Edge computing task scheduling method and system based on genetic simulated annealing algorithm and storage medium

Abstract

The invention relates to the technical field of edge computing, in particular to an edge computing task scheduling method and system based on a genetic simulated annealing algorithm and a storage medium, and the method comprises the following steps: setting an edge computing task scheduling initial moment w, collecting all to-be-scheduled edge computing tasks at the moment w for unified scheduling, and performing unified scheduling processing; determining all participated tasks in the scheduling to form a set N, and obtaining detailed information of each task in the set N; the edge computing task scheduling method and system based on the genetic simulated annealing algorithm and the storage medium provided by the invention have the beneficial effects that the algorithm of fusing the genetic algorithm and the simulated annealing algorithm is adopted, the global search characteristic of the genetic algorithm is combined, and meanwhile, the Metropolis criterion of the simulated annealing algorithm is used for preventing the algorithm from falling into local optimum, so that the task scheduling efficiency is improved. And a better decision result can be obtained.

Description

Technical Field

[0001] The present invention relates to the field of edge computing technology, and specifically to an edge computing task scheduling method, system and storage medium based on a genetic simulated annealing algorithm. Background Art

[0002] With the rapid development of artificial intelligence and big data technologies, the demand for computing resources in industrial equipment is growing. A large number of artificial intelligence technologies such as image recognition and data analysis are widely used to improve industrial production efficiency. However, the decentralized underlying devices have high requirements for computing resources, such as low latency, high bandwidth, and data privacy. Traditional cloud computing architecture cannot meet the real-time requirements of industrial sites due to long transmission links and homogeneous server resources.

[0003] The edge computing network architecture is a new type of network architecture proposed on the basis of cloud computing. The core of this architecture is to deploy servers with certain computing capabilities on the edge side of the network, and at the same time utilize the scattered terminal computing resources on the edge side to achieve deep integration of cloud, network, edge, and end. This architecture is particularly suitable for scenarios that require low latency and high reliability, such as industrial automation, intelligent transportation systems, and the Internet of Things. In terms of task scheduling and resource allocation, important research issues facing edge computing include how to efficiently allocate computing resources on the edge side to meet the real-time computing needs of different industrial equipment. Due to the diversity of industrial field equipment, including different manufacturers, interface protocols, and computing resource requirements, task scheduling algorithms must take into account environmental heterogeneity and optimize resource allocation to improve efficiency.

[0004] At present, the research field of edge computing is mainly concentrated in the fields of smart terminals and Internet of Vehicles. However, there are relatively few studies on edge computing network architecture in the field of industrial Internet, which limits the application of key fields such as edge computing. In terms of research methods, although deep reinforcement learning has been widely studied in task scheduling of edge computing, these methods often rely on a large number of data set samples and may not fully consider the heterogeneity of task devices and server resources. Therefore, for edge computing task scheduling in the field of industrial Internet, more flexible and adaptable algorithms are needed to cope with the diversity of devices and resources. Summary of the invention

[0005] The purpose of the present invention is to provide an edge computing task scheduling method, system and storage medium based on a genetic simulated annealing algorithm to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: an edge computing task scheduling method based on a genetic simulated annealing algorithm, the method comprising the following steps:

[0007] Step 1: Set the initial time w for scheduling edge computing tasks, collect all edge computing tasks to be scheduled at time w, and perform unified scheduling;

[0008] Step 2: Determine all the tasks involved in this scheduling, form a set N, and obtain detailed information for each task in the set N;

[0009] Step 3: Confirm all available edge servers participating in this task scheduling to form a set M, and obtain relevant information of all edge servers in the set M;

[0010] Step 4: According to the information of tasks and servers, decision variables are set to build a nonlinear optimization model, which takes into account the completion time of tasks, the load balance of servers, and the resource matching factors between tasks and servers;

[0011] Step 5: Use the genetic simulated annealing algorithm to solve the established nonlinear optimization model, solve the nonlinear optimization model according to the task set N, the server set M and basic information, and determine the task scheduling decision result.

[0012] Preferably, the basic information of all tasks in the set N includes: the data volume of task i∈N is D i , the amount of calculation is B i , the expected completion time is d i , the required memory size is M fi , the required hard disk capacity is S fi , the required computing module resources are O i .

[0013] Preferably, the basic information of all edge servers in the set M includes: the processing speed of server j∈M is C j , the data transmission speed is W j , memory capacity is M Fj , hard disk capacity is S Fj , the computing module resources are L j .

[0014] Preferably, the server receives multiple tasks but can only process one task at a time, the tasks are executed sequentially in the order of transmission, each industrial field device generates only one task, and each task can only be processed by one server node.

