CN116089091A - Resource allocation and task unloading method based on edge calculation of Internet of things - Google Patents

Resource allocation and task unloading method based on edge calculation of Internet of things Download PDF

Info

Publication number
CN116089091A
CN116089091A CN202310119925.7A CN202310119925A CN116089091A CN 116089091 A CN116089091 A CN 116089091A CN 202310119925 A CN202310119925 A CN 202310119925A CN 116089091 A CN116089091 A CN 116089091A
Authority
CN
China
Prior art keywords
task
edge server
resource allocation
computing
terminal
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.)
Pending
Application number
CN202310119925.7A
Other languages
Chinese (zh)
Inventor
谢昊飞
段如兵
何莉
王平
洪承镐
李昭
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.)
Chongqing University of Post and Telecommunications
Original Assignee
Chongqing University of Post and Telecommunications
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 Chongqing University of Post and Telecommunications filed Critical Chongqing University of Post and Telecommunications
Priority to CN202310119925.7A priority Critical patent/CN116089091A/en
Publication of CN116089091A publication Critical patent/CN116089091A/en
Pending legal-status Critical Current

Links

Images

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
    • G06F9/485—Task life-cycle, e.g. stopping, restarting, resuming execution
    • 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
    • G06F9/505—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 considering the load
    • 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]
    • G06F9/5061—Partitioning or combining of resources
    • G06F9/5072—Grid computing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming

Abstract

The invention relates to a resource allocation and task unloading method based on the edge calculation of the Internet of things, which belongs to the technical field of the Internet of things and comprises the following steps: s1: constructing an edge computing system of the Internet of things based on an edge server; s2: constructing an effect function of the system; s3: decomposing the system effect function into a resource allocation optimization function under the initial given task unloading decision and a task unloading optimization function based on a resource allocation optimization result; s4: secondarily decomposing the resource allocation optimization function into a power allocation optimization function of the terminal user equipment and a computing resource allocation optimization function of the edge server; s5: solving to obtain an optimal transmission power distribution scheme of the terminal user equipment; s6: solving to obtain an optimal computing resource allocation scheme; s7: and (3) bringing the optimal allocation scheme back to the original problem system effect function, and solving to obtain an optimal task unloading strategy.

