015-12-29 | 2016-05-25 | 北京物联远信息技术有限公司 | IOT-oriented fog computing architecture |
| US20180159745A1 (en) * | 2016-12-06 | 2018-06-07 | Cisco Technology, Inc. | Orchestration of cloud and fog interactions |
| CN109379727A (en) * | 2018-10-16 | 2019-02-22 | 重庆邮电大学 | Task distribution formula unloading in car networking based on MEC carries into execution a plan with cooperating |
| CN109547541A (en) * | 2018-11-12 | 2019-03-29 | 安徽师范大学 | Mist calculates the node low overhead collaboration method under environment based on filtering and distribution mechanism |
| CN109783233A (en) * | 2018-12-24 | 2019-05-21 | 中山大学 | A method of task unloading in mobile edge calculations is provided |
-
2019
- 2019-07-03 CN CN201910592703.0A patent/CN110351352B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104899404A (en) * | 2015-07-06 | 2015-09-09 | 广州特种机电设备检测研究院 | Simulation cloud platform and implementation method |
| CN105610944A (en) * | 2015-12-29 | 2016-05-25 | 北京物联远信息技术有限公司 | IOT-oriented fog computing architecture |
| US20180159745A1 (en) * | 2016-12-06 | 2018-06-07 | Cisco Technology, Inc. | Orchestration of cloud and fog interactions |
| CN109379727A (en) * | 2018-10-16 | 2019-02-22 | 重庆邮电大学 | Task distribution formula unloading in car networking based on MEC carries into execution a plan with cooperating |
| CN109547541A (en) * | 2018-11-12 | 2019-03-29 | 安徽师范大学 | Mist calculates the node low overhead collaboration method under environment based on filtering and distribution mechanism |
| CN109783233A (en) * | 2018-12-24 | 2019-05-21 | 中山大学 | A method of task unloading in mobile edge calculations is provided |
Non-Patent Citations (2)
| Title |
|---|
| SIQI LUO: ""Dewing in Fog: Incentive-Aware Micro Computing Cluster Formation for Fog Computing"", 《2018 IEEE 24TH INTERNATIONAL CONFERENCE ON PARALLEL AND DISTRIBUTED SYSTEMS (ICPADS)》 * |
| 李恩: ""边缘智能:边缘计算驱动的深度学习加速技术"", 《自动化博览》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111163178A (en) * | 2020-01-10 | 2020-05-15 | 中国地质大学（武汉） | Game theory-based service deployment and task unloading method in edge computing |
| CN111163178B (en) * | 2020-01-10 | 2021-03-30 | 中国地质大学（武汉） | Game theory-based service deployment and task unloading method in edge computing |
| CN111277669A (en) * | 2020-03-03 | 2020-06-12 | 重庆邮电大学 | Internet of things equipment collaborative intention intelligent sensing method |
| CN111277669B (en) * | 2020-03-03 | 2022-04-19 | 重庆邮电大学 | Internet of things equipment collaborative intention intelligent sensing method |
| CN112235387A (en) * | 2020-10-10 | 2021-01-15 | 华北电力大学（保定） | Multi-node cooperative computing unloading method based on energy consumption minimization |
| CN112235387B (en) * | 2020-10-10 | 2022-12-13 | 华北电力大学（保定） | Multi-node cooperative computing unloading method based on energy consumption minimization |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110351352B (en) | 2022-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Socially aware dynamic computation offloading scheme for fog computing system with energy harvesting devices | |
| Yang et al. | POMT: Paired offloading of multiple tasks in heterogeneous fog networks | |
| CN110351352A (en) | Edge calculations or mist calculate micro- computing cluster forming method based on incentive mechanism under environment | |
| Luo et al. | Incentive-aware micro computing cluster formation for cooperative fog computing | |
| Hu et al. | Heterogeneous edge offloading with incomplete information: A minority game approach | |
| Wang et al. | QoS multicast routing protocol oriented to cognitive network using competitive coevolutionary algorithm | |
| Xu et al. | Trust based incentive scheme to allocate big data tasks with mobile social cloud | |
| Chatzopoulos et al. | OPENRP: A reputation middleware for opportunistic crowd computing | |
| Huang et al. | Multi-agent reinforcement learning for cost-aware collaborative task execution in energy-harvesting D2D networks | |
| Chatzopoulos et al. | Have you asked your neighbors? A hidden market approach for device-to-device offloading | |
| Na et al. | An evolutionary game approach on IoT service selection for balancing device energy consumption | |
| Xie et al. | Dynamic d2d multihop offloading in multi-access edge computing from the perspective of learning theory in games | |
| Liu et al. | Population game based energy and time aware task offloading for large amounts of competing users | |
| Liu et al. | Virtual reality streaming in blockchain-enabled fog radio access networks | |
| Morillo et al. | Ensuring the performance and scalability of peer-to-peer distributed virtual environments | |
| Ahmed et al. | Improving online gaming experience using location awareness and interaction details | |
| Bastani et al. | A social node model for realising information dissemination strategies in delay tolerant networks | |
| Hu et al. | P2P consistency support for large-scale interactive applications | |
| Yang et al. | Non-cooperative game approach for task offloading in edge clouds | |
| Luo et al. | Dewing in fog: Incentive-aware micro computing cluster formation for fog computing | |
| Zhang et al. | A case for coalitions in data swarming systems | |
| CN110061862B (en) | Fairness-based distributed multi-task crowd sensing method in dense network | |
| He et al. | A leader–follower controlled Markov stopping game for delay tolerant and opportunistic resource sharing networks | |
| Vik et al. | Dynamic group membership management for distributed interactive applications | |
| Del Val et al. | Emergence of cooperation through structural changes and incentives in service-oriented mas |
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 |