CN112162863B - Edge unloading decision method, terminal and readable storage medium - Google Patents

Edge unloading decision method, terminal and readable storage medium Download PDF

Info

Publication number
CN112162863B
CN112162863B CN202011127194.3A CN202011127194A CN112162863B CN 112162863 B CN112162863 B CN 112162863B CN 202011127194 A CN202011127194 A CN 202011127194A CN 112162863 B CN112162863 B CN 112162863B
Authority
CN
China
Prior art keywords
task
unloading
service terminal
edge service
edge
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
Application number
CN202011127194.3A
Other languages
Chinese (zh)
Other versions
CN112162863A (en
Inventor
张展
左德承
冯懿
封威
付国栋
张道鹏
尚江卫
王启航
刘宏伟
董剑
舒燕君
罗丹彦
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Harbin Institute of Technology
Original Assignee
Harbin Institute of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Harbin Institute of Technology filed Critical Harbin Institute of Technology
Priority to CN202011127194.3A priority Critical patent/CN112162863B/en
Publication of CN112162863A publication Critical patent/CN112162863A/en
Application granted granted Critical
Publication of CN112162863B publication Critical patent/CN112162863B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806—Task transfer initiation or dispatching
    • G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
    • G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2209/00—Indexing scheme relating to G06F9/00
    • G06F2209/50—Indexing scheme relating to G06F9/50
    • G06F2209/509—Offload

Abstract

The invention provides an edge unloading decision method, a terminal and a readable storage medium, wherein the edge unloading decision method comprises the following steps: acquiring an offloadable task and monitoring state information of a wearable terminal; establishing communication connection with an edge service terminal, and receiving edge service terminal state information sent by the edge service terminal; selecting the edge service terminal and sending an unloading request of the task capable of being unloaded; after receiving the unloading instruction, sending the task information capable of unloading to the edge service terminal; and receiving an execution result obtained by executing the task capable of unloading. In this way, the wearable terminal performs primary allocation direction on the task to be allocated, and then the edge service terminal performs secondary analysis according to own computing resources to give feedback to the task which can be executed, so that the situation that the application of the wearable terminal is blocked due to incapability of executing the task which needs to be executed by the edge service terminal in time caused by excessive task which can be executed by the edge service terminal is avoided.

