Cloud edge collaborative network-oriented computing unloading and resource joint optimization method and device

Abstract

The invention discloses a cloud edge collaborative network-oriented computing unloading and resource joint optimization method and device, and relates to the technical field of cloud edge collaborative resource optimization. The method comprises the following steps: constructing a minimum total energy consumption model by taking minimum total energy consumption as an optimization target; performing dynamic iterative optimization on the link grouping to obtain an optimal link grouping coefficient; carrying out distributed convex iteration optimization on the transmitting power; carrying out linear programming segmentation optimization on the current residual task load; performing convex optimization segmentation on the network bandwidth; and on the basis of a minimum total energy consumption model, according to the optimal link grouping coefficient, the forward link optimal transmitting power, the return link optimal transmitting power, the first optimal allocation task, the second optimal allocation task and the forward link bandwidth optimal allocation coefficient, carrying out convergence verification to obtain a current optimal resource allocation scheme. The computing unloading and resource joint optimization method is based on the cloud edge collaborative network and is high in efficiency and adaptability.

Description

Technical Field

[0001] This invention relates to the field of cloud-edge collaborative resource optimization technology, and in particular to a method and apparatus for computational offloading and joint resource optimization for cloud-edge collaborative networks. Background Technology

[0002] With the rapid development of mobile communication technology, the problem of insufficient infrastructure coverage in remote areas and emergency communication scenarios has become increasingly prominent. Traditional satellite-to-ground networks are unable to meet users' demands for high performance, high reliability, and low latency in computing tasks. Mobile edge computing (MEC) effectively alleviates the problem of limited local resources in mobile user devices (MUDs) by pushing computing resources to the network edge, but it still has shortcomings in resource pooling and security. Mobile cloud computing (MCC), while possessing powerful centralized resource processing capabilities, struggles to meet low latency requirements due to its long transmission distances. In recent years, unmanned aerial vehicles (UAVs) have become important platforms for auxiliary communication and computing due to their high mobility and flexible deployment, while low Earth orbit (LEO) satellites provide reliable network access services to ground users with their wide-area coverage capabilities. With the widespread application of cloud-edge computing networks that combine UAVs and LEO satellites in remote areas and emergency communications, limited terminal computing power, low efficiency of multi-node resource collaboration, and high system energy consumption have become core technical problems that urgently need to be solved.

[0003] Existing computation offloading methods primarily rely on fixed ground infrastructure as edge service nodes, alleviating terminal resource pressure by offloading all or part of the MUD's computational tasks to edge servers or cloud servers. However, these methods have significant shortcomings in practical applications: The node collaboration mechanism is simplistic, only considering a two-layer offloading architecture between the terminal and the edge or between the terminal and the cloud, failing to effectively integrate the collaborative computing capabilities among ground users, UAV edge nodes, and LEO satellite cloud, thus hindering the full utilization of the cloud-edge collaborative network architecture advantages; resource allocation strategies are relatively simple, typically employing static configuration or allocation schemes based on local information, unable to adapt to dynamic environmental changes caused by link time-varying characteristics, resulting in low system resource utilization efficiency; and there is a lack of effective global optimization mechanisms, with existing methods often optimizing for single performance indicators, failing to coordinate multi-dimensional resources such as computing and communication at the system level, leading to overall system performance falling short of optimal levels.

[0004] In existing technologies, there is a lack of a highly efficient and adaptable method for joint optimization of computational offloading and resources based on cloud-edge collaborative networks. Summary of the Invention

[0005] To address the technical problems of existing technologies, such as a single node collaboration mechanism, relatively simple resource allocation strategies, and a lack of effective global optimization mechanisms, this invention provides a method and apparatus for joint optimization of computation offloading and resources in cloud-edge collaborative networks. The technical solution is as follows:

[0006] On the one hand, a method for joint optimization of computation offloading and resources in cloud-edge collaborative networks is provided. This method is implemented by a device for joint optimization of computation offloading and resources, and includes:

[0007] Based on the cloud-edge collaborative network, the system obtains the first computing task information of the local mobile device, the second computing task information of the drone edge server, the third computing task information of the low-orbit satellite cloud server, the current remaining task volume, and the network bandwidth.

[0008] Based on the quasi-static model, under the resource-latency feasible region constraint, with the optimization objective of minimizing total energy consumption, a minimum total energy consumption model is constructed based on the first computing task information, the second task computing information, and the third task computing information.

[0009] Based on the minimum total energy consumption model, the link grouping is dynamically iteratively optimized according to the preset initial resource allocation scheme to obtain the optimal link grouping coefficient;

[0010] Based on the minimum total energy consumption model, the transmit power is optimized by distributed convex iteration according to the optimal link grouping coefficient and the preset initial resource allocation scheme to obtain the optimal transmit power of the fronthaul link and the optimal transmit power of the backhaul link.

[0011] Based on the minimum total energy consumption model, and according to the calculation information of the second task, the calculation information of the third task, the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, and the preset initial resource allocation scheme, the current remaining task quantity is optimized by linear programming to obtain the first optimal allocation task and the second optimal allocation task.

[0012] Based on the minimum total energy consumption model, the network bandwidth is convexly optimized and partitioned according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task and the second optimal allocation task to obtain the optimal allocation coefficient of the fronthaul link bandwidth.

[0013] Based on the minimum total energy consumption model, convergence verification is performed according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task, the second optimal allocation task, and the optimal allocation coefficient of the fronthaul link bandwidth to obtain the current optimal resource allocation scheme.

[0014] On the other hand, a compute offloading and resource joint optimization device for cloud-edge collaborative networks is provided. This device is applied to a compute offloading and resource joint optimization method for cloud-edge collaborative networks. The device includes:

[0015] The network information acquisition module is used to acquire, based on the cloud-edge collaborative network, the first computing task information of the local mobile device, the second computing task information of the UAV edge server, the third computing task information of the low-orbit satellite cloud server, the current remaining task volume, and the network bandwidth.

