Novel IIOT-MEC task unloading method oriented to time delay and energy consumption joint optimization
By establishing mathematical models and nonlinear multi-constraint optimization problems in the industrial Internet of Things, combined with the improved Gray Wolf Algorithm (FIGWO), the problem of offload delay and energy consumption optimization of mobile edge computing tasks is solved, and the total system overhead is reduced and Improvement of energy efficiency.
Patent Information
- Application Number
- CN202411456749.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-18
- Publication Date
- 2024-11-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the industrial Internet of Things, when mobile edge computing tasks are unloaded, the existing technology is difficult to effectively optimize delay and energy consumption, resulting in a large total system overhead, affecting the energy continuity of industrial mobile devices and the energy consumption efficiency of edge nodes.
A new IIoT-MEC task unloading method that is optimized for delay and energy consumption is adopted. By establishing mathematical models and nonlinear multi-constraint optimization problems, combined with the improved Gray Wolf Algorithm (FIGWO), the task unloading strategy is studied, and the Lai is integrated into The control parameter mechanism of dimension flight distribution and adaptive adjustment optimizes the total system overhead.
It effectively reduces the total system overhead, improves the delay and energy consumption trade-offs of task offloading, improves the energy efficiency and computing resource utilization of the system, and ensures the energy continuity of industrial mobile devices and the low-energy-consuming operation of edge nodes.
Smart Images

Figure CN118972903A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of mobile edge computing task offloading in industrial Internet of Things, and in particular to a novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption. Background Art
[0002] The combination of IoT technology and industrial technology has led to the rapid development of the Industrial Internet of Things (IIoT). More and more industrial intelligent mobile devices (such as mobile robots, automatic guided vehicles, etc.) have emerged, accelerating the transformation and development of Industry 4.0. However, the computing power and battery power of mobile devices are limited, and the disadvantages of the cloud computing model that offloads computing tasks to remote cloud processing are becoming increasingly apparent. Mobile Edge Computing (MEC) has emerged as an extension of cloud computing, offloading delay-sensitive or computing-intensive tasks to edge nodes close to industrial mobile devices for processing, effectively reducing the transmission delay of tasks sent to remote cloud centers and reducing the energy consumption of mobile devices.
[0003] According to the energy conservation and carbon reduction policy of industrial enterprises, considering the energy continuity of industrial mobile equipment and the high energy consumption of edge nodes, optimizing system energy consumption is of great research significance. Therefore, the present invention conducts research on IIoT-MEC task offloading for the joint optimization of latency and energy consumption. Summary of the invention
[0004] The purpose of the present invention is to provide a novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption, so as to solve the optimization problem of minimizing the total system overhead under the IIOT-MEC architecture.
[0005] To achieve the above objectives, the present invention provides a novel fusion improved grey wolf algorithm to study the task offloading strategy for joint optimization of delay and energy consumption, comprising the following steps: S1. Establish a mathematical model based on the industrial Internet of Things mobile edge computing architecture scenario, using four computing offloading methods: local mobile device computing, offloading computing to MEC servers, offloading computing to idle MEC servers, and offloading computing to idle mobile devices; S2, establish a nonlinear multi-constraint optimization problem based on the S1 mathematical model; S3. Combine the LFD algorithm and integrate the improved grey wolf optimization FIGWO algorithm to solve nonlinear multi-constraint optimization problems.
[0006] Preferably, the specific process of establishing the mathematical model in S1 is as follows: according to the characteristic parameters of IIOT-MEC involving mobile edge servers, industrial mobile devices, wireless transmission channels and MEC servers, a mathematical model is established according to local calculation, offloading to MEC servers, offloading to idle mobile devices and offloading to idle MEC servers.
[0007] Preferably, the specific process of establishing the nonlinear optimization problem in S2 is as follows: S21, establishing a local mobile device calculation expression; S22. Establishing a MEC server task offloading expression; S23, establishing an idle mobile device task offloading expression; S24, establishing an idle MEC server task offloading expression; S25, establishing an optimization problem constraint expression; S26, establishing an optimization objective function expression; S27. Under the constraints in S25, the entire optimization problem is established with the goal of maximizing the expression of the objective function in S26.
