Distributed computing cooperative resource allocation method, system, device and storage medium

Abstract

The application discloses a distributed computing cooperative resource allocation method, system, device and storage medium, and belongs to the technical field of communication, and comprises the following steps: establishing an edge node distributed computing system model; according to the edge node distributed computing system model, an overall programming problem is constructed with the target of minimizing the average delay of all users in the distributed computing network; the overall programming problem is solved to determine an offloading dependent task strategy, and the distributed computing cooperative resource is allocated according to the offloading dependent task strategy. The method, system, device and storage medium can minimize the average delay of all users while realizing cooperative resource allocation between edge nodes.

Description

Technical Field

[0001] This invention belongs to the field of communication technology and relates to a method, system, device and storage medium for distributed computing collaborative resource allocation. Background Technology

[0002] Distributed computing deploys computing resources to the network edge, offloading complex computing tasks from users (terminals) to nearby edge servers for execution. This provides an effective solution for latency-sensitive and computationally expensive mobile applications and has been widely applied in cloud platforms and cloud services. The latency caused by offloading computationally intensive tasks to edge servers directly impacts the user's service quality. To minimize latency while ensuring a good user experience, researchers have conducted extensive research using optimization theory, Markov decision processes, game theory, reinforcement learning, and heuristic algorithms.

[0003] Ning et al. formulated the single-user computation offloading problem as a mixed-integer linear programming (MILP) problem and solved it using a branch-and-bound algorithm. Chen et al., considering global information from all edge servers, BSs, and tasks, formulated the latency-minimizing task offloading problem as a mixed-integer nonlinear programming (MINLP) problem and simplified it into a computation offloading scheduling and resource allocation problem. Taking advantage of the improved transmission efficiency of non-orthogonal multiple access (NOMA), Liu et al. reduced the latency of computation offloading by selecting appropriate wireless channels for each vehicle. They modeled the offloading problem among multiple vehicles as a game and achieved Nash equilibrium. Zhu et al. studied the computation offloading problem of vehicles in a multi-task computing environment to minimize the long-term task processing latency of vehicles. This problem has been modeled as a multi-agent DRL problem to select appropriate edge servers from multiple vehicles.

[0004] The latency minimization schemes mentioned in the existing work have a single task model and do not take into account the task dependency problem that exists in the real world. In fact, whether or not task dependency is considered has a huge impact on the practicality of the computation offloading scheme, so the accuracy is poor and it cannot minimize the average latency of all users while realizing collaborative resource allocation between edge nodes. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a distributed computing collaborative resource allocation method, system, device and storage medium that can minimize the average latency of all users while realizing collaborative resource allocation among edge nodes.

[0006] In one aspect, the present invention provides a distributed computing collaborative resource allocation method, comprising:

[0007] Establish a distributed computing system model for edge nodes;

[0008] Based on the edge node distributed computing system model, an overall planning problem is constructed with the goal of minimizing the average latency of all users in the distributed computing network.

[0009] Solve the overall planning problem to determine the strategy for unloading dependent tasks, and allocate distributed computing collaborative resources according to the strategy for unloading dependent tasks.

[0010] The further improvement of the distributed computing collaborative resource allocation method described in this invention lies in:

[0011] The overall planning problem is:

[0012]

[0013]

[0014] Among them, T m M is the difference between the time it takes for all subtasks to complete and the time the task was created, where M is the number of users. E represents the set of users. u E represents the total energy of edge node u. max This represents the maximum energy of the edge node. Denotes the UAV set, x u (m, ta) j ) represents the subtask ta of task m. j The unloading decision for being unloaded to edge node u, y u′ (m, ta) j ) indicates that the subtask ta of task m is... j The decision variables for moving from u to other edge nodes u′ Let represent the set of dependent subtasks in task m, x represent the time when all subtasks are completed, and y represent the task creation time.

[0015] The difference T between the time it takes for all subtasks in task m to be completed and the task creation time. m for:

[0016]

[0017] in, This represents subtask ta in task m. j Final completion time, This represents subtask ta in task m. j The start time, N m Let m be the total number of subtasks in task m.