[0016] The energy consumption model construction module is used to construct a minimum total energy consumption model based on a quasi-static model, under the resource-latency feasible region constraint, with the optimization objective of minimizing total energy consumption, and based on the first computing task information, the second task computing information, and the third task computing information.

[0017] The grouping coefficient optimization module is used to dynamically iteratively optimize link grouping based on the minimum total energy consumption model and according to the preset initial resource allocation scheme to obtain the optimal link grouping coefficient.

[0018] The transmit power optimization module is used to perform distributed convex iterative optimization of transmit power based on the minimum total energy consumption model, the optimal link grouping coefficient, and the preset initial resource allocation scheme to obtain the optimal transmit power of the fronthaul link and the optimal transmit power of the backhaul link.

[0019] The task allocation optimization module is used to perform linear programming to partition and optimize the remaining task quantity based on the minimum total energy consumption model, according to the second task calculation information, the third task calculation information, the optimal link grouping coefficient, the optimal transmission power of the fronthaul link, the optimal transmission power of the backhaul link, and the preset initial resource allocation scheme, so as to obtain the first optimal allocation task and the second optimal allocation task.

[0020] The bandwidth allocation optimization module is used to perform convex optimization segmentation of network bandwidth based on the minimum total energy consumption model, according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task and the second optimal allocation task, to obtain the optimal allocation coefficient of the fronthaul link bandwidth.

[0021] The allocation scheme determination module is used to perform convergence verification based on the minimum total energy consumption model, according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task, the second optimal allocation task, and the optimal allocation coefficient of the fronthaul link bandwidth, to obtain the current optimal resource allocation scheme.

[0022] On the other hand, a computational offloading and resource co-optimization device is provided, the computational offloading and resource co-optimization device comprising: a processor; a memory storing computer-readable instructions, wherein when the computer-readable instructions are executed by the processor, any one of the above-described computational offloading and resource co-optimization methods for cloud-edge collaborative networks is implemented.

[0023] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored therein, the at least one instruction being loaded and executed by a processor to implement any of the above-described methods for computation offloading and joint resource optimization for cloud-edge collaborative networks.

[0024] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0025] This invention proposes a computation offloading and resource joint optimization method for cloud-edge collaborative networks. It provides computing services to terrestrial MUDs through collaboration between UAV edge nodes and LEO satellite cloud centers, overcoming the limitations of traditional two-layer "terminal-edge" or "terminal-cloud" architectures. By establishing a three-layer computation offloading model, it achieves dynamic allocation and parallel processing of computing tasks among the three computing entities. Addressing the specific needs of remote areas and emergency communication scenarios, the system can intelligently select local computing, edge computing, or cloud computing methods based on network status and task characteristics, providing reliable computing service support in complex environments.

[0026] By comprehensively considering multiple performance indicators such as system energy consumption, task latency, and resource utilization, a multi-objective collaborative optimization is achieved by establishing a joint optimization objective function and constraints. Compared with traditional methods with a single optimization objective, the feedback iterative optimization algorithm, through alternating solutions to sub-problems of link grouping, power allocation, task orchestration, and bandwidth management, significantly reduces the total system energy consumption while ensuring the timeliness of task execution. Based on the multi-dimensional optimization model, the system's operating state can be more systematically and accurately represented, thereby achieving more intelligent resource allocation and task scheduling, providing a new optimization paradigm for computation offloading in cloud-edge collaborative networks.

[0027] This invention decomposes a complex global optimization problem into multiple subproblems that can be solved in parallel. Each node, through local information exchange and iterative updates, gradually approaches the global optimum. This not only reduces system communication overhead but also significantly improves the algorithm's adaptability and convergence speed in dynamic environments. This mechanism effectively enhances the system's stability and operational efficiency under conditions of high node mobility and fluctuating network states, providing a reliable solution for computational offloading and resource allocation in cloud-edge collaborative networks. This invention is a highly efficient and adaptable method for joint computational offloading and resource optimization based on cloud-edge collaborative networks. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a flowchart of a computation offloading and resource joint optimization method for cloud-edge collaborative networks provided by an embodiment of the present invention;

[0030] Figure 2 This is a block diagram of a computation offloading and resource joint optimization device for cloud-edge collaborative networks provided in an embodiment of the present invention;

[0031] Figure 3 This is a schematic diagram of the structure of a computational offloading and resource joint optimization device provided in an embodiment of the present invention. Detailed Implementation

[0032] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0033] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0034] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0035] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0036] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0037] This invention provides a method for joint optimization of computation offloading and resources in cloud-edge collaborative networks. This method can be implemented by a device for joint optimization of computation offloading and resources, which can be a terminal or a server. Figure 1 The flowchart shown is for a computation offloading and resource joint optimization method for cloud-edge collaborative networks. The processing flow of this method may include the following steps:

[0038] S1. Based on the cloud-edge collaborative network, obtain the first computing task information of the local mobile device, the second computing task information of the drone edge server, the third computing task information of the low-orbit satellite cloud server, the current remaining task volume and network bandwidth.

[0039] The first computing task information includes the local task volume, local computing power, local maximum transmit power, local computing load, and local maximum tolerable transmission latency of the local mobile device.

[0040] The second task calculation information includes the edge computing capability of the UAV edge server, the maximum transmit power of the fronthaul link, the edge computing load, and the maximum tolerable latency of edge transmission.

[0041] The third task's computational information includes the low-orbit satellite server's cloud computing capabilities, the maximum transmission power of the backhaul link, the cloud computing load, and the maximum tolerable latency of cloud transmission.

[0042] In one feasible implementation, this invention constructs a cloud-edge collaborative network architecture assisted by UAVs and LEO satellites. The MUDs are randomly distributed within the UAV coverage area. Network parameters are initialized, and tasks are processed collaboratively through local computing, UAV edge computing, and satellite cloud computing. Considering constraints such as latency, the system establishes a corresponding optimization model with the objective of minimizing total energy consumption.