Description

Resource allocation and task unloading method based on edge calculation of Internet of things
Technical Field
The invention belongs to the technical field of the Internet of things, and relates to a resource allocation and task unloading method based on the edge calculation of the Internet of things.
Background
With the continuous development of the internet of things, a large number of terminal devices of the internet of things are accessed, the traditional mode of taking a centralized server as a data processing center is difficult to meet the service quality requirement of the internet of things, the centralized server system not only causes a large amount of resource waste and communication delay, but also cannot respond to delay sensitive tasks in time, for the internet of things, the occurrence of edge calculation means that a plurality of control functions are completed through local devices without depending on a cloud, task processing is completed in an edge calculation layer, and the core idea is that the processing of data is closer to an edge terminal. Unlike available centralized system, the distributed computing is closer to the source of task generation, and is favorable to the management of industrial production process, and the edge server is mainly used in data aggregation, data transmission and other task processing.
Because the resources and the computing power of the edge computing node are limited, only partial tasks can be processed, a large number of edge devices in the Internet of things are accessed, the generation of a large number of tasks brings serious challenges to the management of limited resources of an edge server, the computing and unloading process involves computing and other resources, and unreasonable computing and unloading strategies can cause resource waste and even performance loss. Designing rational computational offloading and resource allocation strategies by balancing computational and communication overhead is key to exploiting edge computational potential. The problems of low resource utilization rate, large time delay, unbalanced load and the like caused by unreasonable resource allocation are solved, and in the face of the problems, how to reasonably dispatch the terminal tasks to the corresponding edge computing nodes for unloading, and the problem of resource allocation of a plurality of edge servers to be solved, the key of realizing efficient task allocation is to execute resource allocation and unloading decision. Some researches are from resource allocation through a deployment strategy of an edge server, but the position of the edge server in some scenes of the internet of things is fixed; still other studies currently involve single resource allocation, mainly focused on allocation of computational or communication resources, without joint consideration of task offloading, and most studies do not consider task offloading policies and resource allocation jointly. However, the task offloading and resource allocation of edge computation are generally related and must be considered jointly.
Disclosure of Invention
In view of the above, the present invention aims to provide a resource allocation and task offloading method based on internet of things edge computation, which solves the task offloading and resource allocation problems of multiple internet of things terminals, multiple tasks and multiple edge servers by jointly optimizing task offloading and resource allocation.
In order to achieve the above purpose, the present invention provides the following technical solutions:
a resource allocation and task offloading method based on the edge calculation of the Internet of things comprises the following steps:
s1: an edge server-based internet of things edge computing system is constructed, and the system comprises a task unloading distribution processing model, a computing resource and power consumption distribution model and a time delay model;
s2: based on the optimization target of resource allocation, unloading, power and time delay are combined, and the task completion time delay of all terminal user equipment and the energy consumption weight of the equipment are combined to construct an effect function of the system;
s3: decomposing a joint optimization problem of resource allocation and task unloading, and decomposing the system effect function into a resource allocation optimization function under an initial given task unloading decision and a task unloading optimization function based on a resource allocation optimization result;
s4: the resource allocation optimization function under the initial given task unloading decision is secondarily decomposed into a power allocation optimization function of the terminal user equipment and a computing resource allocation optimization function of the edge server;
S5: according to the characteristics of the power distribution optimization function of the terminal user equipment, adopting a dichotomy to solve to obtain an optimal transmission power distribution scheme of the terminal user equipment;
s6: solving by using KKT conditions according to the characteristics of the computing resource allocation optimization function of the edge server to obtain an optimal computing resource allocation scheme;
s7: and (3) after the optimal transmission power distribution scheme and the optimal calculation distribution scheme obtained in the steps S5 and S6 are carried back to the original problem system effect function, and the optimal task unloading strategy is obtained by solving the beneficial user-hybrid improved genetic algorithm.
Further, in step S1, the task unloading allocation processing model is as follows:
the method comprises the steps of setting an Internet of things terminal equipment set to be E= {1,2, …, E }, assuming that each Internet of things terminal can generate a calculation task within a time period, and the calculation task is independently completed, cannot be divided and is marked as T e ＝{d e ，c e }，d e Representing the size of the task input data, c e Representing the workload of a task, namely the calculated amount, in particular the CPU period required by calculating the task;
the edge server set is expressed as M= {1,2, …, M }, the uplink transmission adopts orthogonal frequency division multiple access as an access mode of uploading a terminal task to the edge server, all working frequency bands are divided into N mutually orthogonal channels, and N= {1,2, …, N } is the total number of the mutually orthogonal channels;
The task offloading decision set is expressed as:
Figure BDA0004079661310000021
wherein
Figure BDA0004079661310000022
A decision binary variable representing the offloading of the task of terminal device e to edge server m, wherein,
Figure BDA0004079661310000023
indicating that terminal task e is offloaded to edge server m via channel j, otherwise +.>
Figure BDA0004079661310000024
The set of end users for which tasks need to be offloaded is represented as:
E off ＝e m∈M E m
for any one terminal task to either execute locally or offload to an edge server, the following constraints are met:
Figure BDA0004079661310000031
the number of end users that any edge server m can serve is constrained by the number of channels available:
Figure BDA0004079661310000032
wherein qm Representing the number of terminal devices that can be served by one edge server at the same time at most;
for any edge server m, any channel n can be allocated to at most one end user device, so the following constraints are met:
Figure BDA0004079661310000033
/>
further, the computing resource and power consumption allocation model is:
the set of transmission power allocation policies for the end user device is expressed as:
Figure BDA0004079661310000034
wherein Pe Indicating that terminal e is to discharge the taskTransmission power, P, carried to edge server emax Representing the maximum transmission power corresponding to the terminal equipment, E off A terminal device set offloaded to a server for a computing task;
the set of allocation policies for computing resources is expressed as:
F＝{f em |e∈E,m∈M}
wherein fem Representing the computing resources allocated by the edge server m to the terminal device e;
the computing resource allocation of each edge server cannot exceed its own maximum computing power, with the following constraints:
Figure BDA0004079661310000035
meaning that for all terminal tasks assigned to any one edge server, the sum of the computing resources assigned to it by the edge server cannot exceed its own maximum computing power, where
Figure BDA0004079661310000036
Representing the maximum computing power of the edge server m, in particular the number of CPU cycles per second, E m Representing the set of all end user devices accessing the edge server m.
Further, the time delay model is:
the time for the terminal task to be completed consists of two parts, namely the time for the terminal user equipment to be completed locally or the time required by the terminal user equipment to be executed at the far end of the edge server;
Figure BDA0004079661310000037
the task of the terminal e is represented by the local completion time:
Figure BDA0004079661310000038
wherein ,
Figure BDA0004079661310000039
representing the own computing power of the end user device, in particular the number of CPU cycles per second that can be run, c e A workload, i.e., the number of CPU cycles required, for a computing task; t is t e The time required for the task of terminal e to be offloaded to the edge server remote execution is represented by +.>
Figure BDA0004079661310000041
and />
Figure BDA0004079661310000042
Composition (S)/(S)>
Figure BDA0004079661310000043
For the time delay in the task upload to the edge server,