[0008] Preferably, the local mobile device of S21 calculates the delay And the energy consumption is as follows: (1) (2) in, The amount of task data allocated to the local mobile device for offloading decisions, represents the cycle frequency of the local mobile device, represents the task to be calculated on the nth mobile device, Indicates the computing power of the mobile device or MEC server, Representation Task The amount of data that needs to be unloaded, Indicates the proportion of tasks offloaded to local mobile devices, Indicates the rated power of the local mobile device.
[0009] Preferably, the task data transmission delay of S22 , Task calculation delay , Waiting for return delay , the total delay of task offloading to MEC server Total energy consumption as follows: (3) (4) (5) (6) (7) in, The amount of task data allocated to the local mobile device for offloading decisions, Indicates the task transfer rate, Indicates the cycle frequency of the MEC server, Indicates the MEC server processing task The CPU frequency at Indicates the MEC server output data volume coefficient, is the upload power of the local device, is the rated power of the MEC server.
[0010] Preferably, the task data transmission delay in S23 , calculate delay , Waiting return delay for task offloading on idle mobile devices , the total delay of task offloading to idle mobile devices Total energy consumption as follows: (8) (9) (10) (11) (12) Among them, the amount of task data assigned to the idle mobile device by the offloading decision is , the average transmission delay between local mobile devices and idle mobile devices , represents the cycle frequency of idle mobile devices, Indicates idle mobile device processing tasks The CPU frequency at represents the idle mobile device output data volume coefficient, is the power rating of the idle mobile device.
[0011] Preferably, the task data transmission delay in S24 , computing delay of idle MEC servers , waiting time for return of idle MEC server task offloading , the total delay of task offloading to the idle MEC server Total energy consumption as follows: (13) (14) (15) (16) (17) in, The amount of task data allocated to the idle MEC server for offloading decision, Indicates the cycle frequency of idle MEC servers, Indicates the idle MEC server output data volume coefficient, Indicates idle MEC server processing tasks The CPU frequency at is the rated power of the idle MEC server.
[0012] Preferably, the constraint expressions and objective function expressions of S25 and S26 are as follows: (18) (19) (20) in, represents the total system delay, represents the total energy consumption of the system, represents the delay weight factor of task processing in the system utility, represents the energy consumption weight factor of task processing in the system utility, , Indicates computing tasks on mobile devices, Indicates the transmission bandwidth, Represents four task offloading schemes, and constraint C1 represents the weight factor and Value constraints; C2 and C3 represent tasks The sum of the ratios of offloading to local mobile devices, MEC servers, idle mobile devices, and idle MEC servers is 1; C4 is the system communication bandwidth constraint, is the upper limit of the system communication bandwidth; C5 means that the computing resources of the MEC server allocated to each task are not greater than its own; C6 means that the computing resources of the idle mobile devices allocated to each task are not greater than its own; C7 means that the resources of the idle MEC server allocated to each task are not greater than its own; C8 means that the sum of the resources allocated to all tasks does not exceed itself.