[0043] A cloud-edge collaborative network architecture assisted by UAVs and LEO satellites was established, in which... A number of MUDs with local computing capabilities and located in areas without infrastructure are randomly distributed within the coverage area of ​​UAVs, with their local computing capabilities set to [value missing]. Maximum transmission power Collect computational task information for each MUD, including task data packet size. (Unit: bits), computational load (Unit: CPU cycles / bit) and maximum tolerable latency (Unit: seconds). The UAV is equipped with an edge server, configured with a NOMA signal processing module and a SiC decoding module, and the computing power allocated to the MUD by the edge server is set to... (Unit: CPU cycles / second), Maximum transmit power The LEO satellite cloud center is equipped with cloud servers deployed at... Altitude orbit ensures the satellite maintains line-of-sight communication with all UAVs; OMA resource allocation module is configured; cloud server computing power is allocated to the MUD. UAV edge node task assignment Edge computing is offloaded via the fronthaul link, and the link grouping coefficients are initialized. (This means that any MUD is served by a single UAV) and the fronthaul link bandwidth allocation coefficient. LEO Satellite Cloud Center assigns tasks Cloud computing offloading is completed via the backhaul link; backhaul link bandwidth allocation coefficient is set. Network bandwidth It is fully allocated to the fronthaul and backhaul communication links.

[0044] S2. Based on the quasi-static model, under the resource-latency feasible region constraint, with the optimization objective of minimizing total energy consumption, a minimum total energy consumption model is constructed based on the first computing task information, the second task computing information, and the third task computing information.

[0045] Among them, the resource-latency feasible domain constraints include task completion latency constraints, maximum drone load constraints, and cloud server computing resource constraints.

[0046] In one feasible implementation, a quasi-static model is used, where the network topology of the UAV-assisted LEO and the MUD remain relatively static during the execution of the computation task, and only deterministic tasks in the current task offload request queue are considered. The minimum total energy consumption model corresponding to the optimization problem is constructed as follows (1): (1);

[0047] in, For the variable parameters to be optimized, Indicates the transmit power of the MUD. Indicates the UAV's transmission power. , , These represent local computing energy consumption, UAV edge computing energy consumption, and satellite cloud center computing energy consumption, respectively. Under the constraints of maximum tolerable latency, transmission power, transmission rate, link grouping coefficient, bandwidth allocation coefficient, and the processing capacity of computing nodes, the objective is to minimize the total system energy consumption.

[0048] S3. Based on the minimum total energy consumption model, the link grouping is dynamically iteratively optimized according to the preset initial resource allocation scheme to obtain the optimal link grouping coefficient.

[0049] Optionally, based on the minimum total energy consumption model and according to a preset initial resource allocation scheme, the link grouping is dynamically iteratively optimized to obtain the optimal link grouping coefficients, including:

[0050] Based on the minimum total energy consumption model, according to the preset initial resource allocation scheme, the Lagrange dual method is used to relax the link correlation variables and construct a decomposable dual problem.

[0051] Based on the dual problem, the subgradient method is used to solve each subproblem in a distributed iterative manner, and the link correlation variables and Lagrange multipliers are updated.

[0052] The optimal link grouping coefficients are obtained by projecting the simplex feasible region onto the updated link association variables and Lagrange multipliers.

[0053] One feasible implementation aims to minimize total system energy consumption and addresses the issue of ground-based MUDs dynamically selecting the optimal UAV edge server for task offloading, due to binary variables. The existence of coupling constraints necessitates relaxing Boolean variables into continuous variables. Furthermore, the hard constraints are relaxed to the objective function using the Lagrange duality method, eliminating the coupling of variables. The Lagrange function is constructed as follows (2):

[0054] (2);

[0055] in, Calculate the energy consumption for the objective in the link grouping problem. To address one of the constraints considered when grouping links, These are Lagrange multipliers associated with different constraints;

[0056] By introducing Lagrange multipliers, a dual function with convex properties is constructed. The lower bound of the original problem can be obtained directly by solving it through the Convex Optimization Toolbox (CVX: Matlab Software for Disciplined Convex Programming, CVX). The dual function is defined as follows (3):

[0057] (3);

[0058] in, It is the dual function. This indicates the lower bound of the Lagrange function calculated within the range of values ​​for the link grouping coefficients.

[0059] Therefore, the original problem is transformed into a Lagrange dual problem, that is... .

[0060] According to the Lagrange weak duality, the solution to the dual problem approximates the lower bound of the primal problem by maximizing the dual function. (The variables after relaxation...) Continuity transforms the optimization problem into a problem about The linear function is a convex function with at least one strictly feasible solution that satisfies the Slater condition and has a dual gap of 0. That is, the feasible solution of the dual problem is strictly consistent with the optimal solution of the original problem. Since the dual function is not differentiable, the subgradient method is used to update the Lagrange multipliers. The process is as follows: (4)

[0061] (4);

[0062] in, It's about updating the step size. It is an iterative index. Equivalent to Each iteration first calculates the subgradient of the dual function, then updates the Lagrange multipliers based on the direction and step size of the subgradient. As the iterations continue, it eventually converges to the global optimum or a solution close to the optimum.

[0063] S4. Based on the minimum total energy consumption model, according to the optimal link grouping coefficient and the preset initial resource allocation scheme, the transmit power is optimized by distributed convex iteration to obtain the optimal transmit power of the fronthaul link and the optimal transmit power of the backhaul link.

[0064] Optionally, based on the minimum total energy consumption model, and according to the optimal link grouping coefficient and the preset initial resource allocation scheme, the transmit power is optimized by distributed convex iteration to obtain the optimal transmit power for the fronthaul link and the optimal transmit power for the backhaul link, including:

[0065] Based on the minimum total energy consumption model, according to the optimal link grouping coefficient and the preset initial resource allocation scheme, the quadratic transformation method is used to decouple variables and construct a separable quadratic function.

