A method for scheduling satellite-ground cooperative tasks with consideration of energy consumption and quality of experience fairness

Abstract

The application discloses a satellite-ground cooperative task scheduling method considering energy consumption and experience quality fairness, and belongs to the technical field of edge computing, comprising the following steps: a task scheduling model is constructed based on a mobile edge computing architecture, time delay cost, energy consumption cost and experience quality fairness indexes of each to-be-scheduled task under different scheduling paths are calculated, and a satellite-ground cooperative task scheduling optimization problem with task time delay and resource capacity as constraints is established; for a scenario with a constellation scale within a preset range, a mixed integer nonlinear programming task scheduling method is adopted to solve the satellite-ground cooperative task scheduling optimization problem; otherwise, a fairness-aware hybrid column generation algorithm is adopted to solve the satellite-ground cooperative task scheduling optimization problem, so as to obtain an optimal scheduling scheme of the satellite-ground cooperative task. The application solves the problem that in the existing satellite edge computing system, energy consumption and experience quality fairness cannot be considered while meeting the time delay constraint, and the satellite-ground cooperative task cannot be jointly scheduled and unloaded.

Description

Technical Field

[0001] This invention relates to a satellite-ground collaborative task scheduling method that balances energy consumption and user experience quality fairness, and belongs to the field of edge computing technology. Background Technology

[0002] With the rapid development of low-Earth orbit (LEO) satellite constellations, onboard processors, and inter-satellite laser links, LEO satellites are evolving from simple communication relay nodes into edge nodes capable of providing computing services. In a satellite edge computing system comprised of LEO satellites, a ground-based cloud center, and mobile terminals, latency-sensitive tasks generated by users can be processed locally on the access satellite, forwarded to other satellites via inter-satellite links, or further downlinked to the ground-based cloud center for processing. Since different links have significant differences in bandwidth, propagation distance, satellite queuing load, and computing power, selecting appropriate transmission paths and processing nodes for different tasks becomes a key issue determining system performance.

[0003] Existing methods mainly include rule-based heuristics, exact optimization methods, and deep reinforcement learning methods. While rule-based heuristics offer high computational speed, they struggle to simultaneously consider global energy consumption, task latency, and inter-task service balance. Exact optimization methods can achieve high-quality solutions in small-scale systems, but as the number of satellites and tasks increases, the combinatorial space expands rapidly, making real-time solutions difficult. Deep reinforcement learning methods possess some adaptability, but with increased action space and workload, explicit control over fairness is often challenging, and they suffer from high training costs and limited generalization capabilities. Therefore, existing satellite edge computing systems generally cannot simultaneously meet task latency constraints while reducing energy consumption and improving user experience fairness, making efficient joint scheduling and offloading of satellite-ground collaborative tasks difficult. Summary of the Invention

[0004] The purpose of this invention is to provide a satellite-ground collaborative task scheduling method that balances energy consumption and user experience quality fairness. By constructing a task scheduling model, the delay cost, energy consumption cost, and user experience quality fairness index of each task to be scheduled under different scheduling paths are calculated. A satellite-ground collaborative task scheduling optimization problem is established with task delay and resource capacity as constraints. For the differences in constellation scale, a hybrid integer nonlinear programming and fairness-aware hybrid column generation algorithm are used to solve the satellite-ground collaborative task scheduling optimization problem. This solves the problem that existing satellite edge computing systems cannot simultaneously meet delay constraints and balance energy consumption and user experience quality fairness, and can meet the requirements for joint scheduling and offloading of satellite-ground collaborative tasks.

[0005] To achieve the above objectives, the present invention is implemented using the following technical solution.

[0006] This invention provides a space-ground collaborative mission scheduling method that balances energy consumption and user experience quality fairness, including:

[0007] Construct a satellite-assisted mobile edge computing architecture;

[0008] Based on the mobile edge computing architecture, a task scheduling model is constructed by acquiring a set of tasks to be scheduled, a set of processing nodes, and the status of inter-satellite links and satellite-to-ground links. The set of processing nodes includes a set of satellites and a set of ground cloud centers.