[0013] Preferably, the improvement steps of the Grey Wolf optimization algorithm in S3 are as follows: SA1. Combine the Levy flight distribution with the Grey Wolf optimization algorithm, as follows: FIGWO algorithm is obtained by integrating LFD algorithm and random probability factor on the basis of GWO algorithm, and gray wolf individual selection is performed to define the random step length formula of Levy flight. LFD ( s ): (twenty one) in, and Obey and ,parameter Take a constant on the interval [0, 2], Calculated by formula (22): (twenty two) According to the LFD algorithm rules, a random probability factor is introduced ,when When , the gray wolf group updates the individual gray wolf positions according to the optimal solution of the optimization problem and the hunting behavior of the LFD algorithm; define represents the individual position vector of the gray wolf, is the global optimal solution of the algorithm, is the constant coefficient in the screw equation, is the current iteration number, is a random number in [-1,1]; When , the algorithm individual performs encirclement calculation according to formula (23): (twenty three) when When , the algorithm individuals perform encirclement calculations according to formulas (24)-(27); where is the optimal solution, 、 are the second and third best solutions respectively. , , Represents the three leading grey wolves in the algorithm; and is the convergence factor of the algorithm, the control parameter As the number of iterations increases, it decreases from 2 to 0. and is a vector of random values in [0, 1]; (twenty four) (25) (26) In summary, the gray wolf individual position update formula for: (27) SA2. Improved control parameters of FIGWO algorithm , as follows: FIGWO algorithm has a mechanism for adaptively adjusting control parameters. With the number of iterations The increase of decreases linearly from 2 to 0, making the convergence factor exist[- n , n ] changes within; when When , the wolf pack disperses to conduct a global search; when When , the wolf pack concentrates near the global optimal solution; in the GWO algorithm, Using linear decrease: (28) in, T represents the maximum number of iterations, , ; A nonlinear control parameter strategy is proposed for the convergence of FIGWO algorithm, as shown in formula (29). Non-linear decreasing: (29) SA3. Update the position weights in the GWO algorithm as follows: Introducing a proportional weight, the dynamic update formula of the individual position of the gray wolf is shown in (30) and (31), where: , the weight setting has different influence ratios on the optimal solution for different gray wolf individuals: (30) (31) in, , , Indicates the weight of the optimal solution’s impact ratio; , , Indicates the location of the gray wolf.
[0014] Therefore, the present invention adopts the above-mentioned novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption. The present invention combines the LFD algorithm to improve and propose a fusion-improved gray wolf optimization, which mainly includes Levy flight distribution combined with the gray wolf optimization algorithm, improved control parameters and improved position weight update scheme.
[0015] In order to reduce the total system overhead under the IIoT-MEC architecture, the present invention conducts a task offloading study for the joint optimization of latency and energy consumption. According to the four task offloading methods, the latency model and energy consumption model are respectively constructed, and the two are integrated into the total system overhead. The weight factors of latency and energy consumption are introduced to weigh the two, and the constraints such as communication bandwidth, mobile devices and MEC server computing power are considered to minimize the total system overhead as the objective function. According to the hunting method of the GWO algorithm combined with the LFD algorithm, the FIGWO algorithm is further proposed to solve the optimization problem. The proposed algorithm can effectively reduce the total system overhead while having a fast convergence speed and convergence accuracy.
[0016] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 This is a flow chart of a novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption according to the present invention; Figure 2 This is a system architecture diagram of a novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption in the present invention. DETAILED DESCRIPTION
[0018] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.
[0019] Reference Figure 1-Figure 2 The present invention provides a novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption, and the steps are as follows: Step S1, establish a mathematical model according to the industrial Internet of Things mobile edge computing architecture scenario, use four computing offloading methods, local mobile device computing, offloading to MEC server computing, offloading to idle MEC server computing, and offloading to idle mobile device computing. The specific methods are as follows: like Figure 2,The tasks that need to be offloaded are initiated by mobile devices on the industrial site, including four ways of offloading: local mobile device computing, offloading to MEC servers, offloading to idle MEC servers, and offloading to idle mobile devices. The MEC server makes offloading decisions based on task information, its own computing power, and carrying capacity. If the task is offloaded, the amount of tasks that can be offloaded is calculated. When the computing resources of the MEC server are insufficient, the task is sent to a nearby MEC server with sufficient resources or an idle mobile device for collaborative computing offloading, and the offloading decision is sent back to the local mobile device.
[0020] System Included Industrial field mobile equipment, define the industrial field mobile equipment offloading task set , assuming that the tasks are divisible, each task can be fully or partially offloaded to any computing device. , , and They represent the proportion of tasks offloaded to local mobile devices, MEC servers, idle mobile devices, and idle MEC servers. Each task has the following attributes, which are represented by the set express, Indicates n tasks to be computed on mobile devices, is the cycle frequency, that is, the number of CPU cycles required to calculate 1 bit of data, Indicates the computing power of the mobile device or MEC server, Representation Task The amount of data that needs to be unloaded, Indicates processing task The required delay.