Figure BDA0004079661310000044
the required latency is calculated in the edge server for the task:
Figure BDA0004079661310000045
Figure BDA0004079661310000046
Figure BDA0004079661310000047
wherein xem Representing the task offload variables,
Figure BDA0004079661310000048
indicating the uplink transmission rate of the terminal task,
Figure BDA0004079661310000049
the following are provided:
Figure BDA00040796613100000410
further, the effect function of the system in step S2 is:
Figure BDA00040796613100000411
wherein ,
Figure BDA00040796613100000412
E loc representing a locally computed set of end user devices, E off Representing an offloaded set of end user devices, weighting factor +.>
Figure BDA00040796613100000413
and />
Figure BDA00040796613100000414
Belonging to the range 0 to 1 and->
Figure BDA00040796613100000415
If the task is urgent, the user can increase the delay weight +.>
Figure BDA00040796613100000416
Otherwise, the terminal can increase the energy consumption weighting factor under the condition of low electric quantity>
Figure BDA00040796613100000417
The computational offloading decision and resource allocation joint system effect optimization problem is expressed as:
Figure BDA00040796613100000418
Figure BDA00040796613100000419
Figure BDA00040796613100000420
Figure BDA00040796613100000421
Figure BDA00040796613100000422
Figure BDA00040796613100000423
Figure BDA00040796613100000424
Figure BDA00040796613100000425
wherein Q represents a system benefit function, X represents a task offloading decision set, P represents a transmission power allocation decision set, F represents a computing resource allocation decision set, and constraint Cl represents an offloading decision as a binary variable; constraint C2 indicates that the task of the terminal of the Internet of things is unloaded to an edge server to be executed or is executed locally; constraint C3 indicates that each edge server's sub-bandwidth channels can be allocated to at most one terminal device; constraint C4 represents the number of terminal devices that can offload tasks to the edge server at most; constraint C5 is the power constraint of the vehicle, P e Representing the transmission power, P, of terminal device e offloading tasks to edge server emax Representing the maximum transmission power corresponding to the terminal equipment; constraint C6 indicates that the computing resources allocated by the edge server to the terminal device must be positive, f em Representing the computing resources allocated by the edge server m to the terminal device e; constraint C7 represents the task site offloaded to the edge serverThe sum of the required computing resources does not exceed the computing power of the edge server, f m Representing the computing resources owned by edge server m.
Further, the step S3 specifically includes:
the system effect function is decomposed into a resource allocation optimization function under the initial task unloading decision and a task unloading optimization function based on a resource allocation optimization result, which are respectively expressed as:
Figure BDA0004079661310000051
Figure BDA0004079661310000052
Figure BDA0004079661310000053
/>
Figure BDA0004079661310000054
Figure BDA0004079661310000055
where V (X) represents an optimal value function of the transmission power allocation problem of the end user device and the computing resource allocation problem of the edge server under a specific offload task decision, as follows:
Figure BDA0004079661310000056
Figure BDA0004079661310000057
Figure BDA0004079661310000058
Figure BDA0004079661310000059
further, the power allocation optimization function of the end user device in step S4 is expressed as:
Figure BDA00040796613100000510
Figure BDA00040796613100000511
the computing resource allocation optimization function of the edge server is expressed as:
Figure BDA0004079661310000061
Figure BDA0004079661310000062
Figure BDA0004079661310000063
further, in step S5, the optimal transmission power allocation scheme of the end user device is obtained by solving the characteristics of the power allocation optimization function of the end user device by using a dichotomy, which specifically includes:
The power distribution optimizing function features that it is one strict quasi-convex function with optimal solution in the constraint boundary or zero crossing of the first derivative, the binary method includes setting initial interval of initial power and its tolerance, judging whether the boundary value is the optimal solution of the quasi-convex function, if not, performing binary calculation, and calculating the first order of quasi-convex function in each binary iterationThe value of the derivative, making a decision as to whether or not it is the optimal solution, until a solution to the transmission power allocation problem is obtained
Figure BDA0004079661310000064
Further, in step S6, the optimal computing resource allocation scheme is obtained by using KKT condition solution according to the characteristics of the computing resource allocation optimization function of the edge server, which specifically includes:
Figure BDA0004079661310000065
wherein ,
Figure BDA0004079661310000066
and representing the calculation resource allocation result under the target optimization function.
Further, in step S7, an optimal transmission power allocation scheme is obtained
Figure BDA0004079661310000067
And optimal computational allocation scheme->
Figure BDA0004079661310000068
And then, carrying back the original problem system effect function to obtain a task unloading optimization function, wherein the task unloading optimization function comprises the following steps of: />
Figure BDA0004079661310000069
s.t.C1:
Figure BDA00040796613100000610
m∈M,j∈N
Figure BDA00040796613100000611
Figure BDA00040796613100000612
Figure BDA00040796613100000613
Further, in step S7, to solve the unloading decision set X, a beneficiary user-hybrid improvement genetic algorithm is used to solve the problem of calculating the unloading decision, including whether the end user wants to unload and to which edge server, specifically as follows:
Firstly, the problem of whether an end user needs to be unloaded is solved, the difference value of all users through user spending is divided into an unloading user set and a local computing user set, and the difference value of the unloading computing and the local computing spending of the end user is defined as follows:
Figure BDA0004079661310000071
if U e <0 means that the end user e has less offloading overhead than the local computing overhead, and this class of users is defined as beneficiary users who are to be offloaded; if U e More than or equal to 0 means that the offloading is more expensive than the local computation, such end users are defined as non-beneficiary users, such users do not offload, so such user offloading decisions are performed locally;
after solving the unloading user and the local user set, solving the edge server matching problem of all the unloading users, namely the unloading user is unloaded to which edge server, and solving the unloading decision of the part through a genetic algorithm.
The invention has the beneficial effects that: according to the method, the resource allocation and the task unloading of the Internet of things are determined from the unloading decision, the transmission power allocation and the calculation resource allocation, the method for unloading the tasks of the terminals of the Internet of things and optimally allocating the resources of the edge servers in the Internet of things scene is realized, the optimal solution of the resource allocation is solved by respectively using the dichotomy under the specific unloading strategy, the KKT condition is brought back to the system effect function, the unloading strategy with the optimal system effect function is obtained according to the proposed beneficiary user-hybrid improved genetic algorithm, and the system performance is improved.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objects and other advantages of the invention may be realized and obtained by means of the instrumentalities and combinations particularly pointed out in the specification.
Drawings
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in the following preferred detail with reference to the accompanying drawings, in which:
FIG. 1 is a schematic diagram of an edge computing system of the Internet of things according to the present invention;
FIG. 2 is a flow chart of a method for resource allocation and task offloading based on edge computation of the Internet of things according to the present invention;
FIG. 3 is a flowchart of a beneficial user-hybrid improved genetic algorithm in accordance with the present invention.
Detailed Description
Other advantages and effects of the present invention will become apparent to those skilled in the art from the following disclosure, which describes the embodiments of the present invention with reference to specific examples. The invention may be practiced or carried out in other embodiments that depart from the specific details, and the details of the present description may be modified or varied from the spirit and scope of the present invention. It should be noted that the illustrations provided in the following embodiments merely illustrate the basic idea of the present invention by way of illustration, and the following embodiments and features in the embodiments may be combined with each other without conflict.
