CN116668447B - Edge computing task unloading method based on improved self-learning weight - Google Patents
Edge computing task unloading method based on improved self-learning weight Download PDFInfo
- Publication number
- CN116668447B CN116668447B CN202310956562.2A CN202310956562A CN116668447B CN 116668447 B CN116668447 B CN 116668447B CN 202310956562 A CN202310956562 A CN 202310956562A CN 116668447 B CN116668447 B CN 116668447B
- Authority
- CN
- China
- Prior art keywords
- edge
- task
- unloading
- resources
- equipment
- 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
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000004891 communication Methods 0.000 claims abstract description 102
- 230000005540 biological transmission Effects 0.000 claims abstract description 49
- 238000012544 monitoring process Methods 0.000 claims abstract description 8
- 238000012546 transfer Methods 0.000 claims description 62
- 238000012545 processing Methods 0.000 claims description 17
- 239000000654 additive Substances 0.000 claims description 12
- 230000000996 additive effect Effects 0.000 claims description 12
- 230000001934 delay Effects 0.000 claims description 6
- 239000000463 material Substances 0.000 claims description 6
- 230000002596 correlated effect Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 238000005265 energy consumption Methods 0.000 description 4
- 238000005457 optimization Methods 0.000 description 2
- 238000013468 resource allocation Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
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
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1004—Server selection for load balancing
- H04L67/1008—Server selection for load balancing based on parameters of servers, e.g. available memory or workload
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
- H04L45/121—Shortest path evaluation by minimising delays
-
- 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
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1029—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers using data related to the state of servers by a load balancer
-
- 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
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The invention discloses an edge computing task unloading method based on improved self-learning weight, which comprises the following steps: collecting system resource data of the edge equipment through a Prometaus monitoring tool; calculating global share of four resources according to the idle quantity of the resources and the total quantity of the resources, and initializing weights of the four resources; selecting the edge equipment with the highest overall score and the transmission path with the smallest communication delay to carry out task unloading; updating the idle quantity of resources, and entering into the unloading decision of the next round of tasks until all tasks are unloaded; according to the method and the device, four resources of the edge devices are jointly optimized, the candidate edge devices are scored according to the resource load condition, the difference of unloading decisions among the edge devices is quantized, and certain types of computing tasks with large resource demands are unloaded to the edge devices with more idle resources, so that the method and the device can adapt to the scheduling demands of the tasks in different scenes, the task unloading efficiency is greatly improved, and the method and the device are high in reliability and good in applicability.
Description
Technical Field
The invention relates to the field of computer edge calculation, in particular to an edge calculation task unloading method based on improved self-learning weight.
Background
The basic idea of edge computing is to transfer the computing tasks generated on the mobile device from the original offload to the cloud to the offload to the network edge, thereby meeting the low latency requirements of computationally intensive applications. The unloading of the computing task is a key research problem in edge computing, namely whether the computing task should be locally executed or unloaded to an edge node or a cloud end, wherein edge equipment receives data collected by a monitoring tool in a multithreading mode, and Prometaus is an open-source service monitoring system and a time sequence database; in the current hot unloading system, the edge equipment is provided with a multi-core CPU, the disk space and the bandwidth are sufficient, retransmission can be carried out after task transmission fails, tasks generated from the terminal can be thermally migrated to the edge equipment for processing, and different task unloading schemes have great influence on the time delay of task completion and equipment energy consumption. Most of the prior art schemes take time delay as an optimization target, the time delay is divided into transmission time delay and calculation time delay, and decision is made on task unloading by improving the calculation rate of edge equipment and selecting different task transmission links, so that the transmission time delay and the calculation time delay are reduced.
The invention discloses a task unloading method based on mobile edge computing, which is disclosed by a patent document with the publication number of CN111148155A, and is characterized in that mobile equipment is divided into clusters based on graph division, then task unloading decision problems among multiple users are converted into multi-user game problems, and the game results meet Nash equilibrium, so that computing and communication loads of a cloud core network can be reduced, redundant computing resources, mobile equipment and internet of things (IoT) equipment at the edge of a network can be fully utilized, task completion delay and energy consumption of edge computing task scheduling can be reduced, but only time delay is considered as an optimization target, the time delay is divided into transmission time delay and computing time delay, the influence of other resources on task unloading efficiency is not considered to be large.
The method comprises the steps of providing a task unloading method with a publication number of CN114564304A and a name of edge calculation, calculating the total unloading time delay and transmission energy consumption of a current task by constructing a network model comprising a plurality of mobile devices and a plurality of MEC servers and a time delay calculation model, constructing scene satisfaction models of different application scenes, and constructing an objective function of the total cost of unloading the current task by combining a penalty function; according to the method, the optimal position of task unloading is solved, the unloading of the current task is completed, energy consumption, time delay and satisfaction can be optimized, however, the situation that the resource load condition in the task unloading of each round is different is not considered, a certain resource is excessively allocated to cause unbalance of the whole cluster resource allocation, and the resource utilization rate is low.