[0066] Based on quadratic functions, DC programming is used to transform the non-convex optimization problem into a convex form, and a first-order Taylor expansion is used to linearly approximate the negative convex term to obtain a convex optimization approximation problem.

[0067] Based on the convex optimization approximation problem, the CVX toolbox is used to perform a distributed iterative solution for the link transmit power. When the iteration converges, the optimal transmit power of the fronthaul link and the optimal transmit power of the backhaul link are obtained.

[0068] In one feasible implementation, to address the wireless resource allocation problem, the transmit power of the fronthaul and backhaul links is adjusted. The allocation process considers the coupling relationships between variables, which leads to non-convexity and thus increases the computational complexity of the problem. A quadratic transformation is used to convert the coupled variables in the optimization problem into a DC programming problem (i.e., the difference between two convex functions). DC programming problems are typically non-convex; the linear approximation obtained through a first-order Taylor expansion is an affine function, allowing the optimization process to progressively approximate the optimal solution of the original problem. Since the optimization problem is transformed into a convex optimization problem through the quadratic transformation and DC programming, it can be directly solved using the CVX Convex Optimization Toolbox.

[0069] S5. Based on the minimum total energy consumption model, according to the calculation information of the second task, the calculation information of the third task, the optimal link grouping coefficient, the optimal transmission power of the fronthaul link, the optimal transmission power of the backhaul link, and the preset initial resource allocation scheme, the current remaining task quantity is optimized by linear programming to obtain the first optimal allocation task and the second optimal allocation task.

[0070] Optionally, based on the minimum total energy consumption model, and according to the calculation information of the second task, the calculation information of the third task, the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, and the preset initial resource allocation scheme, the remaining task quantity is optimized by linear programming to obtain the first optimal allocation task and the second optimal allocation task, including:

[0071] Calculations are performed based on the second task calculation information and the third task calculation information to obtain the upper and lower bounds of the first and second task allocations.

[0072] Based on the minimum total energy consumption model, the upper and lower bounds of the first and second allocation tasks, linear constraint modeling is performed according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, and the preset initial resource allocation scheme to obtain the linear programming problem.

[0073] Based on the linear programming problem, and according to the remaining task volume, the CVX toolkit is used for distributed solution to obtain the first optimal task allocation and the second optimal task allocation.

[0074] In one feasible implementation, the task orchestration problem can be transformed into The linear optimization problem is addressed by calculating strict upper and lower bounds on the amount of task that can be allocated in both the UAV edge nodes and the LEO satellite cloud center, based on the maximum tolerable latency of the task and the processing capacity of each computing node. Considering the scenario where the MDU can complete task computation locally without needing to connect to any UAV for task offloading, the following results are obtained. The upper bound of is given by the mathematical expression (5):

[0075] (5);

[0076] in, It is the upper bound of the buffer calculated based on the maximum tolerable delay. This refers to the size of the task data packet.

[0077] Similarly, we can obtain The upper and lower bounds are as follows: (6) and (7):

[0078] (6);

[0079] (7);

[0080] in, and The upper and lower bounds of the buffer are obtained based on the maximum tolerable latency and the processing capacity of each computing node.

[0081] Since both the objective function and the constraints are linear, this problem is a linear programming problem and can be solved using the standard convex optimization tool CVX.

[0082] S6. Based on the minimum total energy consumption model, according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task and the second optimal allocation task, the network bandwidth is convexly optimized and segmented to obtain the optimal allocation coefficient of the fronthaul link bandwidth.

[0083] In one feasible implementation, a bandwidth management algorithm is used to rationally allocate system bandwidth to meet the rate requirements of the fronthaul and backhaul links. The original non-convex structure is transformed into a convex optimization problem by utilizing the constraint property of the bandwidth allocation coefficient, which can then be solved directly using convex optimization tools. To ensure the numerical feasibility of the constraints, strict inequalities are treated as closed intervals with tolerance, allowing the problem to be stably handled during optimization.

[0084] To address the wireless resource optimization problem, further rational allocation of system bandwidth is implemented. Under limited spectrum resources, the bandwidth ratio between the fronthaul and backhaul links is managed to ensure that the transmission rate requirements of both are effectively met. This is because both the optimization objective and constraints include bandwidth allocation coefficients. The fractional terms are usually concave. Based on the bandwidth allocation coefficient constraint, this term is transformed into a convex function within the feasible region. Therefore, the objective function and constraints are... The entire feasible region satisfies convexity. The optimization objective is a convex optimization problem, which can be solved directly using the CVX toolbox.

[0085] Bandwidth allocation coefficient constraints are generally strict inequalities, which are often difficult to handle directly in numerical optimization. Optimization solvers typically require constraints on closed sets; therefore, tolerance is introduced. (Right now (where a value is a very small positive number) transforms the open interval problem into a closed interval constraint, ensuring the numerical feasibility of the forward and backhaul links.

[0086] S7. Based on the minimum total energy consumption model, convergence verification is performed according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task, the second optimal allocation task, and the optimal allocation coefficient of the fronthaul link bandwidth to obtain the current optimal resource allocation scheme.

[0087] Optionally, based on the minimum total energy consumption model, convergence verification is performed according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task, the second optimal allocation task, and the optimal allocation coefficient of the fronthaul link bandwidth to obtain the current optimal resource allocation scheme, including:

[0088] Based on the minimum total energy consumption model, the current minimum total energy consumption is calculated according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task, the second optimal allocation task, and the optimal allocation coefficient of the fronthaul link bandwidth.

[0089] Based on a preset energy consumption threshold, convergence verification is performed according to the preset historical minimum total energy consumption and the current minimum total energy consumption to obtain the verification result;

[0090] When the verification result is converged, the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task, the second optimal allocation task, and the optimal allocation coefficient of the fronthaul link bandwidth are determined as the current optimal resource allocation scheme.