Wherein the drawings are for illustrative purposes only and are shown in schematic, non-physical, and not intended to limit the invention; for the purpose of better illustrating embodiments of the invention, certain elements of the drawings may be omitted, enlarged or reduced and do not represent the size of the actual product; it will be appreciated by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The same or similar reference numbers in the drawings of embodiments of the invention correspond to the same or similar components; in the description of the present invention, it should be understood that, if there are terms such as "upper", "lower", "left", "right", "front", "rear", etc., that indicate an azimuth or a positional relationship based on the azimuth or the positional relationship shown in the drawings, it is only for convenience of describing the present invention and simplifying the description, but not for indicating or suggesting that the referred device or element must have a specific azimuth, be constructed and operated in a specific azimuth, so that the terms describing the positional relationship in the drawings are merely for exemplary illustration and should not be construed as limiting the present invention, and that the specific meaning of the above terms may be understood by those of ordinary skill in the art according to the specific circumstances.
The invention provides a resource allocation and task unloading method based on internet of things edge calculation. The scheme is that the power consumption and the time delay of a system are considered in an optimization and combination mode, the time delay and the system effect function of the system are constructed, an optimal task unloading strategy is obtained through optimizing the system effect function, and because integer constraint exists, the optimization problem is a mixed integer nonlinear programming problem, the system effect function is firstly decomposed, the two sub-problems of resource allocation under a specific task unloading strategy and a task unloading scheme based on the resource allocation are decoupled, the resource allocation problem is further decomposed into two sub-problems of calculating resource allocation and power allocation, and the optimal task unloading strategy solution set is returned to be solved on the basis.
As shown in the figure 1 in the embodiment of the invention, the invention is an Internet of things edge computing system consisting of a plurality of base stations, a plurality of edge servers and a plurality of end users. Comprising the following steps:
an Internet of things terminal user layer: the intelligent terminal mainly comprises intelligent terminals such as sensor nodes, production equipment, intelligent sensing equipment and the like, is mainly powered by batteries, has relatively small computing capacity and relatively limited resources, generates computing tasks by the terminals, determines whether to unload the computing tasks to an edge server according to the computing amount of the generated tasks and the self resources, and can upload the tasks to the edge server through a wireless channel.
Edge server layer: the edge server device performs distributed computation, avoids uploading data to the cloud, thereby reducing task time delay and uploading energy consumption, has the advantage of computing resources compared with local computation of a terminal user, and can reduce energy consumption used by the local computation of the terminal user.
Cloud computing layer: the cloud computing platform mainly comprises a large number of cloud servers, and has the most computing and storage resources. The cloud computing layer is mainly used for processing tasks from the edge computing layer offload, and cloud computing can be parallelized to the tasks of the layer by means of huge computing and storage resources. However, since the cloud computing layer is generally geographically distant from the edge computing layer.
When the end user of the internet of things has tasks, the end user can select to execute calculation locally or unload to the edge server through a wireless channel to perform calculation, the local calculation has limited resources, the consumption of the power consumption of the terminal is large, and extra communication time delay and sending power consumption can be generated when the end user unloads to the edge server, so that the end user decides whether to unload the relationship between the end user and the edge server to be balanced or not, the time delay of the whole system is low, the power consumption of all the end users is reduced, and the benefit of the system is improved. This process involves offloading decisions for the terminal tasks, transmission power allocation, and computing resource allocation for the edge servers.
In order to solve the problem, we propose a resource allocation and task offloading method based on the edge calculation of the internet of things. As shown in fig. 2, includes:
s1: and constructing an edge server-based edge computing system architecture of the Internet of things, and constructing a computing time delay model, a power consumption model and a task unloading distribution processing model of the system.
S2: based on the optimization target of resource allocation, unloading, power and time delay are combined, and the task completion time delay of all terminal user equipment and the energy consumption weight of the equipment are combined to construct an effect function of the system;
s3: in order to facilitate the solution, the joint optimization problem of resource allocation and task offloading can be decomposed, and the system effect function described in S2 is decomposed into a resource allocation optimization function under the initial given task offloading decision and a task offloading optimization function based on the resource allocation optimization result;
s4: secondarily decomposing the resource allocation optimization function under the given initial task unloading decision into a power allocation optimization function of the terminal user equipment and a computing resource allocation optimization function of the edge server on the basis of the step S3;
s5: after the power distribution optimization function is obtained in the step S4, solving by adopting a dichotomy according to the characteristics of the power distribution optimization function to obtain an optimal transmission power distribution scheme of the terminal user equipment;
S6: after obtaining a computing resource allocation optimization function of the edge server in the step S4, solving by using KKT conditions according to the characteristics of the computing resource allocation optimization function to obtain an optimal computing resource allocation scheme;
s7: after the optimal transmission power allocation scheme and the optimal calculation allocation scheme obtained in steps S5 and S6, the original problem system effect function is brought back, and the optimal task offloading strategy is obtained by solving the beneficial user-hybrid improved genetic algorithm proposed herein.
In the resource allocation of the embodiment, because the resources of the terminal user of the internet of things are limited, the task of the terminal user is to be comprehensively considered when the terminal user locally executes or uninstalls to the edge server, and meanwhile, if the terminal user decides to reasonably allocate the transmission power of the terminal of the internet of things after the uninstallation, the terminal device energy consumption is saved and the importance is important.
For a plurality of edge servers, computing resources are reasonably allocated for a certain terminal task unloaded to the server to complete computing, so that a plurality of edge servers are prevented from processing too many tasks, and a plurality of servers are idle, and the allocation of a large number of tasks generated by a plurality of unloaded terminals is also a problem to be solved.
Therefore, time and energy consumption are two important indexes, the task processing time generated by the terminal has a certain time delay constraint, meanwhile, the power consumption of the terminal equipment for transmitting and unloading tasks to the edge server cannot be too large, other normal functions are affected, meanwhile, the task is unloaded to which edge server, and how much computing resource is allocated to a certain terminal task by the edge server are all problems to be considered, and here, we construct a weighted system effect function of the task completion time delay and the power consumption of the terminal equipment. The unloading strategy set X of all the terminals of the Internet of things, the transmission power distribution strategy P of all the terminal equipment and the edge server computing resource distribution set F are required, and the purpose is to minimize the weighted system effect function of time delay and power consumption.
The invention sets the set of the terminal equipment of the Internet of things to represent E, E= {1,2, …, E }, and assumes that each terminal of the Internet of things can generate a calculation task within a time period, and the calculation task is independently completed and cannot be divided. Denoted as T e ＝{d e ，c e },d e The unit of the input data representing the task is bits, c e Representing the task workload, i.e., the amount of computation, and the CPU cycles required to compute the task.
The set of edge servers M is denoted as m= {1,2, …, M }, in order to reduce channel interference between different user uploads, the uplink transmission adopts orthogonal frequency division multiple access as an access mode for terminal task uploading to the edge servers, all the working frequency bands are divided into N mutually orthogonal channels, n= {1,2, …, N },
Figure BDA0004079661310000101
decision binary variable representing the offloading of the task of terminal e to edge server m, wherein +.>