Disclosure of Invention
In order to solve the technical problems, the invention adopts a technical scheme that: an edge computing task offloading method based on improved self-learning weights, the method comprising:
s100: by passing throughThe monitoring tool collects system resource data of the edge equipment and the minimum resource demand of the task;
the system resource data includes: equipment resources, resource idle quantity, resource total amount, uploading rate of edge equipment and unit data calculation time;
s200: calculating global shares of four resources of a central processing unit, a memory, a bandwidth and a magnetic disk according to the idle amount of the resources and the total amount of the resources in the system resource data, and initializing weights of the four resources;
s300: creating a candidate device set, and adding edge devices with the resource idle quantity being greater than or equal to the minimum resource demand quantity of the task into the candidate device set;
s400: calculating overall scores of edge devices in the candidate device set according to the weights of the four resources, creating a score queue, sorting the overall scores in a descending order, sequentially adding the overall scores into the score queue, and selecting edge devices with highest overall scores in the score queue as optimal unloading devices;
s500: calculating communication time delays of different task unloading transmission paths, and selecting a transmission path with the smallest communication time delay to unload the task into the optimal unloading equipment;
s600: updating the resource idle quantity of the optimal unloading device, and entering the unloading decision of the next round of tasks until all tasks are unloaded;
the S200 includes:
s210, calculating total idle quantity of four resources in all edge devices；
The calculation formula of the total idle quantity is as follows:
；
wherein ,resource free amount for single edge device, +.>As a single edge device,for device resources including CPU resources, memory resources, bandwidth resources, disk resources, +.>For an unloading task;
s220, calculating the total sum of four resources in all edge devices；
The sum is calculated by the following formula:
；
wherein ,the sum of resources for a single edge device;
s230, calculating idle proportion of four resources in all edge devices;
the calculation formula of the idle proportion is as follows:
；
s240, calculating global shares of four resources in all edge devices；
The calculation formula of the global share is as follows:
；
wherein ,representing the current load conditions of the four resources, wherein the global share is positively correlated with the load conditions;
s250, initializing weights of four resources in all edge devices；
The formula for initializing the weights of the four resources is as follows:
；
wherein ,is the sum of the global shares of the four resources;
the S400 includes:
s410 computing taskResource score for each edge device in the candidate device set +.>；
The calculation formula of the resource score is as follows:
；
wherein ,；
s420, calculating taskComputing delay caused by individual edge devices in the candidate device set +.>；
The calculation formula of the calculation time delay caused by each edge device is as follows:；
wherein ,for tasks->Data volume->Time required for calculating unit data for the respective edge devices +.>；
S430, computing taskDelay score +.>；
The calculation formula of the time delay score is as follows:
；
wherein ,Maximum computation delay for edge devices, +.>Minimum computation delay for the edge device;
s440, computing taskOverall score +.>Creating a scoring queue, and adding the overall scores into the scoring queue in a descending order;
the calculation formula of the overall score is as follows:
；
s450, selecting the edge equipment with the highest overall score from the score queue as the optimal unloading equipment.
Further, the device resourceIncluding central processing unit resources, memory resources, bandwidth resources, disk resources, expressed as:
；
wherein ,refers to a Central Processing Unit (CPU)>Refers to memory, which is->Refers to bandwidth, & gt>Refers to disk data;
the resource idle quantity comprises a central processing unit idle quantity, a memory idle quantity, a bandwidth idle quantity and a disk idle quantity;
the total amount of resources comprises a total amount of a central processing unit, a total amount of memory, a total amount of bandwidth and a total amount of magnetic disk.
Further, the step S300 includes:
s310, creating a candidate device set;
s320, judging the resource idle quantity of the edge equipmentWhether or not it is greater than or equal to task->Minimum demand for four resources +.>If yes, adding the edge equipment into the candidate equipment set; if not, the edge device can not complete the task +.>Is unable to perform task offloading; minimum demand of the four resources +.>Representing the minimum demand of task k on the CPU resource, memory resource, bandwidth resource and disk resource in the edge equipment;
s330, calculating the signal-to-noise ratio of the edge equipment in the candidate equipment set；
The calculation formula of the signal to noise ratio is as follows:
；
wherein ,for edge devices->Upload rate of->For the distance between the end device and the edge device, < >>For the attenuation index>Is the additive white gaussian noise variance.
Further, the S500 includes:
s510, creating a set of edge devices capable of direct communicationSet of non-directly communicable edge devices>；
S520, judging the signal-to-noise ratio of the edge equipmentWhether or not it is greater than or equal to the signal-to-noise threshold +.>If yes, adding the edge device into the set of directly communicable edge devices +.>In (a) and (b); if not, adding the edge device into the non-directly communicable edge device set +.>In (a) and (b); wherein the signal-to-noise ratio threshold +.>5;
s530, judging whether the optimal unloading equipment belongs to the set of directly communicable edge equipmentIf yes, go to S540, if no, go to S550;
s540, if the optimal unloading device belongs to the set of edge devices capable of direct communicationCalculating the communication time delay of two task unloading transmission paths and selecting a transmission path with smaller communication time delay as an optimal task unloading transmission path if the optimal unloading equipment is reachable;
s550, if the optimal unloading device belongs to the set of non-directly communicable edge devicesI.e. indicating that the optimal unloading device is not reachable, selecting the transit device first, task ∈>Firstly unloading the material into transfer equipment, and then transferring the material from the transfer equipment to optimal unloading equipment.
Further, the optimal offloading device is reachable, representing tasksThe system can be directly unloaded into the optimal unloading equipment through a data transmission link, and can be unloaded into the optimal unloading equipment through a transfer equipment;
the optimal offloading device is not reachable, representing tasksThe task can not be directly transmitted to the optimal unloading device through the data transmission link, and the task can only be unloaded through the transfer device.