Description

Edge unloading decision method, terminal and readable storage medium
Technical Field
The present invention relates to the field of edge decision making technologies, and in particular, to an edge unloading decision making method, a terminal, and a readable storage medium.
Background
In recent years, wearable equipment is increasingly widely applied in various fields, and in the running process, sensor data can be continuously collected, and physiological monitoring, target detection, image recognition, virtual reality, augmented reality and other operations are performed through corresponding algorithms. On one hand, the data volume is larger, and on the other hand, more and more complex intelligent application computing tasks consume a large amount of resources and energy of the equipment, so that the storage and computing pressure of the wearable equipment can be shared to an edge server by utilizing an edge computing technology, and the cruising and performance of the wearable equipment are improved.
Therefore, by utilizing the edge computing technology, computing resources and storage resources can be placed on an edge server close to the wearable equipment, the defects in the aspects of computing performance and efficiency of the wearable equipment can be overcome, and the application execution speed is increased.
However, the existing edge computing technology generally distributes tasks to be offloaded to the edge servers directly, which results in excessive offloading tasks distributed by part of the edge servers, and greatly slows down the execution speed of applications on the wearable device.
Disclosure of Invention
The problem solved by the invention is that the existing task allocation needing to be offloaded easily causes excessive task allocation of part of edge servers.
To solve the above problems, the present invention provides an edge offloading decision method, which is applied to a wearable terminal, and includes:
acquiring an offloadable task and monitoring state information of a wearable terminal;
establishing communication connection with an edge service terminal, and receiving edge service terminal state information sent by the edge service terminal;
selecting the edge service terminal according to the wearable terminal state information and the edge service terminal state information and sending an unloading request of the unloading task;
after receiving an unloading instruction sent by the edge service terminal, sending information of an unloading task to the edge service terminal;
and receiving an execution result obtained by executing the task capable of unloading, which is sent by the edge service terminal.
In this way, the wearable terminal performs primary allocation direction on the task to be allocated, and then the edge service terminal performs secondary analysis according to own computing resources to give feedback to the task which can be executed, so that the situation that the application of the wearable terminal is blocked due to incapability of executing the task which needs to be executed by the edge service terminal in time caused by excessive task which can be executed by the edge service terminal is avoided.
Optionally, the acquiring the task capable of unloading includes:
Acquiring a current application to be executed;
dividing the application to be executed into a plurality of tasks to be executed, and dividing the tasks to be executed into the task to be offloadable and the task not to be offloadable according to the execution process of the tasks to be executed.
In this way, the application to be executed is divided into the task which can be unloaded and the task which can not be unloaded and are (approximately) independent of each other through the division, so that the subsequent task unloading can be conveniently carried out, and the execution speed of the application to be executed is increased.
Optionally, the establishing communication connection with the edge service terminal includes:
scanning communication access points of edge service terminals in a current range;
acquiring the signal intensity of each communication access point;
and selecting the communication access points with signal strength larger than preset strength, and establishing a communication connection relation.
Therefore, the communication access point with the signal strength larger than the preset strength is selected, so that the problem that the task capable of being unloaded cannot be smoothly carried out due to weak signals can be avoided, and the execution speed and time of the task capable of being unloaded can be improved.
Optionally, the selecting the edge service terminal according to the wearable terminal state information and the edge service terminal state information and sending an offloading request of the offloadable task includes:
Setting the execution time weight and the execution energy consumption weight of an application to be executed;
establishing a benefit constraint of an application to be executed according to the wearable terminal state information, the edge service terminal state information, the execution time weight and the execution energy consumption weight;
solving the benefit constraint with the maximum benefit of the application to be executed as a target to obtain a solving result;
and determining the selected edge service terminal according to the solving result, and sending an unloading request of the task capable of unloading.
In this way, by setting the execution time weight and the execution energy consumption weight, the execution energy consumption and the execution time can be comprehensively evaluated to evaluate the unloading strategy, so that the decision purpose of comprehensively considering the execution energy consumption or the execution time is achieved, and the distributed edge service terminals of each unloading task under the condition of comprehensively considering the execution energy consumption and the execution time are determined by establishing the profit constraint and the solution, so that proper unloading mechanisms are respectively selected according to different calculation complexity and data quantity, the execution energy consumption and the execution time are effectively reduced, and the execution success rate is improved.
Optionally, the benefit constraint of the application to be executed is:
The maximum profit constraint of the application to be executed is as follows:
wherein,and->Weights expressed as user u for execution time and execution energy consumption of the application, +.>And->The optimized parameters are time delay and energy consumption respectively; />For the local execution time of an application, +.>For offloading to edge service terminals; />For the local execution of an application, energy consumption, +.>For unloading energy consumption; />For wireless transmission power, < >>The time for transmitting application execution data to the edge service terminal for the wearable terminal; the attribute of application i offloaded by user u is +.>Wherein d is i The amount of data needed for execution of application i is tarnished u,s For radio transmission speed x u,s Indicating whether the application currently requested to be uninstalled by user u is executed at edge service terminal s, x u,s =1 denotes execution at edge, x u,s =0 means performed at the mobile end; />For a given time limit; s is an edge end node set, U is a user set, and I is an application set.
In this way, through the benefit constraint, the corresponding relation between the distribution of the task capable of being unloaded and the data capable of being quantized is established, so that the distribution of the task capable of being unloaded can be continuously optimized through the corresponding relation, or the maximum benefit can be directly solved, and the edge service terminal of the distribution of each task capable of being unloaded under the condition of comprehensively considering the execution energy consumption and the execution time is determined, thereby effectively reducing the execution energy consumption and the execution time, and improving the execution success rate.
Next, a wearable terminal is provided, comprising a computer readable storage medium storing a computer program and a processor, which computer program, when read and executed by the processor, implements the edge offload decision method as described above.
In this way, the wearable terminal performs primary allocation direction on the task to be allocated, and then the edge service terminal performs secondary analysis according to own computing resources to give feedback to the task which can be executed, so that the situation that the application of the wearable terminal is blocked due to incapability of executing the task which needs to be executed by the edge service terminal in time caused by excessive task which can be executed by the edge service terminal is avoided.
The edge offloading decision-making method is provided again, and is applied to an edge service terminal, and comprises the following steps:
monitoring state information of an edge service terminal;
establishing communication connection with a wearable terminal and sending the state information of the edge service terminal;
after receiving an unloading request of an unloading task sent by the wearable terminal, selecting the unloading task according to a time delay limiting condition, and sending an unloading instruction of the unloading task;