Figure BDA0004079661310000102
Indicating that terminal task e is offloaded to edge server m via channel j (j e N), otherwise +.>
Figure BDA0004079661310000103
E off ＝e m∈M E m Representing the set of end users for whom the task needs to be offloaded.
Task offloading decision sets are represented as
Figure BDA0004079661310000104
wherein
Figure BDA0004079661310000105
A decision binary variable representing the offloading of the task of terminal device e to edge server m, wherein,
Figure BDA0004079661310000106
indicating that terminal task e is offloaded to edge server m via channel j (j e N), otherwise +. >
Figure BDA0004079661310000107
For any one of the terminal tasks to be performed either locally or offloaded to the edge server, the following constraints must be met:
Figure BDA0004079661310000108
the number of end users that any edge server m can serve is constrained by the number of channels available:
Figure BDA0004079661310000109
wherein qm Representing the number of terminal devices that one edge server can serve at the same time at most.
For any edge server m, any channel n can be allocated to at most one end user device, so the following constraints need to be met:
Figure BDA00040796613100001010
the set of transmission power allocation policies for the end user device is expressed as:
Figure BDA00040796613100001011
wherein Pe Representing the transmission power, P, of terminal device e offloading tasks to edge server emax Representing the maximum transmission power corresponding to the terminal equipment, E off The computing tasks are offloaded to the set of terminal devices of the server.
The set of allocation policies for computing resources is expressed as:
F＝{f em |e∈E,m∈M} (6)
wherein fem Representing the computing resources allocated by edge server m to terminal device e.
Because the computing resources of an edge server are also limited, the computing resource allocation of each edge server cannot exceed its own maximum computing power, so there are the following constraints:
Figure BDA0004079661310000111
meaning that for all terminal tasks assigned to any one edge server, the sum of the computing resources assigned to it by the edge server cannot exceed its own maximum computing power, where
Figure BDA0004079661310000112
The maximum computing power of the edge server m, in particular the number of CPU cycles that can run per second, is expressed in cycles per second. E (E) m Representing the set of all end user devices accessing the edge server m.
The time for the end task to complete consists of two parts, namely the time for the end user device to complete locally or the time required for the end user device to execute remotely at the edge server. The following are provided:
Figure BDA0004079661310000113
the task of the terminal e is represented by the local completion time:
Figure BDA0004079661310000114
wherein ,
Figure BDA0004079661310000115
representing the own computing power of the end user device, in particular the number of CPU cycles that can run per second, in cycles per second, c e Is the workload of a computing task, i.e., the number of CPU cycles required.
Figure BDA0004079661310000116
The energy consumption of the terminal e for executing the task locally is represented;
Figure BDA0004079661310000117
where K represents the CPU power consumption coefficient of the end user device, related to the chip hardware. Let k=10 according to literature -27 。
In the invention, the uplink transmission mode of the terminal task uploading is orthogonal frequency division multiple access, all working frequency bands are divided into N sub-bandwidths w with equal size,
Figure BDA0004079661310000118
signal to noise ratio representing the uploading of the terminal task to the m-edge server:
Figure BDA0004079661310000119
the uplink transmission rate of the terminal task is expressed as:
Figure BDA00040796613100001110
w represents the bandwidth of the sub-channel,
Figure BDA00040796613100001111
Representing the transmission power of the end user device n on channel j,/->
Figure BDA00040796613100001112
Is the corresponding channel gain, sigma 2 Represents noise power +.>
Figure BDA00040796613100001113
Indicating interference and other end-point devices in the same channel band, since each end-point device can only transmit on one subchannel and all subchannel bandwidths are assumed to be the same, all end-point uplink offload transmission rates can be expressed as:
Figure BDA0004079661310000121
t e representing the time required for the task of the terminal e to be offloaded to the remote end of the edge server for execution; from the following components
Figure BDA0004079661310000122
and />
Figure BDA0004079661310000123
The composition of the composite material comprises the components,
Figure BDA0004079661310000124
delay in uploading tasks to edge server +.>
Figure BDA0004079661310000125
The required latency is calculated in the edge server for the task. The following are provided:
Figure BDA0004079661310000126
Figure BDA0004079661310000127
Figure BDA0004079661310000128
/>
wherein ,
Figure BDA0004079661310000129
E e indicating the energy consumption required for the task of terminal e to offload to the edge server's remote execution,
Figure BDA00040796613100001210
where ε represents the power amplification efficiency of the end user.
In summary, the task delay and the energy consumption weighting system effect function of the end user equipment are constructed, and under the condition of ensuring that the system effect function is optimal, under a series of constraint conditions, the task unloading decision set X is solved, the transmission power allocation strategy set P of all the end user equipment is solved, and the calculation resource allocation strategy set F of all the edge servers is solved.
For edge calculation, the task overhead of the end user is mainly time delay and power consumption, and weight factors
Figure BDA00040796613100001211
and />
Figure BDA00040796613100001212
Belonging to the range 0 to 1 and->
Figure BDA00040796613100001213
If the task is urgent, the user can increase the delay weight appropriately>
Figure BDA00040796613100001214
Otherwise, the terminal can properly increase the energy consumption weighting factor under the condition of low power>
Figure BDA00040796613100001215
The task delay and the energy consumption weighting system effect function of the end user device are as follows:
Figure BDA00040796613100001216
wherein ,
Figure BDA00040796613100001217
E loc representing a locally computed set of end user devices, E off Representing an offloaded set of end-user devices. In summary, the system effect function is further processed and substituted as follows:
Figure BDA00040796613100001218
therefore, the above-described computational offloading decision and resource allocation joint optimization problem can be expressed as follows:
Figure BDA0004079661310000131
s.t.C1:
Figure BDA0004079661310000132
m∈M,j∈N
Figure BDA0004079661310000133
Figure BDA0004079661310000134
Figure BDA0004079661310000135
Figure BDA0004079661310000136
Figure BDA0004079661310000137
Figure BDA0004079661310000138
/>
first, the offloading decision set X is a discrete binary variable, the transmission power allocation policy set P of all end user devices and the computing resource allocation policy set F of the edge server are continuous variables, which is a mixed integer nonlinear problem, and as a result, looking at all constraints, the offloading constraint source allocation constraints are not related to each other, even if they are decoupled from each other, so that the original problem can be decomposed into a resource allocation optimization function under a given task offloading set and an offloading decision optimization based on the resource optimization result.
Therefore, the resource optimization function in step S3 is a resource optimization function under a specific task offloading decision satisfying the constraint Cl-constraint C4 and an offloading decision optimization function based on the resource optimization result by decoupling, where the task offloading decision optimization function is expressed as follows:
Figure BDA0004079661310000139
s.t.C1:
Figure BDA00040796613100001310
m∈M,j∈N
Figure BDA00040796613100001311
Figure BDA00040796613100001312
Figure BDA00040796613100001313
where V (X) represents an optimal value function of the transmission power allocation problem of the end user device and the computing resource allocation problem of the edge server under a specific offload task decision, as follows:
Figure BDA0004079661310000141
Figure BDA0004079661310000142
Figure BDA0004079661310000143
Figure BDA0004079661310000144
the decoupling conversion process does not affect the optimality of the original problem solving.
Further, the step S4 specifically includes the following steps:
firstly, the step S4 performs secondary decoupling on the resource optimization function to obtain the transmission power allocation function.
We first solve the resource allocation problem (21).
When determining a specific offloading task decision, i.e. assigning an offloading decision to the original problem, the resource allocation optimization function (21) can be reduced to the following expression:
Figure BDA0004079661310000145
Figure BDA0004079661310000146
Figure BDA0004079661310000147
/>
Figure BDA0004079661310000148
as can be seen from the above formula, the power constraint condition and the computational resource constraint condition are decoupled, and the computational resource and the transmission power in the objective function are decoupled separately, so that the resource allocation problem can be further decoupled secondarily into two independent sub-problems of the transmission power allocation of the end user equipment and the computational resource allocation of the edge server.