Further, the S520 includes:
s521 if the taskDirectly unloading the data to the optimal unloading device through the data transmission link, and calculating communication delay of the optimal unloading device>；
The communication time delay calculation formula of the optimal unloading device is as follows:
；
wherein ,for tasks->Data volume->The data communication rate between the terminal equipment and the optimal unloading equipment is calculated according to the following formula:
；
wherein ,for optimal offloading of bandwidth of the device +.>For the optimal upload rate of the offloading device +.>For the channel gain between the end device and the optimal offloading device, the calculation formula is: />, wherein ,/>For end equipment and optimal removalDistance between the carrier devices->For the attenuation index>Is the additive white gaussian noise variance;
s522 if taskUnloading the data to the optimal unloading device by the transfer device, and calculating the set of directly communicable edge devices +.>Communication delay of edge devices except for the optimal offloading device>；
The communication delay calculation formula of the edge equipment is as follows:
；
wherein ,for tasks->Data volume->For end devices and edge devices->Data communication rate between;
s523, selecting edge equipment with minimum communication time delay as transfer equipment, and calculating tasksTotal communication delay in the case of transmission via a transfer device to an optimal unloading device>；
The total communication time delay calculation formula is as follows:
；
；
wherein j is a transfer device,for the communication latency of task k transmitted from the end device to the relay device,for tasks->The communication time delay transmitted from the transfer device to the optimal unloading device is calculated by the following formula:
；
wherein ,for the task data volume, +.>For the data communication rate between the transfer device and the optimal unloading device, the calculation formula of the data communication rate is as follows:
；
wherein B is the bandwidth of the transit device j,additive white gaussian noise variance for transit device j +.>For the upload rate of the transit device, +.>The calculation formula of the channel gain is as follows: />, wherein ,/>Distance between the transfer device and the optimal unloading device;
s524, judging whether the communication time delay of direct unloading is smaller than the communication time delay of unloading through the transfer equipment, if so, indicating the task to be carried outThe task unloading transmission path directly unloaded to the optimal unloading equipment is better; if not, it means to add the task->Firstly unloading the task to the transfer equipment, and then transferring the task to the optimal unloading equipment from the transfer equipment to obtain a better task unloading transmission path;
s525, selecting a better task unloading transmission path to carry out task unloading.
Further, the S530 includes:
s531 calculating a set of directly communicable edge devicesCommunication delay of edge devices except for the optimal offloading device>；
The communication time delayThe calculation formula of (2) is as follows:
；
wherein ,for tasks->Data volume->For end devices and edge devices->Data communication rate between;
s532 set of edge devices capable of direct communicationThe edge equipment with the smallest communication time delay except the optimal unloading equipment is used as transfer equipment;
s533 computing taskTotal communication delay in case of transfer to optimal unloading device by transfer deviceThe method comprises the steps of carrying out a first treatment on the surface of the The total communication time delay calculation formula is as follows:
；
wherein j is a transfer device,for the communication latency of task k transmitted from the end device to the relay device,for tasks->Transfer from transfer device to optimal offloading device
The prepared communication time delay is calculated according to the following formula:
；
wherein ,for the task data volume, +.>For the data communication rate between the transfer device and the optimal unloading device, the calculation formula of the data communication rate is as follows:
；
wherein B is the bandwidth of the transit device j,additive white gaussian noise variance for transit device j +.>For the upload rate of the transit device, +.>The calculation formula of the channel gain is as follows: />, wherein />Is the distance between the staging device and the optimal unloading device.
Further, the S600 includes:
the resource idle quantity of the optimal unloading equipment is updatedEntering the next round of task unloading decision until all tasks are unloaded;
the calculation formula of the resource idle quantity of the next round of updating optimal unloading equipment is as follows:
；
wherein ,for tasks->Data amount.
In summary, due to the adoption of the technical scheme, the beneficial effects of the invention are as follows:
1. the edge computing task unloading method based on the improved self-learning weight adopts the self-learning resource weight, dynamically adjusts the resource weight when in task unloading decision according to the current idle resource, and avoids that a certain resource is excessively allocated, thereby causing unbalance of the whole cluster resource allocation.
2. According to the edge computing task unloading method based on the improved self-learning weight, provided by the invention, four resources of a central processor, a memory, a bandwidth and a magnetic disk are jointly optimized, a scoring function mechanism is adopted, candidate edge devices are scored according to various resource load conditions, and a computing task with a large resource requirement is unloaded to the edge device with a large resource requirement, so that the task unloading method can adapt to the scheduling requirements of tasks in different scenes, the task unloading efficiency is greatly improved, and the method is high in reliability and good in applicability.
Drawings
FIG. 1 is a flow chart of an edge computing task offloading method based on improved self-learning weights.
Fig. 2 is a flowchart of four kinds of resource weights for initializing an edge computing task offloading method based on improved self-learning weights.
Fig. 3 is a flowchart of a candidate device for screening an edge computing task offloading method based on improved self-learning weights.
Fig. 4 is a flowchart of an optimal unloading device for selecting an edge computing task unloading method based on improved self-learning weights.
Fig. 5 is a schematic diagram of two unloading situations of an edge computing task unloading method based on improved self-learning weights.
Detailed Description
The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making clear and defining the scope of the present invention.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced otherwise than as described herein; it will be apparent that the embodiments in the specification are only some, but not all, embodiments of the invention.
Fig. 1 is a flowchart of an edge computing task offloading method based on improved self-learning weights, where the method includes:
s100: by passing throughThe monitoring tool collects system resource data of the edge devices and the minimum required resource amount of the tasks.
Further, the system resource data includes: device resources, amount of resource free, total amount of resources, upload rate of edge devicesAnd unit data calculation time;
wherein the device resourceIncluding central processing unit resources, memory resources, bandwidth resources, disk resources, expressed as:
；
wherein ,refers to a Central Processing Unit (CPU)>Refers to memory, which is->Refers to bandwidth, & gt>Refers to disk data;
the resource idle quantity comprises a central processing unit idle quantity, a memory idle quantity, a bandwidth idle quantity and a disk idle quantity;
the total amount of resources comprises a total amount of a central processing unit, a total amount of memory, a total amount of bandwidth and a total amount of magnetic disk.