[0021] 1.1 Local Mobile Device Computing When computing on the device, there is no need to transmit task data through the network. Therefore, there is no transmission delay in the local mobile device computing model, only computing delay. The amount of task data allocated to the local mobile device by the offloading decision is , Represents the cycle frequency of the local mobile device and the calculation delay of the local mobile device It can be deduced from formula (1): (1) The energy consumption of local mobile devices when processing tasks As shown in formula (2), Indicates the rated power of the local mobile device.
[0022] (2) 1.2 MEC Server Task Offloading When a task is offloaded to the MEC server for computation, it needs to transmit the task data through the network. The delay of offloading the task to the MEC server is divided into: task data transmission delay , task calculation delay , waiting for return delay The amount of task data assigned to the local mobile device by the offloading decision is , Indicates the cycle frequency of the MEC server, Indicates the MEC server processing task The CPU frequency at Indicates the task transmission rate, taking into account the data return time, introducing Indicates the MEC server output data volume coefficient.
[0023] (3) (4) (5) Total latency of task offloading to MEC server It can be deduced from formula (6) that the total energy consumption As shown in formula (7). is the upload power of the local device, is the rated power of the MEC server.
[0024] (6) (7) 1.3 Task offloading for idle mobile devices When local mobile devices and MEC servers face intensive industrial field tasks and their computing resources are insufficient, the tasks will be offloaded to idle mobile devices for processing. The transmission path is: local mobile device-MEC server-idle mobile device. The delay of task offloading to idle mobile devices includes: task data transmission delay , where the amount of task data assigned to the idle mobile device by the offloading decision is , the average transmission delay between local mobile devices and idle mobile devices ; Calculate delay It can be deduced from formula (9): represents the cycle frequency of idle mobile devices, Indicates idle mobile device processing tasks The CPU frequency at that time; Formula (10) is the waiting time for backhaul when the idle mobile device is performing task offloading , Represents the output data volume coefficient of idle mobile devices.
[0025] (8) (9) (10) Total latency of task offloading to idle mobile devices It can be deduced from formula (11) that the total energy consumption As shown in formula (12). is the power rating of the idle mobile device.
[0026] (11) (12) 1.4 Uninstalling Idle MEC Servers When a task is offloaded to an idle MEC server for computation, data needs to be transmitted over the network. The delay includes: task data transmission delay As shown in formula (8a), the amount of task data allocated to the idle MEC server by the offloading decision is , the distance between MEC servers is short and the communication mode is wired transmission, so the average transmission delay is ignored; the calculation delay of idle MEC servers Calculated by formula (14), Indicates the cycle frequency of idle MEC servers, Indicates idle MEC server processing tasks The CPU frequency at this time; Formula (15) is the waiting return delay of the idle MEC server task offloading , Indicates the idle MEC server output data volume coefficient.
[0027] (13) (14) (15) Total latency of task offloading to idle MEC servers As shown in formula (16), the total energy consumption As shown in formula (17). is the rated power of the idle MEC server.
[0028] (16) (17) Step S2, establish a nonlinear multi-constraint optimization problem based on the mathematical model of S1, the specific method is as follows: The above four task offloading models are constructed as a multi-constrained optimization problem. Equations (18) and (19) represent the total system delay respectively. and total energy consumption : (18) (19) (20) in, Indicates computing tasks on mobile devices, Indicates the transmission bandwidth, Indicates four task offloading schemes.