The transmission power allocation problem for the end user device is expressed as follows:
Figure BDA0004079661310000149
Figure BDA00040796613100001410
the above formula (23) is a non-convex problem, wherein R em Because different end users use the same channel to offload to different edge servers, there is interference between
Figure BDA00040796613100001411
The deployment is as follows:
Figure BDA00040796613100001412
Figure BDA00040796613100001413
according to the literature, in order to eliminate the relevance of different end users, it is possible to use
Figure BDA00040796613100001414
To replace itself by the approximate upper limit of +.>
Figure BDA0004079661310000151
The objective function and the transmit power constraints of each end user are decoupled, so that the power allocation problem (24) can be reduced to a set of sub-problems each independent of the other for optimizing the transmit power of each end user, as follows
Figure BDA0004079661310000152
Figure BDA0004079661310000153
The method can prove that the equation is a strict pseudo-convex function, a dichotomy is usually adopted for solving, and can prove that the objective function is monotonically increased, so that the optimal solution is positioned at the constraint boundary or the zero crossing position of the first derivative, the dichotomy firstly sets the initial interval of the initial power and the tolerance thereof, judges whether the boundary value is the optimal solution of the first derivative of the pseudo-convex function, if not, the method carries out dichotomy, calculates the value of the first derivative of the pseudo-convex function in each dichotomy iteration, and judges whether the solution is the optimal solution. To this end, a solution to the problem of transmission power allocation is obtained
Figure BDA0004079661310000154
Next, solving the computational resource allocation problem, as known from equation (22), the power constraint and the computational resource constraint are decoupled, and the computational resource and the transmission power in the objective function are also decoupled separately, so the computational resource allocation problem can be expressed separately as follows:
Figure BDA0004079661310000155
Figure BDA0004079661310000156
/>
Figure BDA0004079661310000157
from the above equation, the constraint condition is a convex function, then the convexity of the objective function is seen, the second derivative is obtained, and the black plug matrix of the objective function can be obtained to be a semi-positive definite matrix, so that the objective function is also a convex function, and in summary, the equation (26) is a convex optimization problem, and can be generally solved by adopting the KKT condition.
The optimal allocation mode of the computing resources obtained by solving the KKT condition is as follows:
Figure BDA0004079661310000158
finally, obtaining the optimal power distribution scheme
Figure BDA0004079661310000159
And computing resource optimal allocation scheme->
Figure BDA00040796613100001510
From equations (21) - (27), the task offloading decision-making optimization function can be further expressed as:
Figure BDA0004079661310000161
s.t.C1:
Figure BDA0004079661310000162
m∈M,j∈N
Figure BDA0004079661310000163
Figure BDA0004079661310000164
Figure BDA0004079661310000165
and finally, solving an unloading decision set under a certain constraint under the condition that the solved unloading strategy is power allocation and a computing resource allocation scheme is known, so that the system effect is maximum, namely the total time delay and the power consumption of the system are minimum.
From the foregoing, it can be seen that the task overhead of the end user is mainly time delay and power consumption, and the weight factor
Figure BDA0004079661310000166
and />
Figure BDA0004079661310000167
Belonging to the range 0 to 1 and->
Figure BDA0004079661310000168
If the task is urgent, the user can increase the delay weight appropriately>
Figure BDA0004079661310000169
Otherwise, the terminal can properly increase the energy consumption weighting factor under the condition of low power>
Figure BDA00040796613100001610
The end user local computational overhead is expressed as:
Figure BDA00040796613100001611
the end user offload to edge server task overhead is:
Figure BDA00040796613100001612
based on the offloading decision, the computational overhead of the end user can be expressed as:
Figure BDA00040796613100001613
wherein xe ＝∑ m∈M x em 。
To solve the set of computational offload decisions X, here solved with the beneficial user-hybrid improved genetic algorithm proposed by the present invention, the problem of computational offload decisions includes whether the end user is to offload and to which edge server.
Firstly, the problem of whether an end user needs to be unloaded is solved, the difference value of all users through user spending is divided into an unloading user set and a local computing user set, and the difference value of the unloading computing and the local computing spending of the end user is defined as follows:
Figure BDA00040796613100001614
Figure BDA00040796613100001615
if U e <0 means that the end user e has less offloading overhead than the local computing overhead, and this class of users is defined as beneficiary users who are to be offloaded; if U e 0 indicates that offloading is larger than the overhead of local computation, and offloading does not reduce overhead but increases overhead, such end users are defined as non-beneficiary users, such users do not have to offload, so such user offloading decisions are performed locally.
After the unloading user and the local user set are obtained through the first step, then the edge server matching problem of all the unloading users is solved, namely the unloading user is unloaded to which edge server, and the unloading decision of the part is solved through a genetic algorithm. The method comprises the following steps:
1) First chromosome coding
Setting a population size as s, setting E end users needing task unloading, M edge servers, each chromosome representing an unloading scheme, adopting a real number coding mode, each chromosome having N genes, the value of each gene representing the mth edge server, the subscript sequence number of the gene representing the sequence number of the end user, for example, chromosome C= { C 1 ，c 2 ，c 3 ，。。。，c e}, wherein c2 When =a, the task representing the second end user is offloaded to the edge server numbered a. The value range of each gene is [0, m ]]Since there are m edge servers, a value of 0 indicates execution locally.
2) Fitness function
The fitness represents the adaptability of an individual to the environment, and the genetic algorithm judges the index of the individual in the population by the value of the fitness function, which is generally an objective function, but the objective function is the cost for minimizing the task completion of the end user, so we want to represent the fitness function by the inverse of the total cost, the fitness l i The following is shown:
Figure BDA0004079661310000171
the higher the fitness function value is, the smaller the task completion cost is, the smaller the cost for representing the calculation unloading scheme is, namely, the better the scheme is, and the higher the fitness function value is, and can be used as the basis of genetic operation.
3) Selection of
The selection operation in the genetic algorithm is to determine how to select individuals from a parent population in a way that will inherit to the next generation population. The selection operation is used to determine the reorganization or crossover individuals, and how many child generations the selected individuals will produce, the method of selecting operators is numerous, here using roulette selection, with the probability of being selected as:
Figure BDA0004079661310000172
4) Crossover variation
The crossing function is to keep the population stable and genetically evolve towards the optimal direction, and the crossing operator randomly exchanges some genes for two individuals in the population according to the crossing rate to generate a new gene combination and generate a new individual; the mutation has the function of avoiding the population from being in local optimum, judging whether to mutate the set mutation probability of all individuals in the population, and then randomly selecting mutation of mutation positions of mutated individuals, wherein the mutation can improve the population diversity, and a proper cross mutation method is selected according to the coding mode.
The crossover adopts a monomer crossover mode, two chromosomes obtained in the above selection are randomly selected at one crossover point position, so that genes behind the crossover point positions of the two chromosomes are interchanged to obtain two new chromosome individuals, and new chromosome offspring are generated. The crossover probability is set to 0.6.
Variation: and randomly selecting a mutation position of the chromosome according to the mutation probability, and changing the value of the gene at the mutation position to obtain a new chromosome. The probability of variation was set to 0.02.
5) Termination condition
If the set maximum iteration number is reached or the fitness value is kept unchanged for a long time, the genetic algorithm flow is terminated. And outputting to obtain the optimal unloading strategy X and the optimal fitness value.
Finally, it is noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made thereto without departing from the spirit and scope of the present invention, which is intended to be covered by the claims of the present invention.