S200: and calculating global shares of four resources of a central processing unit, a memory, a bandwidth and a magnetic disk according to the idle amount of the resources and the total amount of the resources in the system resource data, and initializing weights of the four resources.
Further, referring to fig. 2, the S200 includes:
s210, calculating total idle quantity of four resources in all edge devices；
The calculation formula of the total idle quantity is as follows:
；
wherein ,resource free amount for single edge device, +.>As a single edge device,for device resources including CPU resources, memory resources, bandwidth resources, disk resources, +.>For an unloading task;
s220, calculating the total sum of four resources in all edge devices；
The sum is calculated by the following formula:
；
wherein ,the sum of resources for a single edge device;
s230, calculating idle proportion of four resources in all edge devices;
the calculation formula of the idle proportion is as follows:
；
s240, calculating global shares of four resources in all edge devices；
The calculation formula of the global share is as follows:
；
wherein ,representation ofThe current load conditions of the four resources, and the global share and the load conditions are positively correlated;
s250, initializing weights of four resources in all edge devices；
The formula for initializing the weights of the four resources is as follows:
；
wherein ,is the sum of the global shares of the four resources.
S300: and creating a candidate device set, and adding the edge devices with the resource idle quantity being greater than or equal to the minimum resource demand quantity of the task into the candidate device set.
Further, referring to fig. 3, the step S300 includes:
s310, creating a candidate device set;
s320, judging the resource idle quantity of the edge equipmentWhether or not it is greater than or equal to task->Minimum demand for four resources +.>If yes, adding the edge equipment into the candidate equipment set; if not, the edge device can not complete the task +.>Is unable to perform task offloading; minimum demand of the four resources +.>Representing task k versus the middle of the edge deviceThe minimum demand of CPU resources, memory resources, bandwidth resources and disk resources;
s330, calculating the signal-to-noise ratio of the edge equipment in the candidate equipment set；
The calculation formula of the signal to noise ratio is as follows:
；
wherein ,for edge devices->Upload rate of->For the distance between the end device and the edge device, < >>For the attenuation index>Is the additive white gaussian noise variance.
S400: and calculating the overall scores of the edge devices in the candidate device set according to the weights of the four resources, creating a score queue, sorting the overall scores in a descending order, sequentially adding the overall scores into the score queue, and selecting the edge device with the highest overall score in the score queue as the optimal unloading device.
Further, referring to fig. 4, the S400 includes:
s410 computing taskResource score for each edge device in the candidate device set +.>；
The calculation formula of the resource score is as follows:
；
wherein ,；
s420, calculating taskComputing delay caused by individual edge devices in the candidate device set +.>；
The calculation formula of the calculation time delay caused by each edge device is as follows:；
wherein ,for tasks->Data volume->Time required for calculating unit data for the respective edge devices +.>；
S430, computing taskDelay score +.>；
The calculation formula of the time delay score is as follows:
；
wherein ,maximum computation delay for edge devices, +.>Minimum computation delay for the edge device;
s440, computing taskOverall score +.>Creating a scoring queue, and adding the overall scores into the scoring queue in a descending order;
the calculation formula of the overall score is as follows:
；
s450, selecting the edge equipment with the highest overall score from the score queue as the optimal unloading equipment.
S500: and calculating the communication time delay of different task unloading transmission paths, and selecting the transmission path with the smallest communication time delay to unload the task into the optimal unloading equipment.
Further, referring to fig. 5, the S500 includes:
s510, creating a set of edge devices capable of direct communicationSet of non-directly communicable edge devices>；
S520, judging the signal-to-noise ratio of the edge equipmentWhether or not it is greater than or equal to the signal-to-noise threshold +.>If yes, adding the edge device into the set of directly communicable edge devices +.>In (a) and (b); if not, adding the edge device into the non-directly communicable edge device set +.>In (a) and (b); wherein the signal-to-noise ratio threshold +.>5;
s530, judging whether the optimal unloading equipment belongs to the set of directly communicable edge equipmentIf yes, go to S540, if no, go to S550;
s540, if the optimal unloading device belongs to the set of edge devices capable of direct communicationCalculating the communication time delay of two task unloading transmission paths and selecting a transmission path with smaller communication time delay as an optimal task unloading transmission path if the optimal unloading equipment is reachable;
s550, if the optimal unloading device belongs to the set of non-directly communicable edge devicesI.e. indicating that the optimal unloading device is not reachable, selecting the transit device first, task ∈>Firstly unloading the material into transfer equipment, and then transferring the material from the transfer equipment to optimal unloading equipment.
Further, provided thatThe optimal unloading device is reachable and represents tasksThe system can be directly unloaded into the optimal unloading equipment through a data transmission link, and can be unloaded into the optimal unloading equipment through a transfer equipment;
the optimal offloading device is not reachable, representing tasksThe task can not be directly transmitted to the optimal unloading device through the data transmission link, and the task can only be unloaded through the transfer device.