[0018] The total energy E of the edge node uu Represented as:

[0019]

[0020] in, The edge node represents the subtask ta in task m. j energy, This represents the transmission energy generated when edge node u transmits part of the task to other edge nodes u′ via a wireless link.

[0021] The edge nodes are used to compute subtask ta in task m. j energy for:

[0022]

[0023] Where, k u To effectively convert capacitors, This indicates that the edge node is in the computation subtask ta. j Energy consumption per CPU cycle For task m, subtask ta j Data input size, c u This represents the number of CPU cycles required for edge node u to execute one bit of data.

[0024] The transmission energy generated when edge node u transmits part of the task to other edge nodes u′ via a wireless link for:

[0025]

[0026] in, For task m, subtask ta j The data input size, v(u, u′) represents the data transmission rate between edge node u and edge node u′, p u This represents the transmission power of edge node u.

[0027] The process of solving the overall planning problem, determining the strategy for unloading dependent tasks, and allocating distributed computing collaborative resources according to the strategy for unloading dependent tasks is as follows:

[0028] The overall planning problem is transformed into a penalty-based approach.

[0029] The whale optimization algorithm is used to solve the overall planning problem, and a strategy for unloading dependent tasks is obtained.

[0030] Distributed computing collaboration resources are allocated according to the aforementioned unloading dependency task strategy.

[0031] In a second aspect, the present invention provides a distributed computing collaborative resource allocation system, comprising:

[0032] Establish a module to build a distributed computing system model for edge nodes;

[0033] The construction module is used to construct an overall planning problem based on the edge node distributed computing system model, with the goal of minimizing the average latency of all users in the distributed computing network;

[0034] The allocation module is used to solve the overall planning problem, determine the strategy for unloading dependent tasks, and allocate distributed computing collaborative resources according to the strategy for unloading dependent tasks.

[0035] In three aspects, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the steps of the distributed computing collaborative resource allocation method.

[0036] In four aspects, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the distributed computing collaborative resource allocation method.

[0037] The present invention has the following beneficial effects:

[0038] In practical operation, the distributed computing collaborative resource allocation method, system, device, and storage medium described in this invention establish an edge node distributed computing system model, enabling edge nodes to share workloads. Simultaneously, with the goal of minimizing the average latency of all users in the distributed computing network, an overall planning problem is constructed and then solved to determine the strategy for unloading dependent tasks. By comprehensively considering the task scheduling of dependent subtasks, collaborative resource allocation among edge nodes is achieved while minimizing the average latency of all users. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the distributed computing scenario of the present invention;

[0040] Figure 2 This is a schematic diagram of the subtask dependency DAG of the present invention;

[0041] Figure 3 This is a system structure diagram of the present invention. Detailed Implementation

[0042] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, not all embodiments, and are not intended to limit the scope of the present invention. Furthermore, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessary confusion regarding the concepts disclosed in the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort should fall within the scope of protection of the present invention.

[0043] The accompanying drawings show structural schematic diagrams according to embodiments disclosed in this invention. These drawings are not drawn to scale, and some details have been enlarged for clarity, and some details may have been omitted. The shapes of the various regions and layers shown in the drawings, as well as their relative sizes and positional relationships, are merely exemplary and may deviate from reality due to manufacturing tolerances or technical limitations. Furthermore, those skilled in the art can design regions / layers with different shapes, sizes, and relative positions as needed.

[0044] Example 1

[0045] The distributed computing collaborative resource allocation method of the present invention includes the following steps:

[0046] 1) Establish a distributed computing system model for edge nodes;

[0047] The operation process of step 1) is as follows:

[0048] 11) Establish task model: Model the task generation and allocation mechanism of the distributed computing system, model the dependency relationship and dependent data volume between subtasks, and consider the start time, execution time, completion time and transmission delay of subtasks.