Claims (10)

1. A resource allocation and task offloading method based on the edge calculation of the Internet of things is characterized in that: the method comprises the following steps:
S1: an edge server-based internet of things edge computing system is constructed, and the system comprises a task unloading distribution processing model, a computing resource and power consumption distribution model and a time delay model;
s2: based on the optimization target of resource allocation, unloading, power and time delay are combined, and the task completion time delay of all terminal user equipment and the energy consumption weight of the equipment are combined to construct an effect function of the system;
s3: decomposing a joint optimization problem of resource allocation and task unloading, and decomposing the system effect function into a resource allocation optimization function under an initial given task unloading decision and a task unloading optimization function based on a resource allocation optimization result;
s4: the resource allocation optimization function under the initial given task unloading decision is secondarily decomposed into a power allocation optimization function of the terminal user equipment and a computing resource allocation optimization function of the edge server;
s5: according to the characteristics of the power distribution optimization function of the terminal user equipment, adopting a dichotomy to solve to obtain an optimal transmission power distribution scheme of the terminal user equipment;
s6: solving by using KKT conditions according to the characteristics of the computing resource allocation optimization function of the edge server to obtain an optimal computing resource allocation scheme;
S7: and (3) after the optimal transmission power distribution scheme and the optimal calculation distribution scheme obtained in the steps S5 and S6 are carried back to the original problem system effect function, and the optimal task unloading strategy is obtained by solving the beneficial user-hybrid improved genetic algorithm.
2. The method for resource allocation and task offloading based on internet of things edge computing according to claim 1, wherein: the task unloading and distributing processing model in the step S1 is as follows:
the method comprises the steps of setting an Internet of things terminal equipment set to be E= {1,2, …, E }, assuming that each Internet of things terminal can generate a calculation task within a time period, and the calculation task is independently completed, cannot be divided and is marked as T e ＝{d e ，c e }，d e Representing the size of the task input data, c e Representing the workload of a task, namely the calculated amount, in particular the CPU period required by calculating the task;
the edge server set is expressed as M= {1,2, …, M }, the uplink transmission adopts orthogonal frequency division multiple access as an access mode of uploading a terminal task to the edge server, all working frequency bands are divided into N mutually orthogonal channels, and N= {1,2, …, N } is the total number of the mutually orthogonal channels;
the task offloading decision set is expressed as:
Figure FDA0004079661300000011
wherein
Figure FDA0004079661300000012
Decision binary variable representing the offloading of the task of terminal e to edge server m, wherein +. >
Figure FDA0004079661300000013
Indicating that terminal task e is offloaded to edge server m via channel j, otherwise +.>
Figure FDA0004079661300000014
The set of end users for which tasks need to be offloaded is represented as:
E off ＝e m∈M E m
for any one terminal task to either execute locally or offload to an edge server, the following constraints are met:
Figure FDA0004079661300000021
the number of end users that any edge server m can serve is constrained by the number of channels available:
Figure FDA0004079661300000022
/>
wherein qm Representing the number of terminal devices that can be served by one edge server at the same time at most;
for any edge server m, any channel n can be allocated to at most one end user device, so the following constraints are met:
Figure FDA0004079661300000023
3. the method for resource allocation and task offloading based on internet of things edge computing according to claim 1, wherein: the computing resource and power consumption allocation model is as follows:
the set of transmission power allocation policies for the end user device is expressed as:
Figure FDA0004079661300000024
wherein Pe Representing the transmission power, P, of terminal device e offloading tasks to edge server emax Representing the maximum transmission power corresponding to the terminal equipment, E off A terminal device set offloaded to a server for a computing task;
the set of allocation policies for computing resources is expressed as:
F＝{f em |e∈E,m∈M}
wherein fem Representing the computing resources allocated by the edge server m to the terminal device e;
the computing resource allocation of each edge server cannot exceed its own maximum computing power, with the following constraints:
Figure FDA0004079661300000025
meaning that for all terminal tasks assigned to any one edge server, the sum of the computing resources assigned to it by the edge server cannot exceed its own maximum computing power, where
Figure FDA0004079661300000026
Representing the maximum computing power of the edge server m, in particular the number of CPU cycles per second, E m Representing the set of all end user devices accessing the edge server m.
4. The method for resource allocation and task offloading based on internet of things edge computing according to claim 1, wherein: the time delay model is as follows:
the time for the terminal task to be completed consists of two parts, namely the time for the terminal user equipment to be completed locally or the time required by the terminal user equipment to be executed at the far end of the edge server;
Figure FDA00040796613000000325
the task of the terminal e is represented by the local completion time:
Figure FDA0004079661300000031
wherein ,
Figure FDA00040796613000000314
representing the own computing power of the end user device, in particular the number of CPU cycles per second that can be run, c e A workload, i.e., the number of CPU cycles required, for a computing task; t is t e The time required for the task of terminal e to be offloaded to the edge server remote execution is represented by +.>
Figure FDA00040796613000000315
and />
Figure FDA00040796613000000316
Composition (S)/(S)>
Figure FDA00040796613000000317
Delay in uploading tasks to edge server +.>
Figure FDA00040796613000000318
The required latency is calculated in the edge server for the task:
Figure FDA0004079661300000032
Figure FDA0004079661300000033
Figure FDA0004079661300000034
/>
wherein xem Representing the task offload variables,
Figure FDA0004079661300000035
uplink transmission rate indicating terminal task, +.>
Figure FDA00040796613000000319
The following are provided:
Figure FDA0004079661300000036
5. the method for resource allocation and task offloading based on internet of things edge computing according to claim 1, wherein: the effect function of the system in step S2 is:
Figure FDA0004079661300000037
wherein ,
Figure FDA0004079661300000038
E loc representing a locally computed set of end user devices, E off Representing an offloaded set of end user devices, weighting factor +.>
Figure FDA00040796613000000320
and />
Figure FDA00040796613000000321
Belonging to the range 0 to 1 and->
Figure FDA00040796613000000322
If the task is urgent, the user can increase the delay weight +.>
Figure FDA00040796613000000323
Otherwise, the terminal can increase the energy consumption weighting factor under the condition of low electric quantity>
Figure FDA00040796613000000324
The computational offloading decision and resource allocation joint system effect optimization problem is expressed as:
Figure FDA0004079661300000039
Figure FDA00040796613000000310
Figure FDA00040796613000000311
Figure FDA00040796613000000312
Figure FDA00040796613000000313
Figure FDA0004079661300000041
Figure FDA0004079661300000042
Figure FDA0004079661300000043
wherein Q represents a system benefit function, X represents a task offloading decision set, P represents a transmission power allocation decision set, F represents a computing resource allocation decision set, and constraint Cl represents an offloading decision as a binary variable; constraint C2 indicates that the task of the terminal of the Internet of things is unloaded to an edge server to be executed or is executed locally; constraint C3 indicates that each edge server's sub-bandwidth channels can be allocated to at most one terminal device; constraint C4 represents the number of terminal devices that can offload tasks to the edge server at most; constraint C5 is the power constraint of the vehicle, P e Representing the transmission power, P, of terminal device e offloading tasks to edge server emax Representing the maximum transmission power corresponding to the terminal equipment; constraint C6 indicates that the computing resources allocated by the edge server to the terminal device must be positive, f em Representing the computing resources allocated by the edge server m to the terminal device e; constraint C7 represents offloading to edgeThe sum of the computing resources required by the tasks of the servers does not exceed the computing power of the edge servers, f m Representing the computing resources owned by edge server m.
6. The method for resource allocation and task offloading based on internet of things edge computing according to claim 1, wherein: the step S3 specifically includes:
the system effect function is decomposed into a resource allocation optimization function under the initial task unloading decision and a task unloading optimization function based on a resource allocation optimization result, which are respectively expressed as:
Figure FDA0004079661300000044
Figure FDA0004079661300000045
Figure FDA0004079661300000046
Figure FDA0004079661300000047
Figure FDA0004079661300000048
where V (X) represents an optimal value function of the transmission power allocation problem of the end user device and the computing resource allocation problem of the edge server under a specific offload task decision, as follows:
Figure FDA0004079661300000049
Figure FDA00040796613000000410
Figure FDA00040796613000000411
Figure FDA00040796613000000412
7. the method for resource allocation and task offloading based on internet of things edge computing according to claim 1, wherein: the power allocation optimization function of the end user device in step S4 is expressed as:
Figure FDA0004079661300000051
Figure FDA0004079661300000052
The computing resource allocation optimization function of the edge server is expressed as:
Figure FDA0004079661300000053
Figure FDA0004079661300000054
Figure FDA0004079661300000055
8. the method for resource allocation and task offloading based on internet of things edge computing according to claim 1, wherein: in step S5, according to the characteristics of the power allocation optimization function of the end user device, a dichotomy is adopted to solve to obtain an optimal transmission power allocation scheme of the end user device, which specifically includes:
the power distribution optimizing function is characterized in that the power distribution optimizing function is a strict quasi-convex function, an optimal solution is positioned at a constraint boundary or zero crossing position of a first derivative, an initial interval of initial power and tolerance thereof are set by a dichotomy, whether a boundary value is the optimal solution of the first derivative of the quasi-convex function is judged, if not, the power distribution optimizing function is divided into two parts, the value of the first derivative of the quasi-convex function is calculated in each divided iteration, and whether the value is the optimal solution is judged until a solution of the transmission power distribution problem is obtained
Figure FDA00040796613000000512
9. The method for resource allocation and task offloading based on internet of things edge computing according to claim 1, wherein: and step S6, solving by using KKT conditions according to the characteristics of the computing resource allocation optimization function of the edge server to obtain an optimal computing resource allocation scheme, wherein the method specifically comprises the following steps:
Figure FDA0004079661300000056
wherein ,
Figure FDA00040796613000000511
and representing the calculation resource allocation result under the target optimization function.
10. The method for resource allocation and task offloading based on internet of things edge computing according to claim 1, wherein: in step S7, an optimal transmission power allocation scheme is obtained
Figure FDA0004079661300000059
And optimal computational allocation scheme->
Figure FDA00040796613000000510
And then, carrying back the original problem system effect function to obtain a task unloading optimization function, wherein the task unloading optimization function comprises the following steps of:
Figure FDA0004079661300000057
Figure FDA0004079661300000058
Figure FDA0004079661300000061
Figure FDA0004079661300000062
Figure FDA0004079661300000063
to solve the set of computational offload decisions X, a beneficiary-hybrid improved genetic algorithm is used to solve the problem of computational offload decisions including whether the end user is to offload and to which edge server, as follows:
firstly, the problem of whether an end user needs to be unloaded is solved, the difference value of all users through user spending is divided into an unloading user set and a local computing user set, and the difference value of the unloading computing and the local computing spending of the end user is defined as follows:
Figure FDA0004079661300000064
if U e <0 means that the end user e has less offloading overhead than the local computing overhead, and this class of users is defined as beneficiary users who are to be offloaded; if U e More than or equal to 0 means that the offloading is more expensive than the local computation, such end users are defined as non-beneficiary users, such users do not offload, so such user offloading decisions are performed locally;
After solving the unloading user and the local user set, solving the edge server matching problem of all the unloading users, namely the unloading user is unloaded to which edge server, and solving the unloading decision of the part through a genetic algorithm.
CN202310119925.7A 2023-02-15 2023-02-15 Resource allocation and task unloading method based on edge calculation of Internet of things Pending CN116089091A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310119925.7A CN116089091A (en) 2023-02-15 2023-02-15 Resource allocation and task unloading method based on edge calculation of Internet of things