Further, the S520 includes:
s521 if the taskDirectly unloading the data to the optimal unloading device through the data transmission link, and calculating communication delay of the optimal unloading device>；
The communication time delay calculation formula of the optimal unloading device is as follows:
；
wherein ,for tasks->Data volume->The data communication rate between the terminal equipment and the optimal unloading equipment is calculated according to the following formula:
；
wherein ,for optimal offloading of bandwidth of the device +.>For the optimal upload rate of the offloading device +.>For the channel gain between the end device and the optimal offloading device, the calculation formula is: />, wherein ,/>For the distance between the end device and the optimal unloading device, < >>For the attenuation index>Is the additive white gaussian noise variance;
s522 if taskUnloading the data to the optimal unloading device by the transfer device, and calculating the set of directly communicable edge devices +.>Communication delay of edge devices except for the optimal offloading device>；
The communication delay calculation formula of the edge equipment is as follows:
；
wherein ,for tasks->Data volume->For end devices and edge devices->Data communication rate between;
s523, selecting edge equipment with minimum communication time delay as transfer equipment, and calculating tasksTotal communication delay in the case of transmission via a transfer device to an optimal unloading device>；
The total communication time delay calculation formula is as follows:
；
；/>
wherein j is a transfer device,for the communication latency of task k transmitted from the end device to the relay device,for tasks->The communication time delay transmitted from the transfer device to the optimal unloading device is calculated by the following formula:
；
wherein ,for the task data volume, +.>For the data communication rate between the transfer device and the optimal unloading device, the calculation formula of the data communication rate is as follows:
；
wherein B is the bandwidth of the transit device j,additive white gaussian noise variance for transit device j +.>For the upload rate of the transit device, +.>The calculation formula of the channel gain is as follows: />, wherein ,/>Distance between the transfer device and the optimal unloading device;
s524, judging whether the communication time delay of direct unloading is smaller than the communication time delay of unloading through the transfer equipment, if so, indicating the task to be carried outThe task unloading transmission path directly unloaded to the optimal unloading equipment is better; if not, it means to add the task->Firstly unloading into transfer equipment and then transferring from the transfer equipmentThe task unloading transmission path migrated to the optimal unloading device is better;
s525, selecting a better task unloading transmission path to carry out task unloading.
Further, the S530 includes:
s531 calculating a set of directly communicable edge devicesCommunication delay of edge devices except for the optimal offloading device>；
The communication time delayThe calculation formula of (2) is as follows:
；
wherein ,for tasks->Data volume->For end devices and edge devices->Data communication rate between;
s532 set of edge devices capable of direct communicationThe edge equipment with the smallest communication time delay except the optimal unloading equipment is used as transfer equipment;
s533 computing taskTransmitted to through transfer equipmentTotal communication delay in case of optimal offloading equipmentThe method comprises the steps of carrying out a first treatment on the surface of the The total communication time delay calculation formula is as follows:
；
wherein j is a transfer device,for the communication latency of task k transmitted from the end device to the relay device,for tasks->The communication time delay transmitted from the transfer device to the optimal unloading device is calculated by the following formula:
；
wherein ,for the task data volume, +.>For the data communication rate between the transfer device and the optimal unloading device, the calculation formula of the data communication rate is as follows:
；/>
wherein B is the bandwidth of the transit device j,additive white gaussian noise variance for transit device j +.>Is transit inUpload rate of device,/->The calculation formula of the channel gain is as follows: />, wherein />Is the distance between the staging device and the optimal unloading device.
S600: and updating the resource idle quantity of the optimal unloading equipment, and entering the unloading decision of the next round of tasks until all tasks are unloaded.
Further, the S600 includes:
the resource idle quantity of the optimal unloading equipment is updatedEntering the next round of task unloading decision until all tasks are unloaded;
the calculation formula of the resource idle quantity of the next round of updating optimal unloading equipment is as follows:
；
wherein ,for tasks->Data amount.
The features and capabilities of the present invention are described in further detail below in connection with the examples.
S100: by passing throughCollecting system resource data of the edge equipment by the monitoring tool;
according toThe monitoring tool collects part of data of system resources of the edge device as follows, wherein the data represents the idle quantity of the resources/the total quantity of the resources:
the minimum required amount of resources for a task is as follows:
the parameters were set as follows:
s200: calculating global shares of four resources according to the idle amount of the resources and the total amount of the resources in the system resource data, and initializing weights of the four resources;
the global shareThe calculation results are as follows:
weights of the four resourcesThe initialization results are as follows:
creating a set of directly communicable edge devicesCalculating the signal-to-noise ratio of each edge device, and making the signal-to-noise ratio larger thanOr an edge device equal to a threshold value joins the set of directly communicable edge devices +.>In (a) and (b);
calculating the signal-to-noise ratio of each edge device, wherein the calculation result is as follows:putting the edge devices with the signal-to-noise ratio greater than or equal to the threshold value 5 into the set +.>I.e. set->. Minimum required amount of resources for task k +.>Putting devices with the free quantity larger than or equal to the minimum demand quantity of the resources into a collection +.>Set->。
S400: calculating the overall scores of all edge devices according to the weights, creating a score queue, adding the overall scores into the score queue in a descending order, and selecting the edge device with the highest overall score in the score queue as the optimal unloading device;
the overall score calculation results are as follows:
the edge device 4 is thus selected as the optimal unloading device.
Calculating communication time delays of different task unloading transmission paths, and selecting a transmission path with the smallest communication time delay to unload the task into the optimal unloading equipment;
the edge deviceThe task k can be directly unloaded, or can be unloaded through the transfer equipment. Firstly, selecting the optimal transit equipment j, namely calculating the set +.>The transmission delays of all edge devices except the edge device 4 are calculated as follows:
；
the edge device 5 is thus considered as a transit device.
Respectively calculating transmission delays of two unloading path conditions, directly unloading tasks and communication delays；
By task offloading the edge device 5 as a relay device, communication latency is increased
，
The time delay of task unloading transmission through the transfer equipment is smaller;
the specific offloading strategy is: task k is offloaded to edge device 5 and then migrated to edge device 4 for processing.