[0029] The research goal of this embodiment is to balance the latency and energy consumption of IIoT-MEC task offloading. A utility function that can reflect the effect of task offloading decision is designed in the objective function of the optimization problem. To represent the total system overhead, that is , represents the delay weight factor of task processing in the system utility, represents the energy consumption weight factor of task processing in the system utility, and requires , and The value of is set according to the needs of the local mobile device. In summary, the system optimization problem considered is modeled as formula (20). Among them, , constraint C1 represents the weight factor and Value constraints; C2 and C3 represent tasks The sum of the ratios of offloading to local mobile devices, MEC servers, idle mobile devices, and idle MEC servers is 1; C4 is the system communication bandwidth constraint, is the upper limit of the system communication bandwidth; C5 means that the computing resources of the MEC server allocated to each task are not greater than its own; C6 means that the computing resources of the idle mobile devices allocated to each task are not greater than its own; C7 means that the resources of the idle MEC server allocated to each task are not greater than its own; C8 means that the sum of the resources allocated to all tasks does not exceed itself.
[0030] Step S3, combining the LFD algorithm with the improved grey wolf optimization FIGWO algorithm to solve the nonlinear multi-constraint optimization problem, the specific method is as follows: The optimization problem (20) is a complex nonlinear multi-constrained optimization problem. Directly using the GWO algorithm to solve it will bring defects. The most obvious feature is that it is prone to premature convergence and the convergence accuracy is not high when facing complex optimization problems. In addition, due to the GWO algorithm The wolf is not necessarily the global optimal solution, which leads to falling into the local optimal solution during the iteration process. This embodiment combines the LFD algorithm and proposes a fusion-improved grey wolf optimization FIGWO algorithm.
[0031] 3.1 Levy flight distribution combined with gray wolf optimization algorithm FIGWO algorithm integrates LFD algorithm and random probability factor to select individual gray wolves based on GWO algorithm, aiming to enrich the diversity of gray wolf population and enhance the search range of the algorithm. LFD algorithm is inspired by the random walk of Levy flight, and uses the random step length of LFD to expand the search range, explore the unknown large search space, and define the Levy random step length formula LFD ( s ): (twenty one) in, and Obey and ,parameter Take a constant on the interval [0, 2], It can be calculated by formula (22): (twenty two) According to the LFD algorithm rules, a random probability factor is introduced ,when When , the gray wolf group updates the position of the individual gray wolf according to the position of the optimal solution of the optimization problem combined with the hunting behavior of the LFD algorithm, and defines represents the individual position vector of the gray wolf, is the global optimal solution of the algorithm, is the constant coefficient in the screw equation, is the current iteration number, is a random number in [-1,1]. When , the algorithm individual performs encirclement calculation according to formula (23): (twenty three) when When , the algorithm individuals perform encirclement calculations according to formulas (24)-(27). Among them, is the optimal solution, 、 are the second and third best solutions respectively. , , Represents the three leading grey wolves in the algorithm; and is the convergence factor of the algorithm, the control parameter As the number of iterations increases, it decreases from 2 to 0. and is a vector of random values in [0, 1].
[0032] (twenty four) (25) (26) Based on the above, the gray wolf individual position update formula is for: (27) 3.2 Improved control parameters and position weights Improved algorithm control parameters :FIGWO algorithm has a mechanism for adaptively adjusting control parameters. With the number of iterations The increase of decreases linearly from 2 to 0, making the convergence factor exist[- n , n ] changes. When , the wolf pack disperses to conduct a global search; when When , the wolf pack concentrates near the global optimal solution. In the traditional GWO algorithm, Using linear decrease: (28) in, T represents the maximum number of iterations, , Since the process of FIGWO algorithm to find the global optimal solution is nonlinear, it is only necessary to linearly reduce It cannot completely balance the search capabilities of the global optimal solution and the local optimal solution. Therefore, a nonlinear control parameter strategy is proposed for the convergence of the FIGWO algorithm, as shown in formula (29). Nonlinear decreasing, when the algorithm starts The decline rate is slower, which enhances the global search ability of the algorithm. The decreasing speed of becomes larger, which improves the convergence speed of the algorithm.