[0091] In one feasible implementation, a distributed collaborative optimization method based on feedback iteration alternately optimizes four sub-problems: link grouping, power allocation, task orchestration, and bandwidth management, gradually approaching the global optimum and ultimately minimizing the total system energy consumption. In each iteration, the link grouping algorithm uses a fixed transmit power... Uninstallation task and bandwidth allocation coefficient The link grouping coefficients are updated. The power allocation algorithm is based on the updated link grouping coefficients. The power allocation optimization solution is obtained by combining the remaining fixed parameters. Similarly, the optimization solutions for offloading tasks and bandwidth allocation are obtained. After each iteration, the total energy consumption of the current system is calculated and compared with the result of the previous iteration. If the convergence threshold is met... If the number of iterations reaches the upper limit, the algorithm stops iterating and outputs an approximate optimal solution; otherwise, the current result is used as the initial value for the next iteration to continue solving the problem, repeating the previous steps.

[0092] Based on the aforementioned joint optimization strategy, the MUD sends mission data proportionally to designated UAV edge nodes and LEO satellite cloud centers. The UAV and LEO satellites respectively perform edge computing and cloud computing tasks, returning the results to the ground terminal. After completing their tasks, they release the occupied communication and computing resources, preparing to receive the next round of mission requests.

[0093] This invention addresses the application needs of limited computing resources and reduced energy consumption in remote areas and emergency communication scenarios. It proposes a method and apparatus for joint optimization of computational offloading and resources in cloud-edge collaborative networks. By decomposing the energy minimization problem into four sub-problems—link grouping, power allocation, task scheduling, and bandwidth management—a distributed collaborative optimization algorithm based on feedback iteration is proposed to solve these problems, enabling dynamic allocation of resources across multiple nodes. This invention provides technical support for efficient resource utilization in cloud-edge collaborative networks, significantly improving the system's energy efficiency and adaptability in dynamic business scenarios.

[0094] This invention proposes a computation offloading and resource joint optimization method for cloud-edge collaborative networks. It provides computing services to terrestrial MUDs through collaboration between UAV edge nodes and LEO satellite cloud centers, overcoming the limitations of traditional two-layer "terminal-edge" or "terminal-cloud" architectures. By establishing a three-layer computation offloading model, it achieves dynamic allocation and parallel processing of computing tasks among the three computing entities. Addressing the specific needs of remote areas and emergency communication scenarios, the system can intelligently select local computing, edge computing, or cloud computing methods based on network status and task characteristics, providing reliable computing service support in complex environments.

[0095] By comprehensively considering multiple performance indicators such as system energy consumption, task latency, and resource utilization, a multi-objective collaborative optimization is achieved by establishing a joint optimization objective function and constraints. Compared with traditional methods with a single optimization objective, the feedback iterative optimization algorithm, through alternating solutions to sub-problems of link grouping, power allocation, task orchestration, and bandwidth management, significantly reduces the total system energy consumption while ensuring the timeliness of task execution. Based on the multi-dimensional optimization model, the system's operating state can be more systematically and accurately represented, thereby achieving more intelligent resource allocation and task scheduling, providing a new optimization paradigm for computation offloading in cloud-edge collaborative networks.

[0096] This invention decomposes a complex global optimization problem into multiple subproblems that can be solved in parallel. Each node, through local information exchange and iterative updates, gradually approaches the global optimum. This not only reduces system communication overhead but also significantly improves the algorithm's adaptability and convergence speed in dynamic environments. This mechanism effectively enhances the system's stability and operational efficiency under conditions of high node mobility and fluctuating network states, providing a reliable solution for computational offloading and resource allocation in cloud-edge collaborative networks. This invention is a highly efficient and adaptable method for joint computational offloading and resource optimization based on cloud-edge collaborative networks.

[0097] Figure 2 This is a block diagram of a computation offloading and resource joint optimization device for cloud-edge collaborative networks provided by an embodiment of the present invention. This device is used for a computation offloading and resource joint optimization method for cloud-edge collaborative networks. (Refer to...) Figure 2 The device includes a network information acquisition module 210, an energy consumption model construction module 220, a grouping coefficient optimization module 230, a transmission power optimization module 240, a task allocation optimization module 250, a bandwidth allocation optimization module 260, and an allocation scheme determination module 270. Among them:

[0098] The network information acquisition module 210 is used to acquire, based on the cloud-edge collaborative network, the first computing task information of the local mobile device, the second computing task information of the UAV edge server, the third computing task information of the low-orbit satellite cloud server, the current remaining task volume, and the network bandwidth.

[0099] The energy consumption model construction module 220 is used to construct a minimum total energy consumption model based on a quasi-static model, under the resource-latency feasible region constraint, with the optimization objective of minimizing total energy consumption, and based on the first computing task information, the second task computing information, and the third task computing information.

[0100] The grouping coefficient optimization module 230 is used to dynamically iteratively optimize the link grouping based on the minimum total energy consumption model and according to the preset initial resource allocation scheme to obtain the optimal link grouping coefficient.

[0101] The transmit power optimization module 240 is used to perform distributed convex iterative optimization of transmit power based on the minimum total energy consumption model, according to the optimal link grouping coefficient and the preset initial resource allocation scheme, to obtain the optimal transmit power of the fronthaul link and the optimal transmit power of the backhaul link.

[0102] The task allocation optimization module 250 is used to perform linear programming partitioning optimization on the current remaining task quantity based on the minimum total energy consumption model, according to the second task calculation information, the third task calculation information, the optimal link grouping coefficient, the optimal transmission power of the fronthaul link, the optimal transmission power of the backhaul link, and the preset initial resource allocation scheme, to obtain the first optimal allocation task and the second optimal allocation task.