after receiving the task information capable of unloading sent by the wearable terminal, executing the task capable of unloading to obtain an execution result;
And sending the execution result to the edge service terminal.
Therefore, by setting the uninstalling instruction, the edge service terminal can determine the to-be-executed uninstalled task according to the self computing resource, computing capacity and other parameters, so that the situation that the processing speed is slow and the execution speed of the wearable terminal is slow due to the fact that the to-be-executed uninstalled task is too much is avoided.
Optionally, after receiving the task unloading request sent by the wearable terminal, selecting the task unloading according to a time delay constraint condition, and sending an task unloading instruction of the task unloading, including:
after receiving an unloading request of the task capable of being unloaded sent by the wearable terminal, counting the received unloading request of the task capable of being unloaded;
setting time delay limiting conditions of the edge service terminal;
establishing an unloading gain constraint of the edge service terminal according to the counted unloading request of the unloading task and the time delay constraint condition of the edge service terminal;
solving the unloading benefit constraint by taking the maximum unloading benefit of the edge service terminal as a target to obtain a solving result;
and determining the selected task capable of being unloaded according to the solving result of the unloading benefit constraint, and sending an unloading instruction of the task capable of being unloaded.
In this way, by establishing the offloading benefit constraint and solving, the offloading task and the offloading sequence which can be offloaded are determined according to the bandwidth resource and the application offloading benefit, and appropriate offloading mechanisms are respectively selected according to different calculation complexity and data quantity, so that the situation that the processing speed is slow and the execution speed of the wearable terminal is slow due to excessive offloading tasks to be executed is avoided, and the execution efficiency is improved.
Optionally, the offloading benefit constraint of the edge service terminal is:
the maximum constraint of the offloading gain of the edge service terminal is as follows:
wherein pi u For the priority of edge service terminals to wearable end users,and->Expressed as weights of user u on application execution time and execution energy consumption, y u Indicating whether the unloading request of the mobile terminal user u is allowed or not, and if the unloading request is 1, indicating that the application which allows the mobile terminal user u to be unloaded currently is executed on the edge terminal; the attribute of application i offloaded by user u is +.>d i Execution for application iThe amount of data required for a row, c i To apply the floating point computational complexity of the inference part of the i model, the obscuration u,s For the acquired wireless transmission speed, +.>Weight for user execution time for application, +.>Local execution time for the application; / >In order to transmit power over the air,the energy consumption is executed for the application locally; f (f) s,i Computing resources allocated for the application i for the edge service terminal s; />Latency at edge service terminal s for the offload application of user u; />For a given time limit, RB s The total bandwidth available to the edge end node, f s Computing resources for the edge node total; u (U) s The user set is an edge service terminal s, and the application set is I.
In this way, through the offloading benefit constraint, the corresponding relation between the allocation of the offloading tasks and the front of the quantifiable data is established, so that the allocation of the computing resources of the edge service terminal can be continuously optimized through the corresponding relation, or the maximum benefit is directly solved, and the offloading tasks and the offloading sequence which can be offloaded are determined.
An edge service terminal is provided, the edge service terminal comprises a computer readable storage medium storing a computer program and a processor, and the computer program is read and run by the processor to realize the edge unloading decision method.
Therefore, by setting the uninstalling instruction, the edge service terminal can determine the to-be-executed uninstalled task according to the self computing resource, computing capacity and other parameters, so that the situation that the processing speed is slow and the execution speed of the wearable terminal is slow due to the fact that the to-be-executed uninstalled task is too much is avoided.
Finally, a computer readable storage medium is provided, where a computer program is stored, where the computer program, when read and executed by a processor, implements the edge offload decision method described above, or implements the edge offload decision method described above.
In this way, the wearable terminal performs primary allocation direction on the task to be allocated, and then the edge service terminal performs secondary analysis according to own computing resources to give feedback to the task which can be executed, so that the situation that the application of the wearable terminal is blocked due to incapability of executing the task which needs to be executed by the edge service terminal in time caused by excessive task which can be executed by the edge service terminal is avoided.
Drawings
FIG. 1 is a flow chart of a wearable terminal side edge offloading decision-making method according to an embodiment of the invention;
FIG. 2 is a flow chart of an edge unload decision method S10 according to an embodiment of the invention;
FIG. 3 is a flowchart of an edge offload decision method S20 according to an embodiment of the present invention;
FIG. 4 is a flowchart of an edge offload decision method S30 according to an embodiment of the invention;
FIG. 5 is a flow chart of an edge service terminal side edge offloading decision method according to an embodiment of the invention;
fig. 6 is a flowchart of an edge offload decision method S300 according to another embodiment of the invention.
Detailed Description
In order that the above objects, features and advantages of the invention will be readily understood, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings.
In recent years, wearable equipment is increasingly widely applied in various fields, and in the running process, sensor data can be continuously collected, and physiological monitoring, target detection, image recognition, virtual reality, augmented reality and other operations are performed through corresponding algorithms. On one hand, the data volume is larger, and on the other hand, more and more complex intelligent application computing tasks consume a large amount of resources and energy of the equipment, so that the storage and computing pressure of the wearable equipment can be shared to an edge server by utilizing an edge computing technology, and the cruising and performance of the wearable equipment are improved. By utilizing the edge computing technology, computing resources and storage resources can be placed on an edge server close to the wearable device, the edge server can provide low-delay service for the wearable device, and coordinate computing with the wearable device, so that on one hand, the defect of computing performance of the wearable device can be solved, the application execution speed is increased, and on the other hand, the defect of efficacy of the wearable device can be solved.
According to the existing configuration information, a plurality of edge servers are generally distributed around one wearable device, so that the wearable device can place resources to be calculated or stored on the surrounding edge servers, and return results after calculation or reading by the edge servers (the stored resources can be placed on the cloud server, and downloading and other operations can be performed by the edge servers when needed).
However, the existing wearable device generally distributes tasks to be offloaded directly to the edge server; however, the computing power and the transmission power of the edge servers are different, so that the direct allocation can cause a part of the edge servers to be in an idle state, and a part of the edge servers are excessively distributed with unloading tasks and cannot be processed in time, so that for the execution application in the wearable device, the middle task data cannot be fed back from the edge servers later, and cannot be smoothly executed, and the regular clamping of the execution application is reflected.
This cluttered situation, in the case of multiple wearable terminals, would be more complex.
The invention will be described in detail below with reference to the drawings in connection with embodiments.
The embodiment of the disclosure provides an edge unloading decision-making method, and a device for executing the method can be integrated in electronic equipment such as a mobile phone, a pad, a mobile computer, a wearable device and the like. FIG. 1 is a flow chart of a wearable terminal side edge offloading decision method according to an embodiment of the invention; the edge unloading decision method is applied to a wearable terminal and comprises the following steps:
s10, acquiring an offloadable task and monitoring state information of a wearable terminal;
the task capable of being unloaded is a task which can be distributed to the edge service terminal for processing; the task may be acquired through a preset manner, or may be acquired through other realizable manners.
The wearable terminal information comprises battery power and CPU utilization rate of the wearable terminal; preferably, the wearable terminal information further includes a wireless transmission speed of the wearable terminal. In this way, this information can be obtained by monitoring, thereby facilitating decision making for subsequent edge offloading.
S20, establishing communication connection with an edge service terminal, and receiving edge service terminal state information sent by the edge service terminal;
The communication connection with the edge service terminal can be WiFi connection or MQTT connection.
The edge service terminal state information comprises computing resources and communication resource states of the edge service terminal.
S30, selecting the edge service terminal according to the wearable terminal state information and the edge service terminal state information and sending an unloading request of the unloading task;
it should be noted that there may be multiple wearable terminals, and there may be multiple edge service terminals; a complex communication connection relationship is established between the plurality of wearable terminals and the plurality of edge service terminals.
In this step, the offloading request to the wearable terminal is performed in units of the wearable terminal, and when only the edge service terminal having communication connection with the wearable terminal is considered to determine that a part of offloading tasks are allocated to one or some corresponding edge service terminals, the offloading request is sent to the corresponding edge service terminals.
Wherein the offload request includes: the type of task that can be offloaded, the wireless transmission speed that can be obtained, and the data transmission amount. Therefore, the method is convenient for the edge service terminal to accurately analyze the unloading condition of the unloading task after receiving, thereby giving a conclusion.
S40, after receiving an unloading instruction sent by the edge service terminal, sending information of an unloading task to the edge service terminal;
the edge service terminal receives unloading requests of the plurality of wearable terminals, analyzes the unloading requests, and can send unloading instructions to the corresponding wearable terminals if the edge service terminal decides to receive part or all of the unloading tasks.
Therefore, by setting the uninstalling instruction, the edge service terminal can determine the to-be-executed uninstalled task according to the self computing resource, computing capacity and other parameters, so that the situation that the processing speed is slow and the execution speed of the wearable terminal is slow due to the fact that the to-be-executed uninstalled task is too much is avoided.
In addition, by setting the unloading instruction, the edge service terminal can be used as an individual to analyze the unloading tasks which can be executed by the edge service terminal, so that the problem of distribution confusion caused by a plurality of wearable terminals is avoided; meanwhile, the wearable terminal only receives the unloading instruction fed back by the edge service terminal, and then sends the information of the task which can be unloaded and is involved in the unloading instruction, so that the situation that the task which can not be unloaded and is processed is sent is avoided.
If only part of the off-loadable tasks feed back the off-load instruction, the off-loadable tasks not fed back can be redistributed to other edge service terminals or the distribution is abandoned and directly executed by the wearable terminal; and the specific execution and distribution conditions can be set in other ways according to the actual requirements.
It should be noted that, the task information that can be sent may be information after the wearable terminal performs data preprocessing (such as compaction, sorting, screening, etc.), so that data that needs to be transmitted may be greatly reduced, thereby reducing transmission time and execution time of the edge service terminal, and improving application execution time of the wearable terminal.
S50, receiving an execution result obtained by executing the task capable of unloading, which is sent by the edge service terminal.
Wherein, for the task which can be unloaded and is distributed to the edge service terminal for processing, the execution result is directly received; for the task which can be directly executed by the wearable terminal and can be unloaded, the execution result is directly obtained through execution, and the execution result or the execution process can be reported to the edge service terminal for storage, so that the cloud node can conveniently check.
In this way, the wearable terminal performs primary allocation direction on the task to be allocated, and then the edge service terminal performs secondary analysis according to own computing resources to give feedback to the task which can be executed, so that the situation that the application of the wearable terminal is blocked due to incapability of executing the task which needs to be executed by the edge service terminal in time caused by excessive task which can be executed by the edge service terminal is avoided.
Optionally, as shown in fig. 2, the step S10 of acquiring the task capable of unloading includes:
s11, acquiring a current application to be executed;
and a plurality of applications are arranged in one wearable terminal, wherein the applications to be unloaded are divided and added into a local computing task queue, and the applications at the head of the queue are the current applications to be executed according to the sequence.
S12, dividing the application to be executed into a plurality of tasks to be executed, and dividing the tasks to be executed into the task to be offloadable and the task not to be offloadable according to the execution process of the tasks to be executed.
Any application is actually composed of a series of operations arranged in sequence; the division of the application is to divide the sequential operation of an application into a plurality of relatively independent tasks according to the set standard or purpose, and the preset purpose of the application can be completed after the tasks are executed.
The specific division is different if the standards or purposes of the application division are different; for example, an application to be executed may be divided into as many tasks to be executed as possible; the method can be divided into tasks to be executed with as few data interactions as possible before each other according to the specific execution situation; the task to be executed can be divided into the content which can be unloaded and the content which cannot be unloaded as far as possible according to the different unloading requirements.
Dividing the task to be executed into the task capable of being unloaded and the task capable of not being unloaded according to the execution process of the task to be executed; that is, if the wearable device has to participate in the execution process, the task is not necessarily an off-loadable task, and the rest of the tasks can be divided into off-loadable tasks (a part of the off-loadable tasks can be considered as the off-loadable tasks according to other actual requirements, such as computing resource allocation, etc.).
For example, for applications of four wearable terminals, such as object-oriented detection, motion recognition, face detection and data fusion, data collection (non-migratable task) and data preprocessing in the applications are divided into non-offloadable tasks, because the data collection in the applications, such as motion recognition, receives data from a motion sensor in real time, redundant data is generated when motion is unchanged, so that the data preprocessing needs to be performed locally in the wearable terminal to reduce unnecessary network transmission; the current action recognition is based on a machine learning model SVM for classification, the output result of a data fusion application is also a category, the face detection and the target detection application are both realized based on a deep learning model convolutional neural network, and all the face detection and the target detection application need to be preprocessed to construct features or readjust input data (such as pictures) into the input size specified by a model, and the preprocessed data is smaller than the preprocessed data, so that the data preprocessing is also carried out locally on a wearable terminal. The model reasoning part of the application is not connected with specific hardware of the terminal, and for the reliability of application execution, the application can choose to unload the whole reasoning part of the model and not divide the model.
In this way, the application to be executed is divided into the task which can be unloaded and the task which can not be unloaded and are (approximately) independent of each other through the division, so that the subsequent task unloading can be conveniently carried out, and the execution speed of the application to be executed is increased.