[0033] (29) Update position weight: In the GWO algorithm, The position of the wolf may not be the global optimal solution, and it is difficult for other gray wolf individuals to jump out of their own local optimal solutions, which leads to a significant increase in algorithm processing time and reduced convergence performance. In order to improve the solution accuracy and convergence speed of the GWO algorithm, a new proportional weight is introduced inspired by the custom weight. The dynamic update formula of the gray wolf individual position is shown in (30) and (31). Among them, , this weight setting has different influence proportions on the optimal solution for different gray wolf individuals, so that the algorithm can search for the optimal solution more effectively.
[0034] (30) (31) in, , , Indicates the weight of the optimal solution’s impact ratio; , , Indicates the location of the gray wolf.
[0035] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.
Claims
1. A novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption, characterized in that: The following steps are involved: S1. Establish a mathematical model based on the industrial Internet of Things mobile edge computing architecture scenario, using four computing offloading methods: local mobile device computing, offloading computing to MEC servers, offloading computing to idle MEC servers, and offloading computing to idle mobile devices; S2, establish a nonlinear multi-constraint optimization problem based on the S1 mathematical model; S3. Combine the LFD algorithm and integrate the improved grey wolf optimization FIGWO algorithm to solve nonlinear multi-constraint optimization problems.
2. According to claim 1, a novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption is characterized in that: The specific process of establishing the mathematical model in S1 is as follows: according to the characteristic parameters of the mobile edge server, industrial mobile device, wireless transmission channel and MEC server involved in IIOT-MEC, a mathematical model is established according to local computing, offloading to MEC server, offloading to idle mobile devices and offloading to idle MEC server.
3. According to claim 1, a novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption is characterized in that: The specific process of the nonlinear optimization problem established in S2 is as follows: S21, establishing a local mobile device calculation expression; S22. Establishing a MEC server task offloading expression; S23, establishing an idle mobile device task offloading expression; S24, establishing an idle MEC server task offloading expression; S25, establishing an optimization problem constraint expression; S26, establishing an optimization objective function expression; S27. Under the constraints in S25, the entire optimization problem is established with the goal of maximizing the expression of the objective function in S26.
4. According to a novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption according to claim 1, it is characterized in that: S21 local mobile device calculation latency And the energy consumption is as follows: (1) (2) in, The amount of task data allocated to the local mobile device for offloading decisions, represents the cycle frequency of the local mobile device, represents the task to be calculated on the nth mobile device, Indicates the computing power of the mobile device or MEC server, Representation Task The amount of data that needs to be unloaded, Indicates the proportion of tasks offloaded to local mobile devices, Indicates the rated power of the local mobile device.
5. According to claim 4, a novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption is characterized in that: S22 task data transmission delay , Task calculation delay , Waiting for return delay , the total delay of task offloading to MEC server Total energy consumption as follows: (3) (4) (5) (6) (7) in, The amount of task data allocated to the local mobile device for offloading decisions, Indicates the task transfer rate; Indicates the cycle frequency of the MEC server, Indicates the MEC server processing task The CPU frequency at Indicates the MEC server output data volume coefficient, is the upload power of the local device, is the rated power of the MEC server.
6. According to claim 5, a novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption is characterized in that: Task data transmission delay in S23 , calculate delay , Waiting return delay for task offloading on idle mobile devices , the total delay of task offloading to idle mobile devices Total energy consumption as follows: (8) (9) (10) (11) (12) Among them, the amount of task data assigned to the idle mobile device by the offloading decision is , the average transmission delay between local mobile devices and idle mobile devices , represents the cycle frequency of idle mobile devices, Indicates idle mobile device processing tasks The CPU frequency at represents the idle mobile device output data volume coefficient, is the power rating of the idle mobile device.
7. A novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption according to claim 6, characterized in that: Task data transmission delay in S24 , computing delay of idle MEC servers , waiting time for return of idle MEC server task offloading , the total delay of task offloading to the idle MEC server Total energy consumption as follows: (13) (14) (15) (16) (17) in, The amount of task data allocated to the idle MEC server for offloading decision, Indicates the cycle frequency of idle MEC servers, Indicates the idle MEC server output data volume coefficient, Indicates idle MEC server processing tasks The CPU frequency at is the rated power of the idle MEC server.