[0103] The bandwidth allocation optimization module 260 is used to perform convex optimization segmentation of network bandwidth based on the minimum total energy consumption model, according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task and the second optimal allocation task, to obtain the optimal allocation coefficient of the fronthaul link bandwidth.

[0104] The allocation scheme determination module 270 is used to perform convergence verification based on the minimum total energy consumption model, according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task, the second optimal allocation task, and the optimal allocation coefficient of the fronthaul link bandwidth, to obtain the current optimal resource allocation scheme.

[0105] The first computing task information includes the local task volume, local computing power, local maximum transmit power, local computing load, and local maximum tolerable transmission latency of the local mobile device.

[0106] The second task calculation information includes the edge computing capability of the UAV edge server, the maximum transmit power of the fronthaul link, the edge computing load, and the maximum tolerable latency of edge transmission.

[0107] The third task's computational information includes the low-orbit satellite server's cloud computing capabilities, the maximum transmission power of the backhaul link, the cloud computing load, and the maximum tolerable latency of cloud transmission.

[0108] Among them, the resource-latency feasible domain constraints include task completion latency constraints, maximum drone load constraints, and cloud server computing resource constraints.

[0109] Optionally, the grouping coefficient optimization module 230 is further used for:

[0110] Based on the minimum total energy consumption model, according to the preset initial resource allocation scheme, the Lagrange dual method is used to relax the link correlation variables and construct a decomposable dual problem.

[0111] Based on the dual problem, the subgradient method is used to solve each subproblem in a distributed iterative manner, and the link correlation variables and Lagrange multipliers are updated.

[0112] The optimal link grouping coefficients are obtained by projecting the simplex feasible region onto the updated link association variables and Lagrange multipliers.

[0113] Optionally, the transmit power optimization module 240 is further used for:

[0114] Based on the minimum total energy consumption model, according to the optimal link grouping coefficient and the preset initial resource allocation scheme, the quadratic transformation method is used to decouple variables and construct a separable quadratic function.

[0115] Based on quadratic functions, DC programming is used to transform the non-convex optimization problem into a convex form, and a first-order Taylor expansion is used to linearly approximate the negative convex term to obtain a convex optimization approximation problem.

[0116] Based on the convex optimization approximation problem, the CVX toolbox is used to perform a distributed iterative solution for the link transmit power. When the iteration converges, the optimal transmit power of the fronthaul link and the optimal transmit power of the backhaul link are obtained.

[0117] Optionally, the task allocation optimization module 250 is further used for:

[0118] Calculations are performed based on the second task calculation information and the third task calculation information to obtain the upper and lower bounds of the first and second task allocations.

[0119] Based on the minimum total energy consumption model, the upper and lower bounds of the first and second allocation tasks, linear constraint modeling is performed according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, and the preset initial resource allocation scheme to obtain the linear programming problem.

[0120] Based on the linear programming problem, and according to the remaining task volume, the CVX toolkit is used for distributed solution to obtain the first optimal task allocation and the second optimal task allocation.

[0121] Optionally, the allocation scheme determination module 270 is further used for:

[0122] Based on the minimum total energy consumption model, the current minimum total energy consumption is calculated according to the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task, the second optimal allocation task, and the optimal allocation coefficient of the fronthaul link bandwidth.

[0123] Based on a preset energy consumption threshold, convergence verification is performed according to the preset historical minimum total energy consumption and the current minimum total energy consumption to obtain the verification result;

[0124] When the verification result is converged, the optimal link grouping coefficient, the optimal transmit power of the fronthaul link, the optimal transmit power of the backhaul link, the first optimal allocation task, the second optimal allocation task, and the optimal allocation coefficient of the fronthaul link bandwidth are determined as the current optimal resource allocation scheme.

[0125] This invention proposes a computation offloading and resource joint optimization method for cloud-edge collaborative networks. It provides computing services to terrestrial MUDs through collaboration between UAV edge nodes and LEO satellite cloud centers, overcoming the limitations of traditional two-layer "terminal-edge" or "terminal-cloud" architectures. By establishing a three-layer computation offloading model, it achieves dynamic allocation and parallel processing of computing tasks among the three computing entities. Addressing the specific needs of remote areas and emergency communication scenarios, the system can intelligently select local computing, edge computing, or cloud computing methods based on network status and task characteristics, providing reliable computing service support in complex environments.

[0126] By comprehensively considering multiple performance indicators such as system energy consumption, task latency, and resource utilization, a multi-objective collaborative optimization is achieved by establishing a joint optimization objective function and constraints. Compared with traditional methods with a single optimization objective, the feedback iterative optimization algorithm, through alternating solutions to sub-problems of link grouping, power allocation, task orchestration, and bandwidth management, significantly reduces the total system energy consumption while ensuring the timeliness of task execution. Based on the multi-dimensional optimization model, the system's operating state can be more systematically and accurately represented, thereby achieving more intelligent resource allocation and task scheduling, providing a new optimization paradigm for computation offloading in cloud-edge collaborative networks.

[0127] This invention decomposes a complex global optimization problem into multiple subproblems that can be solved in parallel. Each node, through local information exchange and iterative updates, gradually approaches the global optimum. This not only reduces system communication overhead but also significantly improves the algorithm's adaptability and convergence speed in dynamic environments. This mechanism effectively enhances the system's stability and operational efficiency under conditions of high node mobility and fluctuating network states, providing a reliable solution for computational offloading and resource allocation in cloud-edge collaborative networks. This invention is a highly efficient and adaptable method for joint computational offloading and resource optimization based on cloud-edge collaborative networks.

[0128] Figure 3 This is a schematic diagram of the structure of a computational offloading and resource joint optimization device provided in an embodiment of the present invention, as shown below. Figure 3 As shown, the computational offloading and resource joint optimization device may include the above-mentioned Figure 2 The illustrated compute offloading and resource joint optimization device for cloud-edge collaborative networks. Optionally, the compute offloading and resource joint optimization device 310 may include a first processor 2001.