[0015] Preferably, the resource matching model between the task and the server node is as follows: the resource matching between the task and the server node is calculated based on the memory resources, hard disk resources, computing module resources of the server node and the memory resources, hard disk resources, computing module resources required by the task; the resource matching calculation formula for offloading the task i∈N to the edge node j∈M is:

[0016] Preferably, the task response model between the task and the server node is as follows: the response delay of the task is calculated according to the data transmission speed, processing speed and data volume and calculation amount of the server node; the response delay of task i∈N offloaded to edge node j∈M can be calculated as:

[0017] Preferably, the specific meaning of the decision variable is:

[0018]

[0019] The objective function of the nonlinear optimization model is:

[0020]

[0021] Among them, the objective function represents minimizing the maximum value of task response delay and resource matching degree in all server nodes;

[0022] Among them, the constraints of the nonlinear optimization model are as follows: Constraint 1 means that any task can only be assigned to one server, Constraint 2 means that each task can only be assigned to a server node that meets the memory requirements, Constraint 3 means that each task can only be assigned to a server node that meets the hard disk requirements, Constraint 4 means that each task can only be assigned to a server node that meets the computing module resource requirements, and Constraint 5 means the value range of the decision variable.

[0023] An edge computing task scheduling system based on a genetic simulated annealing algorithm is applied to an edge computing task scheduling method based on a genetic simulated annealing algorithm, and the system includes:

[0024] The task collection module is used to determine the task set generated by the industrial field equipment and obtain the basic information of the task;

[0025] The server cluster module is used to determine the node set of the server cluster and obtain the basic information of the server nodes;

[0026] The model building module is used to set decision variables and build nonlinear programming models.

[0027] The decision module is used to solve the nonlinear optimization model and determine the task scheduling decision by calling the edge computing task scheduling method based on the genetic simulated annealing algorithm according to the task set N and the server set M as well as the basic information and decision variables.

[0028] A storage medium stores a computer program, which, when executed by a processor, implements an edge computing task scheduling method based on a genetic simulated annealing algorithm.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] The edge computing task scheduling method, system and storage medium based on the genetic simulated annealing algorithm proposed in the present invention adopt an algorithm that integrates the genetic algorithm and simulated annealing, combines the global search characteristics of the genetic algorithm, and uses the Metropolis criterion of the simulated annealing algorithm to avoid the algorithm from falling into the local optimum, so as to obtain better decision-making results; comprehensively consider the transmission delay and execution delay of the task, the resource matching relationship between the task and the server node, and the cumulative comprehensive matching degree of the server node, accurately describe the matching relationship between the task and the server, and can be effectively applied to the edge computing network architecture deployment in the field of industrial Internet; consider the on-site operation equipment manufacturers in the field of industrial Internet, the interface protocol heterogeneity, and the distributed architecture relationship of the server resource nodes, and give the specific definition of the industrial field equipment and the server resource nodes. The proposed algorithm can be applied to the edge computing task scheduling problem in the field of industrial Internet. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a flow chart of the edge computing task scheduling method of the present invention;

[0032] Figure 2 It is a specific flow chart of the genetic simulated annealing algorithm of the present invention. DETAILED DESCRIPTION

[0033] In order to make the purpose and technical solution of the present invention clearly and completely described, and the advantages more clearly understood, the embodiments of the present invention are further described in detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are part of the embodiments of the present invention, rather than all of the embodiments, and are only used to explain the embodiments of the present invention, and are not used to limit the embodiments of the present invention. All other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] For example, see Figure 1 to Figure 2 The present invention provides a technical solution: an edge computing task scheduling method based on a genetic simulated annealing algorithm, the method comprising the following steps:

[0035] Step 1: Set the initial time of edge computing task scheduling to 8:00:00, and uniformly schedule the edge computing tasks collected at this time.

[0036] Step 2: Determine all the tasks that need to be scheduled uniformly this time, form a set N, and obtain the basic information of all tasks in the set N; the basic information of all tasks in the set N includes: the data volume of task i∈N is D i , the amount of calculation is B i , the expected completion time is d i , the required memory size is M fi , the required hard disk capacity is S fi , the required computing module resources are O i .

[0037] In this embodiment, the industrial production line collects 100 computing tasks to be assigned within 8:00:00, and the task numbers are 1-100 respectively. The basic information of the task includes the data volume, in kB; the computing volume, in MI; the memory capacity, in MB; the hard disk capacity, in MB; the number of computing resource modules required by the task is 1, in FLOPS; and the expected completion time of the task.

[0038] Furthermore, each industrial field device generates only one task, and each task can only be processed by one server node.

[0039] Table 1 Task parameter settings

[0040]

[0041]

[0042] Step 3: Determine all available edge servers in this unified scheduling to form a set M, and obtain basic information of all edge servers in the set M; the basic information of all edge servers in the set M includes: the processing speed of server j∈M is C j , the data transmission speed is W j , memory capacity is M Fj , hard disk capacity is S Fj , the computing module resources are L j .