Optionally, as shown in fig. 3, the step S20 of establishing a communication connection with the edge service terminal includes:
s21, scanning communication access points of the edge service terminals in the current range;
the current range of the wearable terminal is a certain range of the position at the current moment, and the coverage range of the current range may be determined by the scanning radius or may be determined by other parameters.
It should be noted that, since the minimum time interval of WiFi scanning by the android system and the like also needs 10 seconds, instead of scanning every time there is a task that can be offloaded, the scanning may be performed again only when the information of the communication access point of the edge service terminal changes.
S22, obtaining the signal intensity of each communication access point;
s23, selecting the communication access point with the signal strength larger than the preset strength, and establishing a communication connection relation.
Taking WiFi as an example, weak WiFi signals may cause that the established communication connection cannot perform effective information transfer, so that by selecting the communication access point with signal strength greater than the preset strength, the incapability of smoothly performing the task capable of being unloaded caused by the weak signals can be avoided, and the speed and time for performing the task capable of being unloaded can be improved.
Optionally, as shown in fig. 4, the step S30 of selecting the edge service terminal according to the wearable terminal status information and the edge service terminal status information and sending an offloading request of the offloadable task includes:
s31, setting execution time weight and execution energy consumption weight of an application to be executed;
the evaluation direction of the unloading strategy of the application to be executed is the execution energy consumption or the execution time generated by unloading the application to be executed, and in the step, the execution energy consumption and the execution time can be integrated by setting the execution time weight and the execution energy consumption weight to evaluate the unloading strategy, so that the decision purpose of comprehensively considering the execution energy consumption or the execution time is achieved.
S32, establishing benefit constraint of an application to be executed according to the wearable terminal state information, the edge service terminal state information, the execution time weight and the execution energy consumption weight;
the benefit of the application to be executed is that the time weight and the execution energy consumption weight are distributed, so that the distribution mode of the task capable of unloading the wearable terminal is converted into quantifiable data which can be corresponding to the task capable of unloading the wearable terminal, and the advantages and disadvantages of the unloading strategy can be directly evaluated through the quantifiable data.
And establishing a benefit constraint of the application to be executed, and establishing a corresponding relation between the allocation of the task capable of unloading and the data capable of quantifying, so that the allocation of the task capable of unloading can be continuously optimized through the corresponding relation.
S33, solving the benefit constraint with the maximum benefit of the application to be executed as a target to obtain a solving result;
and the solving result is an edge service terminal allocated to each task capable of being unloaded, namely whether the task capable of being unloaded requested by the user is executed at a certain edge service terminal or not.
S34, determining the selected edge service terminal according to the solving result, and sending an unloading request of the task capable of unloading.
In this way, by setting the execution time weight and the execution energy consumption weight, the execution energy consumption and the execution time can be comprehensively evaluated to evaluate the unloading strategy, so that the decision purpose of comprehensively considering the execution energy consumption or the execution time is achieved, and the distributed edge service terminals of each unloading task under the condition of comprehensively considering the execution energy consumption and the execution time are determined by establishing the profit constraint and the solution, so that proper unloading mechanisms are respectively selected according to different calculation complexity and data quantity, the execution energy consumption and the execution time are effectively reduced, and the execution success rate is improved.
Optionally, the benefit constraint of the application to be executed is:
the maximum profit constraint of the application to be executed is as follows:
wherein,and->Weights expressed as user u for execution time and execution energy consumption of the application, +.>And->The optimized parameters are time delay and energy consumption respectively; />For the local execution time of an application, +.>For offloading to edge service terminals; />For the local execution of an application, energy consumption, +.>For unloading energy consumption; />For wireless transmission power, < >>The time for transmitting application execution data to the edge service terminal for the wearable terminal; the attribute of application i offloaded by user u is +.>Wherein d is i The amount of data needed for execution of application i is tarnished u,s For radio transmission speed x u,s Indicating whether the application currently requested to be uninstalled by user u is executed at edge service terminal s, x u,s =1 denotes execution at edge, x u,s =0 means performed at the mobile end; />For a given time limit; s is an edge end node set, U is a user set, and I is an application set.
In this way, through the benefit constraint, the corresponding relation between the distribution of the task capable of being unloaded and the data capable of being quantized is established, so that the distribution of the task capable of being unloaded can be continuously optimized through the corresponding relation, or the maximum benefit can be directly solved, and the edge service terminal of the distribution of each task capable of being unloaded under the condition of comprehensively considering the execution energy consumption and the execution time is determined, thereby effectively reducing the execution energy consumption and the execution time, and improving the execution success rate.
The embodiment of the disclosure provides a wearable terminal for executing the edge unloading decision method provided by the invention, wherein the wearable terminal comprises a computer readable storage medium and a processor, wherein the computer readable storage medium stores a computer program, and the computer program is read and run by the processor to realize the edge unloading decision method.
In this way, the wearable terminal performs primary allocation direction on the task to be allocated, and then the edge service terminal performs secondary analysis according to own computing resources to give feedback to the task which can be executed, so that the situation that the application of the wearable terminal is blocked due to incapability of executing the task which needs to be executed by the edge service terminal in time caused by excessive task which can be executed by the edge service terminal is avoided.
The embodiment of the disclosure provides an edge unloading decision-making method, and a device for executing the method can be integrated in electronic equipment such as a mobile phone, a pad, a computer, a server and the like. FIG. 5 is a flow chart of an edge service terminal side edge offloading decision method according to an embodiment of the invention; the edge unloading decision method is applied to an edge service terminal and comprises the following steps:
S100, monitoring state information of an edge service terminal;
the edge service terminal state information comprises computing resources and communication resource states of the edge service terminal.
S200, establishing communication connection with a wearable terminal and sending the state information of the edge service terminal;
the communication connection with the wearable terminal can be WiFi connection or MQTT connection.
S300, after receiving an unloading request of an unloading task sent by the wearable terminal, selecting the unloading task according to a time delay limiting condition, and sending an unloading instruction of the unloading task;
the edge service terminal receives unloading requests of the plurality of wearable terminals, analyzes the unloading requests, and can send unloading instructions to the corresponding wearable terminals if the edge service terminal decides to receive part or all of the unloading tasks.
Therefore, by setting the uninstalling instruction, the edge service terminal can determine the to-be-executed uninstalled task according to the self computing resource, computing capacity and other parameters, so that the situation that the processing speed is slow and the execution speed of the wearable terminal is slow due to the fact that the to-be-executed uninstalled task is too much is avoided.