[0049] like Figure 1 As shown, in a specific area, there is a set of edge nodes that provide communication and computing services to users; this is the UAV set. Indicates a user set Assuming information sharing among edge nodes, during resource allocation, the edge nodes are responsible for selecting UAV offloading tasks and transmitting the decision information to users and other edge nodes. Edge nodes cover designated areas; that is, user requests within that area are handled by the corresponding edge node. Each user can only generate one task at a time; therefore, a task created by user m is sometimes referred to as m. Different users have different needs, and the dependencies between subtasks within a task also differ, such as... Figure 2 As shown, the task can be modeled as a directed acyclic graph. in, Let m be the set of dependent subtasks. N represents the dependency relationship between subtasks in task m. m Let be the number of subtasks in task m. When a subtask has been unloaded to a specific UAV, the execution time of completing the task depends on computing power and the size of the input data for the subtask. This represents subtask ta in task m. i With subtask ta j The amount of dependent data between them. When the preceding subtask of a subtask is on different UAVs, then in subtask ta j Before execution begins, the dependency data of all preceding subtasks needs to be transferred to the server responsible for executing that subtask. j On the UAV, set This indicates that the subtask ta j The set of preceding subtasks, then This represents the dependency data of all preceding subtasks. Of course, not every two tasks have a dependency relationship, and there is no dependency data between subtasks that have no dependency relationship.

[0050] The user creates tasks locally, but the user's mobile device has low computing power and battery life. Therefore, all tasks are transferred to the UAV for execution when the base station makes a decision. This process takes time, and the subtasks are transmitted. j The input data delay is set to Then user m's subtask ta j Arrival time for:

[0051]

[0052] in, Let m be the creation time of task m. The start time of a subtask depends not only on its arrival time but also on the completion time of its preceding subtask and the transmission delay of its dependent data. The relationship between the start time of a subtask and the time variable is as follows:

[0053]

[0054] in, This indicates the transfer from other devices to the execution of subtask ta. j The transmission latency depends on the amount of data, in subtask ta j When multiple result data are required, a certain period of time must be waited until the last preceding subtask completes and is transmitted to the current device. Of course, if the preceding subtask is completed by the current device, the transmission delay is negligible. The transmission delay between different UAVs depends on the amount of data sent and the channel quality between the UAV and the edge node, i.e.:

[0055]

[0056] in, This indicates two subtasks. i The amount of dependent data between them, v(ta) i ,ta j This indicates the data rate between UAVs. Subtask ta j Final completion time For subtask ta j The sum of the start time and the execution time, i.e.:

[0057]

[0058] in, This represents the computation time for the task.

[0059] 12) Establish a communication model: Model the communication channel between the user and the edge node, jointly consider the path loss between user m and edge node u, and determine the channel gain, the transmission delay from user m to edge node u, and the task completion time representation;

[0060] The uplink data rate v(m, u) between user m and edge node u is:

[0061]

[0062] Where, p m Let g(m, u) represent the transmission power of user m, g(m, u) represent the channel gain between the user and the edge node, and σ represent the transmission power of user m. 2 Let B represent the variance of Gaussian white noise, B represent the channel bandwidth, and the channel gain g(m, u) be:

[0063] g(m, u) = 10 -δ(m，u) / 10

[0064] Where δ(m, u) represents the path loss between user m and edge node u. Since there may be buildings or vegetation blocking the path between the edge node and the user or base station, the link needs to consider not only the LOS channel but also the NLOS channel. Therefore, the path loss is a combination of the two path losses, which are defined in the formula as follows:

[0065]

[0066] Among them, f c The carrier frequency is represented by c, the speed of light is represented by L. LoS and L NLoS Let d(m, u) represent the additional loss of the two types of links, and let d(m, u) represent the distance between the user and the edge node. The link between the edge node and the user has a Loss of S (LoS), i.e.:

[0067]

[0068] Among them, h u Let represent the height of edge node u, C and B be variables determined by the environment, and d(m, u) be the distance between edge node u and user u. Then, the path loss δ(m, u) from user u to the associated edge node u is:

[0069] δ(m, u) = P(m, u) LoS ·δ(m, u) LoS +(1-P(m,u) LoS )·δ(m,u) NLoS

[0070] Therefore, the transmission delay t from user m to edge node u tran (m, u) can be determined by the amount of data in the subtask and the data transmission rate, i.e.