[0043] In this embodiment, the number of edge-side server nodes is 5, the server numbers are 1-5 respectively, and the server information includes: the processing speed of the server, in MIPS; the transmission speed of the server, in kB / s; the memory capacity of the server, in MB; the hard disk capacity of the server, in MB; the number of computing resource modules possessed by the server is 1, in FLOPS.

[0044] Table 2 Server parameter settings

[0045]

[0046] Furthermore, the server can receive multiple tasks but can only process one task at a time, and the tasks are executed in sequence according to the transmission order.

[0047] Step 4: Set the decision variables and determine the decision variables, objective function and constraints of the mathematical model.

[0048] Step 5: Through the edge computing task scheduling method based on the genetic simulated annealing algorithm, the nonlinear optimization model is solved according to the task set N, the server set M and basic information to determine the task scheduling decision.

[0049] In a specific example, the processing flow of the genetic simulated annealing algorithm is as follows Figure 2 As shown, the specific process is as follows.

[0050] S1: Initialize genetic simulated annealing algorithm parameters.

[0051] The specific parameters include the population size set to 100, the crossover probability to 0.8, the mutation probability to 0.05, the initial temperature to 100, the termination temperature to 0.1, and the maximum number of algorithm iterations to 200.

[0052] S2: Population initialization.

[0053] Encoding: The algorithm uses natural number encoding. The length of the solution is the number of tasks. Each position in the solution represents a task, and the element at each position represents the server number to which the corresponding task is assigned. For example, a feasible solution to the problem of allocating five tasks to four servers can be encoded as X = (1, 2, 1, 3, 4), which means: tasks 1 and 3 are assigned to server 1, task 2 is assigned to server 2, task 4 is assigned to server 3, and task 5 is assigned to server 4.

[0054] The fitness of each chromosome is equal to the inverse of the maximum value of the comprehensive matching degree among all server nodes. The comprehensive matching degree is the sum of the task execution time and resource matching degree of all tasks on server node j. The execution time of all tasks on any server is the sum of the transmission time and running time of all tasks assigned to the server.

[0055] Population initialization:

[0056] In the greedy solution, it is assumed that the tasks assigned to the same server are transmitted in order of priority from high to low. Given that the set of all tasks is N, the greedy solution generation scheme is as follows:

[0057] Step 1: Calculate the priority of the task according to the expected completion time of the task, the task scheduling time, and the computational amount of the task, and sort all tasks in descending order according to the priority of the task.

[0058] Step 2: Assign the highest priority task in set N to the available server with the smallest sum of task execution time and resource matching degree, and update set N.

[0059] Step 3: If the set N is empty, output the allocation result and terminate the greedy algorithm; otherwise, go to step 1.

[0060] For the remaining chromosomes, this chapter randomly selects any available server from the allocable server nodes to generate the initial solution. The specific algorithm steps are as follows:

[0061] Step 1: Based on the constraints of memory resources, hard disk resources, and computing module resources, determine the set of server nodes that meet the resource constraints for all tasks.

[0062] Step 2: Randomly select an available server node for the task with the smallest task number in set N, and update set N.

[0063] Step 3: If the set N is empty, output the allocation result and terminate the algorithm; otherwise, go to step 2.

[0064] S3: Fitness value calculation.

[0065] The fitness value of the individual is calculated according to the objective function of the nonlinear optimization model.

[0066] S4: Selection operation. A roulette wheel selection method with an elite retention strategy is used.

[0067] S5: Crossover operation. Use multi-point crossover mode.

[0068] S6: Mutation operation. The server node with the largest sum of task execution time and resource matching degree among all current server nodes is selected; a task is randomly selected on the server node, and the set of servers that can execute the task is determined; a server is selected from the set of all servers that can execute the task, and the task is added to the task queue of the server node.

[0069] S7: Perform Metropolis principle operations.

[0070] Acceptance of the individual will be determined based on Metropolis guidelines.

[0071] If the fitness value of an individual is better than that of its parent individual before crossover mutation, the individual is accepted.

[0072] If the fitness value of an individual is inferior to that of its parent individual before crossover mutation, the current simulated annealing temperature is calculated according to the number of iterations, and then the probability of accepting the individual is obtained based on the probability calculation formula and compared with the probability randomly generated by the algorithm. If the calculated probability is greater than the randomly generated probability, the individual is accepted, otherwise it is not accepted.

[0073]

[0074] S8: The number of iterations is accumulated to determine whether the maximum number of iterations has been reached. If not, go to S3.

[0075] S9: Determine the individual fitness values ​​in the current population and output the individual with the largest fitness value.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments can still be modified, or some or all of the technical features therein can be replaced by equivalents. Therefore, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the claims of the present invention.