In addition, by setting the uninstalling instruction, the edge service terminal can be used as an individual to analyze the uninstalling task which can be executed by the edge service terminal, so that the problem of distribution confusion caused by a plurality of wearable terminals is avoided.
S400, after receiving the task information capable of being unloaded sent by the wearable terminal, executing the task capable of being unloaded to obtain an execution result;
the wearable terminal only receives the unloading instruction fed back by the edge service terminal, and then sends the information of the task which can be unloaded and is involved in the unloading instruction, so that the situation that the task which can not be unloaded and is processed is sent is avoided.
It should be noted that, the task information that can be sent may be information after the wearable terminal performs data preprocessing (such as compaction, sorting, screening, etc.), so that data that needs to be transmitted may be greatly reduced, thereby reducing transmission time and execution time of the edge service terminal, and improving application execution time of the wearable terminal.
S500, the execution result is sent to the edge service terminal.
Therefore, by setting the uninstalling instruction, the edge service terminal can determine the to-be-executed uninstalled task according to the self computing resource, computing capacity and other parameters, so that the situation that the processing speed is slow and the execution speed of the wearable terminal is slow due to the fact that the to-be-executed uninstalled task is too much is avoided.
Optionally, as shown in fig. 6, S300, after receiving an offloading request of an offloadable task sent by the wearable terminal, selects the offloadable task according to a time delay constraint condition, and sends an offloading instruction of the offloadable task, including:
s310, after receiving an unloading request of an unloading task sent by the wearable terminal, counting the received unloading request of the unloading task;
after the wearable terminal selects the optimal edge service terminal and judges whether the task which can be offloaded to the edge service terminal can be offloaded, each edge service terminal can acquire the offloading requests of a plurality of tasks which can be offloaded of a plurality of wearable terminals, and the offloading requests need to be counted, so that the follow-up decision is facilitated.
S320, setting time delay limiting conditions of the edge service terminal;
the bandwidth of the edge service terminal is limited, if more wearable terminals are unloaded simultaneously, the time of the wearable terminals for transmitting data to the edge service terminal may exceed the time limit of application unloading, so that the application unloading is failed; meanwhile, the resource allocation strategy of the edge service terminal after being adjusted according to different queuing conditions of the container application service in the last time slot is not suitable for the current queuing conditions of the container application service, so that the queuing time and the service time of certain container application service are overlong.
The time delay limiting condition of the edge service terminal is a time limiting condition for unloading tasks, so that unloading failure caused by unloading overtime is avoided.
S330, establishing an unloading benefit constraint of the edge service terminal according to the counted unloading request of the unloading task and the time delay constraint condition of the edge service terminal;
and defining the unloading benefits of the edge service terminal as the accumulated sum of the unloading benefits of all the wearable terminals according to the weight values, so that the distribution mode of the unloading tasks of the edge service terminal is converted into the corresponding quantifiable data, and the advantages and disadvantages of the unloading strategy can be directly evaluated through the quantifiable data.
And establishing a benefit constraint of the application to be executed, and establishing a corresponding relation between the allocation of the task capable of unloading and the data capable of quantifying, so that the allocation of the task capable of unloading can be continuously optimized through the corresponding relation.
And the corresponding relation between the distribution of the off-loadable tasks and the data before the quantifiable data can be established by constraint, so that the distribution of the off-loadable tasks can be continuously optimized through the corresponding relation.
S340, solving the unloading benefit constraint with the maximum unloading benefit of the edge service terminal as a target to obtain a solving result;
And the solving result is whether the distribution of each task capable of being unloaded is received or not, namely whether the task capable of being unloaded which is requested to be unloaded currently is executed at the edge service terminal or not.
S350, determining the selected task capable of being unloaded according to the solving result of the unloading benefit constraint, and sending an unloading instruction of the task capable of being unloaded.
In this way, by establishing the offloading benefit constraint and solving, the offloading task and the offloading sequence which can be offloaded are determined according to the bandwidth resource and the application offloading benefit, and appropriate offloading mechanisms are respectively selected according to different calculation complexity and data quantity, so that the situation that the processing speed is slow and the execution speed of the wearable terminal is slow due to excessive offloading tasks to be executed is avoided, and the execution efficiency is improved.
Optionally, the offloading benefit constraint of the edge service terminal is:
the maximum constraint of the offloading gain of the edge service terminal is as follows:
wherein pi u For the priority of edge service terminals to wearable end users,and->Expressed as weights of user u on application execution time and execution energy consumption, y u Indicating whether the unloading request of the mobile terminal user u is allowed or not, and if the unloading request is 1, indicating that the application which allows the mobile terminal user u to be unloaded currently is executed on the edge terminal; the attribute of application i offloaded by user u is +. >d i The amount of data needed for execution of application i, c i To apply the floating point computational complexity of the inference part of the i model, the obscuration u,s For the acquired wireless transmission speed, +.>Weight for user execution time for application, +.>Local execution time for the application; />For wireless transmission power, < >>The energy consumption is executed for the application locally; f (f) s,i Computing resources allocated for the application i for the edge service terminal s; />Latency at edge service terminal s for the offload application of user u; />For a given time limit, RB s The total bandwidth available to the edge end node, f s Computing resources for the edge node total; u (U) s The user set is an edge service terminal s, and the application set is I.
In this way, through the offloading benefit constraint, the corresponding relation between the allocation of the offloading tasks and the front of the quantifiable data is established, so that the allocation of the computing resources of the edge service terminal can be continuously optimized through the corresponding relation, or the maximum benefit is directly solved, and the offloading tasks and the offloading sequence which can be offloaded are determined.
The embodiment of the disclosure provides an edge service terminal for executing the edge unloading decision method according to the invention, which comprises a computer readable storage medium storing a computer program and a processor, wherein the computer program is read and run by the processor to realize the edge unloading decision method.
Therefore, by setting the uninstalling instruction, the edge service terminal can determine the to-be-executed uninstalled task according to the self computing resource, computing capacity and other parameters, so that the situation that the processing speed is slow and the execution speed of the wearable terminal is slow due to the fact that the to-be-executed uninstalled task is too much is avoided.
The embodiment of the invention also discloses a computer readable storage medium which stores a computer program, and when the computer program is read and run by a processor, the edge unloading decision method is realized, or the edge unloading decision method is realized.
In this way, the wearable terminal performs primary allocation direction on the task to be allocated, and then the edge service terminal performs secondary analysis according to own computing resources to give feedback to the task which can be executed, so that the situation that the application of the wearable terminal is blocked due to incapability of executing the task which needs to be executed by the edge service terminal in time caused by excessive task which can be executed by the edge service terminal is avoided.
Although the present invention is disclosed above, the present invention is not limited thereto. Various changes and modifications may be made by one skilled in the art without departing from the spirit and scope of the invention, and the scope of the invention should be assessed accordingly to that of the appended claims.