[0129] Optionally, the compute offloading and resource co-optimization device 310 may also include a memory 2002 and a transceiver 2003.

[0130] The first processor 2001, memory 2002, and transceiver 2003 can be connected via a communication bus.

[0131] The following is combined with Figure 3 The components of the computational offloading and resource co-optimization device 310 are described in detail below:

[0132] The first processor 2001 is the control center of the computation offloading and resource joint optimization device 310. It can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 can be one or more central processing units (CPUs), application-specific integrated circuits (ASICs), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).

[0133] Optionally, the first processor 2001 can perform various functions of the computation offloading and resource co-optimization device 310 by running or executing software programs stored in the memory 2002 and calling data stored in the memory 2002.

[0134] In a specific implementation, as one example, the first processor 2001 may include one or more CPUs, for example... Figure 3 CPU0 and CPU1 are shown in the diagram.

[0135] In a specific implementation, as one example, the computational offloading and resource joint optimization device 310 may also include multiple processors, for example... Figure 3 The first processor 2001 and the second processor 2004 are shown in the diagram. Each of these processors can be a single-core processor or a multi-core processor. Here, a processor can refer to one or more devices, circuits, and / or processing cores used to process data (such as computer program instructions).

[0136] The memory 2002 is used to store the software program that executes the present invention, and is controlled by the first processor 2001 to execute it. The specific implementation method can be referred to the above method embodiment, and will not be repeated here.

[0137] Optionally, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or may exist independently, and may be connected via the interface circuit of the computational offloading and resource joint optimization device 310. Figure 3 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this.

[0138] The transceiver 2003 is used to communicate with network devices or with terminal devices.

[0139] Alternatively, transceiver 2003 may include a receiver and a transmitter. Figure 3 (Not shown separately). The receiver is used to implement the receiving function, and the transmitter is used to implement the transmitting function.

[0140] Optionally, the transceiver 2003 can be integrated with the first processor 2001 or exist independently, and its interface circuitry can be optimized by computational offloading and resource co-optimization of the device 310. Figure 3 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this.

[0141] It should be noted that, Figure 3 The structure of the compute offloading and resource co-optimization device 310 shown in the diagram does not constitute a limitation on the router. Actual compute offloading and resource co-optimization devices may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0142] Furthermore, the technical effects of the computation offloading and resource joint optimization device 310 can be referred to the technical effects of the computation offloading and resource joint optimization method for cloud-edge collaborative networks described in the above method embodiments, and will not be repeated here.

[0143] It should be understood that the first processor 2001 in the embodiments of the present invention may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor, or it may be any conventional processor, etc.

[0144] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0145] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0146] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0147] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0148] It should be understood that, in various embodiments of the present invention, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0149] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0150] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0151] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0152] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0153] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0154] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0155] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for computing offloading and resource joint optimization for cloud-edge collaborative networks, characterized in that, The method comprises: Based on the cloud edge collaborative network, the first computing task information of the local mobile device, the second task computing information of the unmanned aerial vehicle edge server, the third task computing information of the low-orbit satellite cloud server, the current residual task quantity and the network bandwidth are acquired; Based on the quasi-static model, under the constraint of the resource-delay feasible region, a minimum total energy consumption model is constructed according to the first computing task information, the second task computing information and the third task computing information, with the optimization target of minimizing the total energy consumption; Based on the minimum total energy consumption model, the optimal link grouping coefficient is obtained by dynamically and iteratively optimizing the link grouping according to the preset initial resource allocation scheme; Based on the minimum total energy consumption model, the optimal forward link transmission power and the optimal backhaul link transmission power are obtained by distributing and iteratively optimizing the transmission power according to the optimal link grouping coefficient and the preset initial resource allocation scheme; Based on the minimum total energy consumption model, the first optimal allocation task and the second optimal allocation task are obtained by linear programming and segmentation optimization of the current residual task quantity according to the second task computing information, the third task computing information, the optimal link grouping coefficient, the optimal forward link transmission power, the optimal backhaul link transmission power and the preset initial resource allocation scheme; Based on the minimum total energy consumption model, the optimal forward link bandwidth allocation coefficient is obtained by convex optimization segmentation of the network bandwidth according to the optimal link grouping coefficient, the optimal forward link transmission power, the optimal backhaul link transmission power, the first optimal allocation task and the second optimal allocation task; Based on the minimum total energy consumption model, the current optimal resource allocation scheme is obtained by convergence verification according to the optimal link grouping coefficient, the optimal forward link transmission power, the optimal backhaul link transmission power, the first optimal allocation task, the second optimal allocation task and the optimal forward link bandwidth allocation coefficient.

2. The method of claim 1, wherein, The first computing task information comprises the local task quantity, the local computing capability, the local maximum transmission power, the local computing load and the local transmission maximum tolerance delay of the local mobile device; The second task computing information comprises the edge computing capability, the forward link maximum transmission power, the edge computing load and the edge transmission maximum tolerance delay of the unmanned aerial vehicle edge server; The third task computing information comprises the cloud computing capability, the backhaul link maximum transmission power, the cloud computing load and the cloud transmission maximum tolerance delay of the low-orbit satellite server.

3. The method of claim 1, wherein, The resource-delay feasible region constraint comprises the task completion delay constraint, the unmanned aerial vehicle maximum load constraint and the cloud server computing resource constraint.

4. The method of claim 1, wherein, Based on the minimum total energy consumption model, the optimal link grouping coefficient is obtained by dynamically and iteratively optimizing the link grouping according to the preset initial resource allocation scheme, comprising: Based on the minimum total energy consumption model, the Lagrange dual method is used to relax the link associated variables and construct a decomposable dual problem according to the preset initial resource allocation scheme; Based on the dual problem, the sub-gradient method is used to distribute and iteratively solve each sub-problem, and the link associated variables and the Lagrange multiplier are updated; The optimal link grouping coefficient is obtained by simplex feasible region projection according to the updated link associated variables and the Lagrange multiplier.