[0071]

[0072] in, Indicates subtask ta j The data input size is N. m Each subtask is a subtask. The base station, acting as the decision-maker, decides which device to offload the subtask to. This offloading decision, representing the subtask being offloaded to the associated edge node u, can be defined as x. u (m, ta) j The unloading decision x u (m, ta) j ) is a binary variable, when x u (m, ta) j When ) is 1, it indicates that user m's subtask ta j It is unloaded to the associated edge node u when x u (m, ta) j When ) is 0, it means that the data is not unloaded to the edge node u, therefore the transmission subtask ta is not performed. j Delay when inputting data for:

[0073]

[0074] Among them, when subtask ta j When unloading to edge node u, the subtask ta is executed by edge node u. j Calculation time for:

[0075]

[0076] in, For edge nodes, for subtask ta j The computing resources are determined based on the input size of all subtasks executed at UAVu, i.e. for:

[0077]

[0078] Among them, f max This represents the maximum available computing resources of the edge node. It is essential to ensure that the total computing power of the UAV, based on the number of subtasks from different users, does not exceed the maximum computing power of the edge node.

[0079]

[0080] When the computing power of edge node u is insufficient to execute the task alone, the edge node will transfer part or all of the task to other edge nodes via a wireless link, defining a binary variable y. u′ (m, ta) j The decision variable y represents whether to transfer user m's computation task from edge node u to another edge node u′. When user m's subtask is transferred to edge node u′ for execution, the decision variable y... u′ (m, ta) j When ) = 1, the transmission delay between edge nodes is... It is determined by the amount of data transmitted and the transmission rate, that is:

[0081]

[0082] Where v(u, u′) represents the data transmission rate between edge nodes, and v(u, u′) is:

[0083]

[0084] Where B represents bandwidth, and g(u, u′) represents channel gain between edge nodes. Since edge nodes are highly efficient, we assume the link between edge nodes is a Loss-of-Stake (LoS) link, then g(u, u′) is:

[0085]

[0086] When subtask ta j When task m is created and unloaded to the collaborative edge node u′, each subtask will eventually be unloaded to an edge node in the network. The decision variables then satisfy the following constraints:

[0087]

[0088] Task completion time T mThis is the difference between the time it takes for all subtasks to complete and the task creation time, i.e.:

[0089]

[0090] 13) Establish an energy model: jointly consider calculation of energy consumption and flight energy consumption;

[0091] Edge nodes in computation subtask ta j The computational energy consumed is [amount] per CPU cycle. Where, k u For effective switching capacitance, k u The value of depends on the chip structure of the edge node; therefore, the edge node is used to compute the subtask ta. j energy for:

[0092]

[0093] Transmission energy is generated when an edge node transmits part of its tasks to other edge nodes via a wireless link. for:

[0094]

[0095] The total energy of edge node u is the sum of the computational energy for executing subtasks and the transmission energy generated by forwarding the input data of the subtasks to adjacent UAVs. The energy consumption of the edge node hovering at a fixed height is mechanical energy, which is not part of the energy consumed by computation and communication, so it is not considered for the time being. Therefore, the total energy E of edge node u is... u :

[0096]

[0097] 2) Minimize the average latency of all users in the distributed computing network and complete the resource allocation for offloading decisions in the collaborative edge node group.

[0098] The operation process for step 2) is as follows:

[0099] 21) The objective problem is an integer programming problem, i.e.

[0100]

[0101]

[0102] In this objective problem, each subtask in the user task is assigned to edge nodes for computation. This is a generalized allocation problem, an NP-hard problem based on combinatorial optimization theory, aiming to find the optimal allocation scheme to maximize or minimize a specific objective function. In the generalized allocation problem, there is a fixed number of tasks and resource sets, each task has a set of available resources and an associated cost or benefit. Heuristic algorithms can be attempted to solve this objective problem. In this invention, the whale optimization algorithm is used to solve the objective problem, and simulations are performed for comparison. Before deploying the algorithm, the unloaded variables x and y are merged into a single vector. Among them, Z m This represents the unloading vector for task m. Indicates subtask ta j The unloading decision instructs the edge node ultimately responsible for executing the subtask, setting... The target problem is then transformed into a new target problem:

[0103]

[0104]

[0105] 22) Employ a penalty method based on the objective function to transform the new objective problem into a penalty form;