The idle amount of four resources on the optimal unloading device 4 is updated, and the updated data are as follows:
each block in the flowchart or block diagrams in the figures may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (7)
1. An edge computing task offloading method based on improved self-learning weights, comprising:
s100: by passing throughThe monitoring tool collects system resource data of the edge equipment and the minimum resource demand of the task;
the system resource data includes: equipment resources, resource idle quantity, resource total amount, uploading rate of edge equipment and unit data calculation time;
s200: calculating global shares of four resources of a central processing unit, a memory, a bandwidth and a magnetic disk according to the idle amount of the resources and the total amount of the resources in the system resource data, and initializing weights of the four resources;
s300: creating a candidate device set, and adding edge devices with the resource idle quantity being greater than or equal to the minimum resource demand quantity of the task into the candidate device set;
s400: calculating overall scores of edge devices in the candidate device set according to the weights of the four resources, creating a score queue, sorting the overall scores in a descending order, sequentially adding the overall scores into the score queue, and selecting edge devices with highest overall scores in the score queue as optimal unloading devices;
s500: calculating communication time delays of different task unloading transmission paths, and selecting a transmission path with the smallest communication time delay to unload the task into the optimal unloading equipment;
s600: updating the resource idle quantity of the optimal unloading equipment, and entering an unloading decision of the next round of tasks until all tasks are unloaded;
the S200 includes:
s210, calculating total idle quantity of four resources in all edge devices；
The calculation formula of the total idle quantity is as follows:
；
wherein ,resource free amount for single edge device, +.>For a single edge device->For device resources including CPU resources, memory resources, bandwidth resources, disk resources, +.>For an unloading task;
s220, calculating the total sum of four resources in all edge devices；
The sum is calculated by the following formula:
；
wherein ,the sum of resources for a single edge device;
s230, calculating idle proportion of four resources in all edge devices;
the calculation formula of the idle proportion is as follows:
；
s240, calculating global shares of four resources in all edge devices；
The calculation formula of the global share is as follows:
；
wherein ,representing four resources asThe previous load condition, the global share and the load condition are positively correlated;
s250, initializing weights of four resources in all edge devices；
The formula for initializing the weights of the four resources is as follows:
；
wherein ,is the sum of the global shares of the four resources;
the S400 includes:
s410 computing taskResource score for each edge device in the candidate device set +.>；
The calculation formula of the resource score is as follows:
；
wherein ,；
s420, calculating taskComputing delay caused by individual edge devices in the candidate device set +.>；
The saidThe calculation formula of the calculation time delay caused by each edge device is as follows:；
wherein ,for tasks->Data volume->Time required for calculating unit data for the respective edge devices +.>；
S430, computing taskDelay score +.>；
The calculation formula of the time delay score is as follows:
；
wherein ,maximum computation delay for edge devices, +.>Minimum computation delay for the edge device;
s440, computing taskIs the least suitable for the personOverall score +.>Creating a scoring queue, and adding the overall scores into the scoring queue in a descending order;
the calculation formula of the overall score is as follows:
；
s450, selecting the edge equipment with the highest overall score from the score queue as the optimal unloading equipment.
2. The edge computing task offloading method of claim 1, wherein the device resource is configured toIncluding central processing unit resources, memory resources, bandwidth resources, disk resources, expressed as:
；
wherein ,refers to a Central Processing Unit (CPU)>Refers to memory, which is->Refers to bandwidth, & gt>Refers to disk data;
the resource idle quantity comprises a central processing unit idle quantity, a memory idle quantity, a bandwidth idle quantity and a disk idle quantity;
the total amount of resources comprises a total amount of a central processing unit, a total amount of memory, a total amount of bandwidth and a total amount of magnetic disk.
3. The edge computing task offloading method of claim 1, wherein S300 comprises:
s310, creating a candidate device set;
s320, judging the resource idle quantity of the edge equipmentWhether or not it is greater than or equal to task->Minimum demand for four resources +.>If yes, adding the edge equipment into the candidate equipment set; if not, the edge device can not complete the task +.>Is unable to perform task offloading; minimum demand of the four resources +.>Representing the minimum demand of task k on the CPU resource, memory resource, bandwidth resource and disk resource in the edge equipment;
s330, calculating the signal-to-noise ratio of the edge equipment in the candidate equipment set；
The calculation formula of the signal to noise ratio is as follows:
；
wherein ,for edge devices->Upload rate of->For the distance between the end device and the edge device, < >>For the attenuation index>Is the additive white gaussian noise variance.
4. The edge computing task offloading method of claim 1, wherein S500 comprises:
s510, creating a set of edge devices capable of direct communicationSet of non-directly communicable edge devices>；
S520, judging the signal-to-noise ratio of the edge equipmentWhether or not it is greater than or equal to the signal-to-noise threshold +.>If yes, adding the edge device into the set of directly communicable edge devices +.>In (a) and (b); if not, edge is formedEdge device joins the set of non-directly communicable edge devices +.>In (a) and (b); wherein the signal-to-noise ratio threshold +.>5;
s530, judging whether the optimal unloading equipment belongs to the set of directly communicable edge equipmentIf yes, go to S540, if no, go to S550;
s540, if the optimal unloading device belongs to the set of edge devices capable of direct communicationCalculating the communication time delay of two task unloading transmission paths and selecting a transmission path with smaller communication time delay as an optimal task unloading transmission path if the optimal unloading equipment is reachable;
s550, if the optimal unloading device belongs to the set of non-directly communicable edge devicesI.e. indicating that the optimal unloading device is not reachable, selecting the transit device first, task ∈>Firstly unloading the material into transfer equipment, and then transferring the material from the transfer equipment to optimal unloading equipment;
the optimal unloading device is reachable and represents tasksThe system can be directly unloaded into the optimal unloading equipment through a data transmission link, and can be unloaded into the optimal unloading equipment through a transfer equipment;
the optimal offloading device is not reachable, representing tasksThe task can not be directly transmitted to the optimal unloading device through the data transmission link, and the task can only be unloaded through the transfer device.