Claims (6)

1. An edge offloading decision-making method applied to a wearable terminal, comprising:
s10, acquiring an offloadable task and monitoring state information of a wearable terminal;
s20, establishing communication connection with an edge service terminal, and receiving edge service terminal state information sent by the edge service terminal;
s30, selecting the edge service terminal according to the wearable terminal state information and the edge service terminal state information and sending an unloading request of the unloading task;
s40, after receiving an unloading instruction sent by the edge service terminal, sending information of an unloading task to the edge service terminal;
s50, receiving an execution result obtained by executing the task capable of unloading, which is sent by the edge service terminal;
wherein, the step S20 of establishing communication connection with the edge service terminal includes: s21, scanning the communication access points of the edge service terminals in the current range; s22, obtaining the signal intensity of each communication access point; s23, selecting the communication access point with the signal strength larger than the preset strength, and establishing a communication connection relation;
The step S30, selecting the edge service terminal according to the wearable terminal status information and the edge service terminal status information, and sending an offloading request of the offloading task, includes: s31, setting execution time weight and execution energy consumption weight of an application to be executed; s32, establishing benefit constraint of an application to be executed according to the wearable terminal state information, the edge service terminal state information, the execution time weight and the execution energy consumption weight; s33, solving the benefit constraint with the maximum benefit of the application to be executed as a target to obtain a solving result; s34, determining the selected edge service terminal according to the solving result, and sending an unloading request of the task capable of unloading;
wherein the benefit constraint of the application to be executed is:
the maximum profit constraint of the application to be executed is as follows:
wherein,and->Weights for user u for application execution time and execution energy consumption +.>And->The optimized parameters are time delay and energy consumption respectively; />For the local execution time of an application, +.>For offloading to edge service terminals; />For the local execution of an application, energy consumption, +.>For unloading energy consumption; / >For wireless transmission power, < >>The time for transmitting application execution data to the edge service terminal for the wearable terminal;the attribute of application i offloaded by user u is +.>Wherein d is i The amount of data needed for application i execution, r u,s For radio transmission speed x u,s Indicating whether the application currently requested to be uninstalled by user u is executed at edge service terminal s, x u,s =1 denotes execution at edge, x u,s =0 means performed at the mobile end; />For a given time limit; s is an edge end node set, U is a user set, and I is an application set.
2. The edge offload decision-making method according to claim 1, wherein S10, acquiring an offloadable task includes:
s11, acquiring a current application to be executed;
s12, dividing the application to be executed into a plurality of tasks to be executed, and dividing the tasks to be executed into the task to be offloadable and the task not to be offloadable according to the execution process of the tasks to be executed.
3. A wearable terminal comprising a computer readable storage medium storing a computer program and a processor, the computer program implementing the edge offload decision method of any of claims 1-2 when read and executed by the processor.
4. An edge offloading decision-making method applied to an edge service terminal is characterized by comprising the following steps:
s100, monitoring state information of an edge service terminal;
s200, establishing communication connection with a wearable terminal and sending the state information of the edge service terminal;
s300, after receiving an unloading request of an unloading task sent by the wearable terminal, selecting the unloading task according to a time delay limiting condition, and sending an unloading instruction of the unloading task;
s400, after receiving the task information capable of being unloaded sent by the wearable terminal, executing the task capable of being unloaded to obtain an execution result;
s500, sending the execution result to the edge service terminal;
the step 300 of selecting the task capable of being offloaded according to a time delay constraint condition after receiving the task capable of being offloaded sent by the wearable terminal and sending an instruction for offloading the task capable of being offloaded, includes:
s310, after receiving an unloading request of an unloading task sent by the wearable terminal, counting the received unloading request of the unloading task;
s320, setting time delay limiting conditions of the edge service terminal;
S330, establishing an unloading benefit constraint of the edge service terminal according to the counted unloading request of the unloading task and the time delay constraint condition of the edge service terminal;
s340, solving the unloading benefit constraint with the maximum unloading benefit of the edge service terminal as a target to obtain a solving result;
s350, determining the selected task capable of being unloaded according to the solving result of the unloading benefit constraint, and sending an unloading instruction of the task capable of being unloaded;
the offloading benefit constraint of the edge service terminal is as follows:
the maximum constraint of the offloading gain of the edge service terminal is as follows:
wherein pi u For the priority of edge service terminals to wearable end users,and->For user u, weight of application execution time and execution energy consumption, y u Indicating whether the unloading request of the mobile terminal user u is allowed or not, and if the unloading request is 1, indicating that the application which allows the mobile terminal user u to be unloaded currently is executed on the edge terminal; the attribute of application i offloaded by user u is +.>d i The amount of data needed for execution of application i, c i To apply the floating point calculation amount of the reasoning part of the i model, r u,s For the acquired wireless transmission speed, +.>Weight for user execution time for application, +. >Local execution time for the application; />For wireless transmission power, < >>The energy consumption is executed for the application locally; f (f) s,i Computing resources allocated for the application i for the edge service terminal s; />Latency at edge service terminal s for the offload application of user u; />For a given time limit, RB s The total bandwidth available to the edge end node, f s Computing resources for the edge node total; u (U) s The user set is an edge service terminal s, and the application set is I.
5. An edge service terminal comprising a computer readable storage medium storing a computer program and a processor, the computer program implementing the edge offload decision method of claim 4 when read and executed by the processor.
6. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program which, when read and executed by a processor, implements the edge offload decision method of any of claims 1-2 or implements the edge offload decision method of claim 4.
CN202011127194.3A 2020-10-20 2020-10-20 Edge unloading decision method, terminal and readable storage medium Active CN112162863B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011127194.3A CN112162863B (en) 2020-10-20 2020-10-20 Edge unloading decision method, terminal and readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011127194.3A CN112162863B (en) 2020-10-20 2020-10-20 Edge unloading decision method, terminal and readable storage medium