8. According to claim 7, a novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption is characterized in that: The constraint expressions and objective function expressions of S25 and S26 are as follows: (18) (19) (20) in, represents the total system delay, represents the total energy consumption of the system, represents the delay weight factor of task processing in the system utility, represents the energy consumption weight factor of task processing in the system utility, , Indicates computing tasks on mobile devices, Indicates the transmission bandwidth, Represents four task offloading schemes, and constraint C1 represents the weight factor and Value constraints; C2 and C3 represent tasks The sum of the ratios of offloading to local mobile devices, MEC servers, idle mobile devices, and idle MEC servers is 1; C4 is the system communication bandwidth constraint, is the upper limit of the system communication bandwidth; C5 means that the computing resources of the MEC server allocated to each task are not greater than its own; C6 means that the computing resources of the idle mobile devices allocated to each task are not greater than its own; C7 means that the resources of the idle MEC server allocated to each task are not greater than its own; C8 means that the sum of the resources allocated to all tasks does not exceed itself.
9. According to claim 8, a novel IIoT-MEC task offloading method for joint optimization of latency and energy consumption is characterized in that: The specific steps for improving the Grey Wolf optimization algorithm in S3 are as follows: SA1. Combine the Levy flight distribution with the Grey Wolf optimization algorithm, as follows: FIGWO algorithm is obtained by integrating LFD algorithm and random probability factor on the basis of GWO algorithm, and gray wolf individual selection is performed to define the random step length formula of Levy flight. LFD ( s ): (21) in, and Obey and ,parameter Take a constant on the interval [0, 2], Calculated by formula (22): (22) According to the LFD algorithm rules, a random probability factor is introduced ,when When , the gray wolf group updates the individual gray wolf positions according to the optimal solution of the optimization problem and the hunting behavior of the LFD algorithm; define represents the individual position vector of the gray wolf, is the global optimal solution of the algorithm, is the constant coefficient in the screw equation, is the current iteration number, is a random number in [-1, 1]; When , the algorithm individual performs encirclement calculation according to formula (23): (23) when When , the algorithm individuals perform encirclement calculations according to formulas (24)-(27); where is the optimal solution, 、 are the second and third best solutions respectively. , , Represents the three leading grey wolves in the algorithm; and is the convergence factor of the algorithm, the control parameter As the number of iterations increases, it decreases from 2 to 0. and is a vector of random values in [0, 1]; (24) (25) (26) In summary, the gray wolf individual position update formula for: (27) SA2. Improved control parameters of FIGWO algorithm , as follows: FIGWO algorithm has a mechanism for adaptively adjusting control parameters. With the number of iterations The increase of decreases linearly from 2 to 0, making the convergence factor exist[- n , n ] changes within; when When , the wolf pack disperses to conduct a global search; when When , the wolf pack concentrates near the global optimal solution; in the GWO algorithm, Using linear decrease: (28) in, T represents the maximum number of iterations, , ; A nonlinear control parameter strategy is proposed for the convergence of FIGWO algorithm, as shown in formula (29). Non-linear decreasing: (29) SA3. Update the position weights in the GWO algorithm as follows: Introducing a proportional weight, the dynamic update formula of the individual position of the gray wolf is shown in (30) and (31), where: , the weight setting has different influence ratios on the optimal solution for different gray wolf individuals: (30) (31) in, , , Indicates the weight of the optimal solution’s impact ratio; , , Indicates the location of the gray wolf.
Citation Information
Patent Citations
Multi-user cooperative computing unloading method based on mobile edge computing
CN111263401A
Task scheduling method and device for server-free edge computing
CN113114758A
MEC-oriented joint computing resource allocation and unloading decision optimization scheme in Internet of Vehicles
CN113364859A
Calculation unloading optimization method based on D2D communication
CN115103405A
Vehicle-mounted computing task unloading method based on deep reinforcement learning strategy
CN115134242A