[0106] Specifically, the total latency, which is the transmission latency between edge nodes, can be calculated using each decision matrix obtained from the search agent. and computational delay The sum of . The transmission delay between edge nodes is the time to transmit a subtask from its associated UAV to other UAVs in the network. The new objective problem is transformed into a penalty form using a penalty method based on the objective function. Specifically, the penalty form is:

[0107]

[0108]

[0109] Wherein, ε(E) max E u Let ε(E) be a step function. When the energy consumption given by the unloading decision exceeds the maximum energy of the edge node, then ε(E) max E u Return 1 if ε(E) is not returned, otherwise ε(E) is returned. max E uReturns 0. In this invention, the right-hand side of the objective function is represented as a penalty term, and λ is the penalty coefficient, which is used to manipulate the penalty value. When the penalty value is too small, the algorithm may converge to an infeasible solution and accept the penalty. On the other hand, a severe penalty will prevent the agent from exploring more promising regions and stay in the comfort zone, i.e., it will not obtain a better solution.

[0110] 3) Employ a metaheuristic approach to find suboptimal solutions: Use the Whale Optimization Algorithm (WOA) to determine the strategy for unloading dependent tasks;

[0111] The specific process of step 3) is as follows:

[0112] First, in the prey-encirclement phase, the optimal solution is unknown at the start of the search. The WOA algorithm uses the current optimal solution as the target prey so that other search agents can attempt to move towards the optimal position and update their own positions.

[0113]

[0114]

[0115] in, This represents the current best solution. If any agent can find a better target, that agent will update the coefficient vector in each iteration. and t represents the current iteration number, where the vector and for:

[0116]

[0117]

[0118] in, During the iteration process, the range decreases linearly within [0,2]. For random vectors, The range is [0,1].

[0119] During the hunting phase, whales can locate their prey and swim towards it along a spiral path. This hunting mechanism prevents the prey from escaping and can be modeled as follows:

[0120]

[0121]

[0122] in, Let be the distance between the i-th whale and the currently obtained optimal solution, b be a constant of logarithmic spiral shape, and l be a random number in the range [-1, 1].

[0123]

[0124] The equation above assigns an equal probability of selection to both behaviors. The whale optimization algorithm sets the probability of selection when... At that time, the whale attacks its prey and finally obtains the optimal solution.

[0125] Example 2

[0126] Two aspects of the present invention, with reference to Figure 3 This invention provides a distributed computing collaborative resource allocation system, comprising:

[0127] Module 1 is established to build a distributed computing system model for edge nodes;

[0128] Module 2 is used to construct an overall planning problem based on the edge node distributed computing system model, with the goal of minimizing the average latency of all users in the distributed computing network.

[0129] Allocation module 3 is used to solve the overall planning problem, determine the strategy for unloading dependent tasks, and allocate distributed computing collaborative resources according to the strategy for unloading dependent tasks.

[0130] As a preferred technical solution in this embodiment, the allocation module 3 includes:

[0131] Transformation module 31 is used to transform the overall planning problem into a penalty form using a penalty method;

[0132] Solver module 32 is used to solve the overall planning problem using the whale optimization algorithm to obtain the strategy for unloading dependency tasks;

[0133] The control module 33 is used to allocate distributed computing collaborative resources according to the unloading dependency task strategy.

[0134] Example 3

[0135] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a distributed computing collaborative resource allocation method. The memory may include main memory, such as high-speed random access memory (RAM), and may also include non-volatile memory, such as at least one disk storage device. The processor, network interface, and memory are interconnected via an internal bus, which may be an industry-standard architecture bus, a peripheral component interconnection standard bus, an extended industry-standard architecture bus, etc. The bus may be classified as an address bus, a data bus, a control bus, etc. The memory stores the program; specifically, the program may include program code, which includes computer operation instructions. The memory may include main memory and non-volatile memory, and provides instructions and data to the processor.

[0136] Example 4

[0137] A computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the distributed computing cooperative resource allocation method. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include read-only memory (ROM), hard disk, flash memory, optical disk, magnetic disk, etc.

[0138] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0139] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1A device that provides the functions specified in one or more boxes.