5. The method for offloading edge computing tasks based on improved self-learning weights as recited in claim 4, wherein said S520 comprises:
s521 if the taskDirectly unloading the data to the optimal unloading device through the data transmission link, and calculating communication delay of the optimal unloading device>；
The communication time delay calculation formula of the optimal unloading device is as follows:
；
wherein ,for tasks->Data volume->The data communication rate between the terminal equipment and the optimal unloading equipment is calculated according to the following formula:
；
wherein ,for optimal offloading of bandwidth of the device +.>For the optimal upload rate of the offloading device +.>For the channel gain between the end device and the optimal offloading device, the calculation formula is: />, wherein ,/>For the distance between the end device and the optimal unloading device, < >>For the attenuation index>Is the additive white gaussian noise variance;
s522 if taskUnloading the data to the optimal unloading device by the transfer device, and calculating the set of directly communicable edge devices +.>Communication delay of edge devices except for the optimal offloading device>；
The communication delay calculation formula of the edge equipment is as follows:
；
wherein ,for tasks->Data volume->For end devices and edge devices->Data communication rate between;
s523, selecting edge equipment with minimum communication time delay as transfer equipment, and calculating tasksTotal communication delay in the case of transmission via a transfer device to an optimal unloading device>；
The total communication time delay calculation formula is as follows:
；
；
wherein j is a transfer device,for the communication latency of task k transmitted from the end device to the relay device,for tasks->The communication time delay transmitted from the transfer device to the optimal unloading device is calculated by the following formula:
；
wherein ,for the task data volume, +.>For the data communication rate between the transfer device and the optimal unloading device, the calculation formula of the data communication rate is as follows:
；
wherein B is the bandwidth of the transit device j,additive white gaussian noise variance for transit device j +.>For the upload rate of the transit device, +.>The calculation formula of the channel gain is as follows: />, wherein ,/>Distance between the transfer device and the optimal unloading device;
s524, judging whether the communication time delay of direct unloading is smaller than the communication time delay of unloading through the transfer equipment, if so, indicating the task to be carried outThe task unloading transmission path directly unloaded to the optimal unloading equipment is better; if not, it means to add the task->Firstly unloading the task to the transfer equipment, and then transferring the task to the optimal unloading equipment from the transfer equipment to obtain a better task unloading transmission path;
s525, selecting a better task unloading transmission path to carry out task unloading.
6. The edge computing task offloading method of claim 4, wherein S530 comprises:
s531 calculating a set of directly communicable edge devicesCommunication delay of edge devices except for optimal unloading device；
The communication time delayThe calculation formula of (2) is as follows:
；
wherein ,for tasks->Data volume->For end devices and edge devices->Data communication rate between;
s532 set of edge devices capable of direct communicationThe edge equipment with the smallest communication time delay except the optimal unloading equipment is used as transfer equipment;
s533 computing taskTotal communication delay in the case of transmission via a transfer device to an optimal unloading device>The method comprises the steps of carrying out a first treatment on the surface of the The total communication time delay calculation formula is as follows:
；
wherein j is a transfer device,for the communication latency of task k transmitted from the end device to the relay device,for tasks->The communication time delay transmitted from the transfer device to the optimal unloading device is calculated by the following formula:
；
wherein ,for the task data volume, +.>For the data communication rate between the transfer device and the optimal unloading device, the calculation formula of the data communication rate is as follows:
；
wherein B is the bandwidth of the transit device j,additive white gaussian noise variance for transit device j +.>For the upload rate of the transit device, +.>The calculation formula of the channel gain is as follows: />, wherein />Is the distance between the staging device and the optimal unloading device.