[0077] Embodiment 2, based on embodiment 1, proposes an edge computing task scheduling system based on genetic simulated annealing algorithm, which is applied to the edge computing task scheduling method based on genetic simulated annealing algorithm. The system includes:

[0078] The task collection module is used to determine the task set generated by the industrial field equipment and obtain the basic information of the task;

[0079] The server cluster module is used to determine the node set of the server cluster and obtain the basic information of the server nodes;

[0080] The model building module is used to set decision variables and build nonlinear programming models.

[0081] The decision module is used to solve the nonlinear optimization model and determine the task scheduling decision by calling the edge computing task scheduling method based on the genetic simulated annealing algorithm according to the task set N and the server set M as well as the basic information and decision variables.

[0082] Embodiment 3, based on embodiment 2, proposes a storage medium on which a computer program is stored. When the computer program is executed by a processor, an edge computing task scheduling method based on a genetic simulated annealing algorithm is implemented.

[0083] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for scheduling edge computing tasks based on genetic simulated annealing algorithm, characterized in that: The method comprises the following steps: Step 1: Set the initial time w for scheduling edge computing tasks, collect all edge computing tasks to be scheduled at time w, and perform unified scheduling; Step 2: Determine all the tasks involved in this scheduling, form a set N, and obtain detailed information for each task in the set N; Step 3: Confirm all available edge servers participating in this task scheduling to form a set M, and obtain relevant information of all edge servers in the set M; Step 4: According to the information of tasks and servers, decision variables are set to build a nonlinear optimization model, which takes into account the completion time of tasks, the load balance of servers, and the resource matching factors between tasks and servers; Step 5: Use the genetic simulated annealing algorithm to solve the established nonlinear optimization model, solve the nonlinear optimization model according to the task set N, the server set M and basic information, and determine the task scheduling decision result.

2. The edge computing task scheduling method based on genetic simulated annealing algorithm according to claim 1 is characterized in that: The basic information of all tasks in the set N includes: The data volume of task i∈N is D i , the amount of calculation is B i , the expected completion time is d i , the required memory size is M fi , the required hard disk capacity is S fi , the required computing module resources are O i .

3. The edge computing task scheduling method based on genetic simulated annealing algorithm according to claim 1 is characterized in that: The basic information of all edge servers in the set M includes: the processing speed of server j∈M is C j , the data transmission speed is W j , memory capacity is M Fj , hard disk capacity is S Fj , the computing module resources are L j .

4. The edge computing task scheduling method based on genetic simulated annealing algorithm according to claim 1 is characterized in that: The server receives multiple tasks but can only process one task at a time. Tasks are executed in the order of transmission. Each industrial field device generates only one task, and each task can only be processed by one server node.

5. The edge computing task scheduling method based on genetic simulated annealing algorithm according to claim 4 is characterized in that: The resource matching model between tasks and server nodes is as follows: the resource matching between tasks and server nodes is calculated based on the memory resources, hard disk resources, computing module resources of the server node and the memory resources, hard disk resources, computing module resources required by the task; The resource matching calculation formula for offloading task i∈N to edge node j∈M is:

6. The edge computing task scheduling method based on genetic simulated annealing algorithm according to claim 5 is characterized in that: The task response model between tasks and server nodes is as follows: the response delay of the task is calculated according to the data transmission speed, processing speed, data volume and computational amount of the server node; the response delay of task i∈N offloaded to edge node j∈M can be calculated as:

7. The edge computing task scheduling method based on genetic simulated annealing algorithm according to claim 1 is characterized in that: The specific meanings of the decision variables are: The objective function of the nonlinear optimization model is: Among them, the objective function represents minimizing the maximum value of task response delay and resource matching degree in all server nodes; Among them, the constraints of the nonlinear optimization model are as follows: Constraint 1 means that any task can only be assigned to one server, Constraint 2 means that each task can only be assigned to a server node that meets the memory requirements, Constraint 3 means that each task can only be assigned to a server node that meets the hard disk requirements, Constraint 4 means that each task can only be assigned to a server node that meets the computing module resource requirements, and Constraint 5 means the value range of the decision variable.

8. An edge computing task scheduling system based on genetic simulated annealing algorithm, applied to the edge computing task scheduling method based on genetic simulated annealing algorithm according to any one of claims 1 to 7, characterized in that: The system comprises: The task collection module is used to determine the task set generated by the industrial field equipment and obtain the basic information of the task; The server cluster module is used to determine the node set of the server cluster and obtain the basic information of the server nodes; The model building module is used to set decision variables and build nonlinear programming models. The decision module is used to solve the nonlinear optimization model and determine the task scheduling decision by calling the edge computing task scheduling method based on the genetic simulated annealing algorithm according to the task set N and the server set M as well as the basic information and decision variables.

9. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the edge computing task scheduling method based on the genetic simulated annealing algorithm described in any one of claims 1 to 7 is implemented.

Images