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310119925.7A CN116089091A (en) 2023-02-15 2023-02-15 Resource allocation and task unloading method based on edge calculation of Internet of things

Publications (1)

Publication Number Publication Date
CN116089091A true CN116089091A (en) 2023-05-09

Family

ID=86186815

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310119925.7A Pending CN116089091A (en) 2023-02-15 2023-02-15 Resource allocation and task unloading method based on edge calculation of Internet of things

Country Status (1)

Country Link
CN (1) CN116089091A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116993183A (en) * 2023-09-27 2023-11-03 电子科技大学中山学院 Service optimization method for probabilistic computation offloading in unmanned aerial vehicle auxiliary edge computation

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116993183A (en) * 2023-09-27 2023-11-03 电子科技大学中山学院 Service optimization method for probabilistic computation offloading in unmanned aerial vehicle auxiliary edge computation
CN116993183B (en) * 2023-09-27 2023-12-29 电子科技大学中山学院 Service optimization method for probabilistic computation offloading in unmanned aerial vehicle auxiliary edge computation

Similar Documents

Publication Publication Date Title
Bozorgchenani et al. Multi-objective computation sharing in energy and delay constrained mobile edge computing environments
CN111586720B (en) Task unloading and resource allocation combined optimization method in multi-cell scene
CN109684075B (en) Method for unloading computing tasks based on edge computing and cloud computing cooperation
CN111585816B (en) Task unloading decision method based on adaptive genetic algorithm
CN111182570B (en) User association and edge computing unloading method for improving utility of operator
CN109600178B (en) Optimization method for energy consumption, time delay and minimization in edge calculation
CN110233755B (en) Computing resource and frequency spectrum resource allocation method for fog computing in Internet of things
CN112286677A (en) Resource-constrained edge cloud-oriented Internet of things application optimization deployment method
CN113781002B (en) Low-cost workflow application migration method based on agent model and multiple group optimization in cloud edge cooperative network
CN111787543B (en) 5G communication system resource allocation method based on improved wolf optimization algorithm
CN112637907B (en) Combined optimization method for user multi-association and downlink power distribution in millimeter wave network
CN113918240A (en) Task unloading method and device
CN113364859A (en) MEC-oriented joint computing resource allocation and unloading decision optimization scheme in Internet of vehicles
Zhu et al. Computing offloading strategy using improved genetic algorithm in mobile edge computing system
CN113573363A (en) MEC calculation unloading and resource allocation method based on deep reinforcement learning
CN116089091A (en) Resource allocation and task unloading method based on edge calculation of Internet of things
Lan et al. Deep reinforcement learning for computation offloading and caching in fog-based vehicular networks
CN113590279A (en) Task scheduling and resource allocation method for multi-core edge computing server
Zhou et al. Multi-server federated edge learning for low power consumption wireless resource allocation based on user QoE
Zhang et al. A deep reinforcement learning approach for online computation offloading in mobile edge computing
Dou et al. Mobile edge computing based task offloading and resource allocation in smart grid
Wang Collaborative task offloading strategy of UAV cluster using improved genetic algorithm in mobile edge computing
CN108965034A (en) Small-cell base station super-intensive deployment under user-association to network method
CN116828534B (en) Intensive network large-scale terminal access and resource allocation method based on reinforcement learning
CN114615705B (en) Single-user resource allocation strategy method based on 5G network

Legal Events

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