Publications (2)

Publication Number Publication Date
CN112162863A CN112162863A (en) 2021-01-01
CN112162863B true CN112162863B (en) 2024-04-02

Family

ID=73867724

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011127194.3A Active CN112162863B (en) 2020-10-20 2020-10-20 Edge unloading decision method, terminal and readable storage medium

Country Status (1)

Country Link
CN (1) CN112162863B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113125675A (en) * 2021-04-19 2021-07-16 北京物资学院 Storage yard coal spontaneous combustion early warning device and early warning method of edge computing framework
CN113037805B (en) * 2021-05-11 2022-07-08 福州大学 Multi-instance micro-service migration method facing mobile edge calculation
CN113342418B (en) * 2021-06-24 2022-11-22 国网黑龙江省电力有限公司 Distributed machine learning task unloading method based on block chain
CN116132436A (en) * 2021-11-15 2023-05-16 中国移动通信有限公司研究院 Task unloading method, device, equipment and readable storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016161677A1 (en) * 2015-04-07 2016-10-13 北京邮电大学 Traffic offload method and system
CN110096318A (en) * 2019-05-08 2019-08-06 北京邮电大学 A kind of task discharging method and device based on mobile edge calculations
CN111372314A (en) * 2020-03-12 2020-07-03 湖南大学 Task unloading method and task unloading device based on mobile edge computing scene

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016161677A1 (en) * 2015-04-07 2016-10-13 北京邮电大学 Traffic offload method and system
CN110096318A (en) * 2019-05-08 2019-08-06 北京邮电大学 A kind of task discharging method and device based on mobile edge calculations
CN111372314A (en) * 2020-03-12 2020-07-03 湖南大学 Task unloading method and task unloading device based on mobile edge computing scene

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
移动边缘计算环境中基于能耗优化的深度神经网络计算任务卸载策略;高寒;李学俊;周博文;刘晓;徐佳;;计算机集成制造系统;20200615(第06期);全文 *

Also Published As

Publication number Publication date
CN112162863A (en) 2021-01-01

Similar Documents

Publication Publication Date Title
CN112162863B (en) Edge unloading decision method, terminal and readable storage medium
Abd Elaziz et al. IoT workflow scheduling using intelligent arithmetic optimization algorithm in fog computing
CN111212106B (en) Edge computing task processing and scheduling method and device in industrial internet environment
CN105159782A (en) Cloud host based method and apparatus for allocating resources to orders
CN114816721B (en) Multitask optimization scheduling method and system based on edge calculation
CN112261120B (en) Cloud-side cooperative task unloading method and device for power distribution internet of things
CN111506434B (en) Task processing method and device and computer readable storage medium
CN113961264B (en) Intelligent unloading algorithm and system for video monitoring cloud edge cooperation
CN110213327A (en) A kind of resource regulating method based on edge calculations, apparatus and system
CN112966608A (en) Target detection method, system and storage medium based on edge-side cooperation
CN114489977A (en) Task unloading method and device for mobile edge computing system
CN113553160A (en) Task scheduling method and system for edge computing node of artificial intelligence Internet of things
CN106570011B (en) Distributed crawler URL seed distribution method, scheduling node and capturing node
CN117271100B (en) Algorithm chip cluster scheduling method, device, computer equipment and storage medium
CN114697324A (en) Real-time video analysis and processing method based on edge cloud cooperation
CN112559078B (en) Method and system for hierarchically unloading tasks of mobile edge computing server
CN112306696B (en) Energy-saving and efficient edge computing task unloading method and system
KR102350195B1 (en) Energy Optimization Scheme of Mobile Devices for Mobile Augmented Reality Applications in Mobile Edge Computing
CN117135131A (en) Task resource demand perception method for cloud edge cooperative scene
CN114489963A (en) Management method, system, equipment and storage medium of artificial intelligence application task
CN116627771A (en) Log acquisition method, device, electronic equipment and readable storage medium
CN114936089A (en) Resource scheduling method, system, device and storage medium
CN116069498A (en) Distributed computing power scheduling method and device, electronic equipment and storage medium
CN114035906A (en) Virtual machine migration method and device, electronic equipment and storage medium
CN113115072A (en) Video target detection tracking scheduling method and system based on end cloud 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