7. The edge computing task offloading method of claim 1, wherein S600 comprises:
the resource idle quantity of the optimal unloading equipment is updatedEntering the next round of task unloading decision until all tasks are unloaded;
the calculation formula of the resource idle quantity of the next round of updating optimal unloading equipment is as follows:
；
wherein ,for tasks->Data amount.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310956562.2A CN116668447B (en) | 2023-08-01 | 2023-08-01 | Edge computing task unloading method based on improved self-learning weight |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310956562.2A CN116668447B (en) | 2023-08-01 | 2023-08-01 | Edge computing task unloading method based on improved self-learning weight |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116668447A CN116668447A (en) | 2023-08-29 |
| CN116668447B true CN116668447B (en) | 2023-10-20 |
Family
ID=87715782
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310956562.2A Active CN116668447B (en) | 2023-08-01 | 2023-08-01 | Edge computing task unloading method based on improved self-learning weight |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116668447B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111148155A (en) * | 2019-12-26 | 2020-05-12 | 曲阜师范大学 | Task unloading method based on mobile edge calculation |
| CN111262906A (en) * | 2020-01-08 | 2020-06-09 | 中山大学 | Method for unloading mobile user terminal task under distributed edge computing service system |
| CN113950103A (en) * | 2021-09-10 | 2022-01-18 | 西安电子科技大学 | Multi-server complete computing unloading method and system under mobile edge environment |
| CN114024970A (en) * | 2021-09-28 | 2022-02-08 | 国网辽宁省电力有限公司锦州供电公司 | Power internet of things work load distribution method based on edge calculation |
| CN114564304A (en) * | 2022-02-15 | 2022-05-31 | 广州杰赛科技股份有限公司 | Task unloading method for edge calculation |
| WO2023024219A1 (en) * | 2021-08-24 | 2023-03-02 | 苏州大学 | Joint optimization method and system for delay and spectrum occupancy in cloud-edge collaborative network |
| CN116032935A (en) * | 2022-11-08 | 2023-04-28 | 北京邮电大学 | Task unloading method and device, electronic equipment and storage medium |
| CN116112488A (en) * | 2022-12-15 | 2023-05-12 | 山东大学 | Fine-grained task unloading and resource allocation method for MEC network |
| CN116390161A (en) * | 2023-03-20 | 2023-07-04 | 重庆邮电大学 | Task migration method based on load balancing in mobile edge calculation |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111245651B (en) * | 2020-01-08 | 2022-03-29 | 上海交通大学 | Task unloading method based on power control and resource allocation |
-
2023
- 2023-08-01 CN CN202310956562.2A patent/CN116668447B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111148155A (en) * | 2019-12-26 | 2020-05-12 | 曲阜师范大学 | Task unloading method based on mobile edge calculation |
| CN111262906A (en) * | 2020-01-08 | 2020-06-09 | 中山大学 | Method for unloading mobile user terminal task under distributed edge computing service system |
| WO2023024219A1 (en) * | 2021-08-24 | 2023-03-02 | 苏州大学 | Joint optimization method and system for delay and spectrum occupancy in cloud-edge collaborative network |
| CN113950103A (en) * | 2021-09-10 | 2022-01-18 | 西安电子科技大学 | Multi-server complete computing unloading method and system under mobile edge environment |
| CN114024970A (en) * | 2021-09-28 | 2022-02-08 | 国网辽宁省电力有限公司锦州供电公司 | Power internet of things work load distribution method based on edge calculation |
| CN114564304A (en) * | 2022-02-15 | 2022-05-31 | 广州杰赛科技股份有限公司 | Task unloading method for edge calculation |
| CN116032935A (en) * | 2022-11-08 | 2023-04-28 | 北京邮电大学 | Task unloading method and device, electronic equipment and storage medium |
| CN116112488A (en) * | 2022-12-15 | 2023-05-12 | 山东大学 | Fine-grained task unloading and resource allocation method for MEC network |
| CN116390161A (en) * | 2023-03-20 | 2023-07-04 | 重庆邮电大学 | Task migration method based on load balancing in mobile edge calculation |
Non-Patent Citations (5)
| Title |
|---|
| Privacy-Preserving Blockchain-Based Federated Learning for IoT Devices;Yang Zhao等;《IEEE Internet of Things Journal》;第08卷(第03期);1817-1829 * |
| 一种基于强化学习的多节点MEC计算资源分配方案;余萌迪等;《通信技术》(第12期);88-93 * |
| 基于深度强化学习的多用户边缘计算任务卸载调度与资源分配算法;邝祝芳等;《计算机学报》;第45卷(第04期);812-824 * |
| 超密集网络中基于MEC的任务卸载和资源分配;陈发堂等;《南京邮电大学学报（自然科学版）》;第42卷(第02期);22-30 * |
| 面向多用户资源均衡分配的MEC计算卸载策略;卢敏等;《计算机集成制造系统》;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116668447A (en) | 2023-08-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110493360B (en) | Mobile edge computing unloading method for reducing system energy consumption under multiple servers | |
| CN111930436B (en) | Random task queuing unloading optimization method based on edge calculation | |
| CN111240701B (en) | Task unloading optimization method for end-side-cloud collaborative computing | |
| CN110557732B (en) | Vehicle edge computing network task unloading load balancing system and balancing method | |
| US11831708B2 (en) | Distributed computation offloading method based on computation-network collaboration in stochastic network | |
| CN114662661A (en) | Method for accelerating multi-outlet DNN reasoning of heterogeneous processor under edge calculation | |
| CN114422349B (en) | Cloud-edge-end-collaboration-based deep learning model training and reasoning architecture deployment method | |
| CN113961264B (en) | Intelligent unloading algorithm and system for video monitoring cloud edge cooperation | |
| CN113918240A (en) | Task unloading method and device | |
| CN112860429A (en) | Cost-efficiency optimization system and method for task unloading in mobile edge computing system | |
| CN115562756A (en) | Multi-access edge computing vehicle task unloading method and system | |
| CN114168328A (en) | Mobile edge node calculation task scheduling method and system based on federal learning | |
| CN116668447B (en) | Edge computing task unloading method based on improved self-learning weight | |
| CN114374949A (en) | Power control mechanism based on information freshness optimization in Internet of vehicles | |
| CN111611069B (en) | Multi-type task migration method among multiple data centers | |
| CN116916386A (en) | Large model auxiliary edge task unloading method considering user competition and load | |
| CN114615705B (en) | Single-user resource allocation strategy method based on 5G network | |
| CN116017570A (en) | Edge computing system resource management method based on block chain | |
| CN115499875A (en) | Satellite internet task unloading method and system and readable storage medium | |
| CN114928611A (en) | Internet of vehicles energy-saving calculation unloading optimization method based on IEEE802.11p protocol | |
| CN114742166A (en) | Communication network field maintenance model migration method based on time delay optimization | |
| Hoiles et al. | Risk-averse caching policies for YouTube content in femtocell networks using density forecasting | |
| CN113590211A (en) | Calculation unloading method based on PSO-DE algorithm | |
| CN117114113B (en) | Collaborative reasoning acceleration method based on queuing theory | |
| CN112203309A (en) | Joint task unloading and caching method based on server cooperation |
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 |