5. The method of claim 1, wherein, The distributed convex iteration optimization of the transmit power is performed based on the minimum total energy consumption model according to the optimal link grouping coefficient and the preset initial resource allocation scheme, and the optimal transmit power of the front-haul link and the optimal transmit power of the back-haul link are obtained, including: Based on the minimum total energy consumption model, the variable decoupling is performed using the quadratic transformation method according to the optimal link grouping coefficient and the preset initial resource allocation scheme, and a separable quadratic function is constructed; Based on the quadratic function, the non-convex optimization problem is converted into a convex difference form by using DC programming, and the linear approximation of the negative convex term is performed through the first-order Taylor expansion, and a convex optimization approximation problem is obtained; Based on the convex optimization approximation problem, the CVX toolbox is used to perform distributed iteration solution on the link transmit power, and when the iteration converges, the optimal transmit power of the front-haul link and the optimal transmit power of the back-haul link are obtained.

6. The method of claim 1, wherein, The linear programming partition optimization of the current remaining task quantity is performed based on the minimum total energy consumption model according to the second task calculation information, the third task calculation information, the optimal link grouping coefficient, the optimal transmit power of the front-haul link, the optimal transmit power of the back-haul link and the preset initial resource allocation scheme, and the first optimal allocation task and the second optimal allocation task are obtained, including: The first allocation task upper and lower bounds and the second allocation task upper and lower bounds are obtained by calculating according to the second task calculation information and the third task calculation information; Based on the minimum total energy consumption model, the first allocation task upper and lower bounds and the second allocation task upper and lower bounds, the linear constraint modeling is performed according to the optimal link grouping coefficient, the optimal transmit power of the front-haul link, the optimal transmit power of the back-haul link and the preset initial resource allocation scheme, and a linear programming problem is obtained; Based on the linear programming problem, the first optimal allocation task and the second optimal allocation task are obtained by using the CVX toolbox for distributed solution according to the remaining task quantity.

7. The method of claim 1, wherein, The convergence verification is performed based on the minimum total energy consumption model according to the optimal link grouping coefficient, the optimal transmit power of the front-haul link, the optimal transmit power of the back-haul link, the first optimal allocation task, the second optimal allocation task and the front-haul link bandwidth optimal allocation coefficient, and the current optimal resource allocation scheme is obtained, including: Based on the minimum total energy consumption model, the current minimum total energy consumption is obtained by calculating according to the optimal link grouping coefficient, the optimal transmit power of the front-haul link, the optimal transmit power of the back-haul link, the first optimal allocation task, the second optimal allocation task and the front-haul link bandwidth optimal allocation coefficient; Based on the preset energy consumption threshold, the convergence verification is performed according to the preset historical minimum total energy consumption and the current minimum total energy consumption, and a verification result is obtained; When the verification result is convergence, the optimal link grouping coefficient, the optimal transmit power of the front-haul link, the optimal transmit power of the back-haul link, the first optimal allocation task, the second optimal allocation task and the front-haul link bandwidth optimal allocation coefficient are determined as the current optimal resource allocation scheme.

8. A device for computing offloading and resource joint optimization for cloud-edge collaborative network, the device for computing offloading and resource joint optimization for cloud-edge collaborative network is used to implement the method for computing offloading and resource joint optimization for cloud-edge collaborative network according to any one of claims 1-7, characterized in that, The device comprises: A network information acquisition module is configured to acquire, based on a cloud-edge collaborative network, first calculation task information of a local mobile device, second task calculation information of a UAV edge server, third task calculation information of a low-orbit satellite cloud server, a current remaining task quantity and a network bandwidth. The energy consumption model construction module is configured to construct a minimum total energy consumption model based on the quasi-static model, under the constraint of the resource-latency feasible region, and according to the first computing task information, the second computing task information and the third computing task information, with the optimization target of minimizing the total energy consumption. The grouping coefficient optimization module is configured to perform dynamic iterative optimization on the link grouping based on the minimum total energy consumption model and according to the preset initial resource allocation scheme, to obtain an optimal link grouping coefficient. The transmit power optimization module is configured to perform distributed convex iterative optimization on the transmit power based on the minimum total energy consumption model, the optimal link grouping coefficient and the preset initial resource allocation scheme, to obtain optimal transmit power of the front-haul link and optimal transmit power of the back-haul link. The task allocation optimization module is configured to perform linear programming segmentation optimization on the current remaining task amount based on the minimum total energy consumption model, the second computing task information, the third computing task information, the optimal link grouping coefficient, the optimal transmit power of the front-haul link, the optimal transmit power of the back-haul link and the preset initial resource allocation scheme, to obtain a first optimal allocated task and a second optimal allocated task. The bandwidth allocation optimization module is configured to perform convex optimization segmentation on the network bandwidth based on the minimum total energy consumption model, the optimal link grouping coefficient, the optimal transmit power of the front-haul link, the optimal transmit power of the back-haul link, the first optimal allocated task and the second optimal allocated task, to obtain a front-haul link bandwidth optimal allocation coefficient. The allocation scheme determination module is configured to perform convergence verification based on the minimum total energy consumption model, the optimal link grouping coefficient, the optimal transmit power of the front-haul link, the optimal transmit power of the back-haul link, the first optimal allocated task, the second optimal allocated task and the front-haul link bandwidth optimal allocation coefficient, to obtain a current optimal resource allocation scheme.

9. A computing offload and resource joint optimization device, comprising: The computing offloading and resource joint optimization device comprises: a processor; a memory having computer readable instructions stored thereon, the computer readable instructions being executed by the processor to implement the method of any one of claims 1 to 7.

10. A computer readable storage medium, characterized in that, The computer readable storage medium stores program codes, which can be invoked by the processor to execute the method of any one of claims 1 to 7.

Images