[0140] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0141] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0142] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A distributed computing collaborative resource allocation method, characterized in that, include: An edge node distributed computing system model is established, which includes a task model, a communication model, and an energy model. Based on the edge node distributed computing system model, an overall planning problem is constructed with the goal of minimizing the average latency of all users in the distributed computing network. Solve the overall planning problem to determine the strategy for unloading dependent tasks, and allocate distributed computing collaborative resources according to the strategy for unloading dependent tasks; The overall planning problem is: in, This is the difference between the time it takes for all subtasks to complete and the task creation time. For the number of users, Represents a set of users. For edge nodes Total energy, This represents the maximum energy of the edge node. UAV collection Indicates task subtasks Unloaded to edge nodes The decision to uninstall Indicates the task subtasks from Move to other edge nodes Decision variables, Indicates task The set of dependent subtasks, where x represents the time when all subtasks are completed and y represents the task creation time; The difference between the time it takes for all subtasks in task m to complete and the task creation time. for: in, Indicates task Neutron mission Final completion time, Indicates task Neutron mission The start time, For the task The total number of neutron missions; The edge node Total energy Represented as: in, Edge nodes are used for computation tasks. Neutron mission energy, Represents edge nodes Part of the task is transmitted to other edge nodes via wireless link. The energy transmitted during this time; The edge nodes are used for computation tasks. Neutron mission energy for: in, To effectively convert capacitors, This indicates that edge nodes are in the computation subtask. Energy consumption per CPU cycle For the task Neutron mission Data input size, Represents edge nodes The number of CPU cycles required to execute a single bit of data; The edge node Part of the task is transmitted to other edge nodes via wireless link. Energy generated during transmission for: in, For the task Neutron mission Data input size, Represents edge nodes With edge nodes Data transfer rate between Represents edge nodes The transmission power.

2. The distributed computing collaborative resource allocation method according to claim 1, characterized in that, The process of solving the overall planning problem, determining the strategy for unloading dependent tasks, and allocating distributed computing collaborative resources according to the strategy for unloading dependent tasks is as follows: The overall planning problem is transformed into a penalty-based approach. The whale optimization algorithm is used to solve the overall planning problem, and a strategy for unloading dependent tasks is obtained. Distributed computing collaboration resources are allocated according to the aforementioned unloading dependency task strategy.

3. A distributed computing collaborative resource allocation system, characterized in that, include: Establishment module (1) is used to establish an edge node distributed computing system model, which includes a task model, a communication model and an energy model; The construction module (2) is used to construct an overall planning problem based on the edge node distributed computing system model, with the goal of minimizing the average latency of all users in the distributed computing network; The allocation module (3) is used to solve the overall planning problem, determine the strategy for unloading dependent tasks, and allocate distributed computing collaborative resources according to the strategy for unloading dependent tasks. The overall planning problem is: in, This is the difference between the time it takes for all subtasks to complete and the task creation time. For the number of users, Represents a set of users. For edge nodes Total energy, This represents the maximum energy of the edge node. UAV collection Indicates task subtasks Unloaded to edge nodes The decision to uninstall Indicates the task subtasks from Move to other edge nodes Decision variables, Indicates task The set of dependent subtasks, where x represents the time when all subtasks are completed and y represents the task creation time; The difference between the time it takes for all subtasks in task m to complete and the task creation time. for: in, Indicates task Neutron mission Final completion time, Indicates task Neutron mission The start time, For the task The total number of neutron missions; The edge node Total energy Represented as: in, Edge nodes are used for computation tasks. Neutron mission energy, Represents edge nodes Part of the task is transmitted to other edge nodes via wireless link. The energy transmitted during this time; The edge nodes are used for computation tasks. Neutron mission energy for: in, To effectively convert capacitors, This indicates that edge nodes are in the computation subtask. Energy consumption per CPU cycle For the task Neutron mission Data input size, Represents edge nodes The number of CPU cycles required to execute a single bit of data; The edge node Part of the task is transmitted to other edge nodes via wireless link. Energy generated during transmission for: in, For the task Neutron mission Data input size, Represents edge nodes With edge nodes Data transfer rate between Represents edge nodes The transmission power.

4. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the distributed computing collaborative resource allocation method as described in any one of claims 1-2.

5. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the distributed computing collaborative resource allocation method as described in any one of claims 1-2.

Images