[0009] Based on the task scheduling model, the latency cost, energy consumption cost, and experience quality fairness index of each task to be scheduled are calculated under different scheduling paths.

[0010] Based on the latency cost, energy consumption cost, and experience quality fairness index, a space-ground collaborative task scheduling optimization problem is established with task latency and resource capacity as constraints.

[0011] For scenarios where the constellation size is within a preset range, a mixed-integer nonlinear programming task scheduling method is used to solve the satellite-ground collaborative task scheduling optimization problem;

[0012] For scenarios where the constellation size is outside the preset range, a fairness-aware hybrid column generation algorithm is used to solve the satellite-ground collaborative task scheduling optimization problem;

[0013] The solution to the aforementioned satellite-ground collaborative task scheduling optimization problem is adopted as the optimal scheduling scheme for satellite-ground collaborative tasks.

[0014] Furthermore, in the mobile edge computing architecture, each satellite is equipped with computing resources for maintaining a receiving queue, a transmission queue, and a computing queue. After the satellite stores the ground tasks uploaded via satellite links or base station links into the receiving queue, it makes routing decisions. The ground tasks are routed to be executed locally, forwarded via inter-satellite relays, or offloaded to any ground cloud center.

[0015] Define the task to be scheduled The routing decision variables are , ,in, This represents the total number of satellites, of which, Indicates tasks to be scheduled It was forwarded to the ground cloud center for execution. Indicates tasks to be scheduled In satellite Execute locally. Indicates tasks to be scheduled Forwarded to satellite implement, and They represent satellites and satellite The arrangement and numbering within the satellite set;

[0016] Define the task to be scheduled One of the task forwarding paths is , ,in, This indicates the number of hops in the task forwarding path. These represent the first intermediate forwarding node and the second intermediate forwarding node on the task forwarding path, respectively. One intermediate forwarding node;

[0017] Define the task to be scheduled for ,in, Indicates tasks to be scheduled The amount of input data, Indicates tasks to be scheduled Delay constraints, Indicates tasks to be scheduled Number of CPU cycles required to input each bit of data;

[0018] Define satellite and ground cloud center They are respectively and ,in:

[0019] ;

[0020] In the formula, Indicates satellite The x-coordinate of the position, Indicates satellite The ordinate of the position. Indicates satellite The vertical coordinate of the position. Indicates the center of the ground cloud The x-coordinate of the position, Indicates the center of the ground cloud The ordinate of the position. Indicates the center of the ground cloud The vertical coordinate of the position. Represents the cosine function. Represents the sine function. Represents the Earth's radius. Indicates the satellite's orbital altitude. Indicates the satellite's orbital inclination. Indicates the satellite phase angle, Indicates the latitude of the cloud center on the ground. Indicates the longitude of the center of the cloud on the ground.

[0021] Furthermore, based on the aforementioned mobile edge computing architecture, a task scheduling model is constructed by acquiring the set of tasks to be scheduled, the set of processing nodes, and the status of inter-satellite links and satellite-to-ground links.

[0022] For satellite-to-ground links:

[0023] Based on the location of the satellite and the ground cloud center, the carrier frequency, and the propagation environment parameters, a satellite-to-ground link transmission model is established, and the link path loss between the satellite and the ground cloud center is calculated.

[0024] Based on the link path loss between the satellite and the ground cloud center, the satellite antenna gain, the ground cloud center antenna gain, and the carrier wavelength, calculate the channel gain on the corresponding sub-channel.

[0025] Calculate the satellite's download rate to the ground cloud center based on the channel gain, satellite transmit power, system bandwidth, noise power, and co-channel interference on the corresponding sub-channel.

[0026] For inter-satellite links:

[0027] Based on the node positions, carrier frequencies, antenna gains, Euclidean distances, system noise temperatures, link margins, and energy required per bit of effective noise ratio of the satellites at both ends of the inter-satellite link, an inter-satellite link transmission model is established, and the inter-satellite transmission rate between satellites is calculated.

[0028] Furthermore, the link path loss between the satellite and the ground cloud center is expressed as:

[0029] ;

[0030] ;

[0031] In the formula, For satellite With ground cloud center Link path loss between It is a logarithmic function with base 10. For satellite With ground cloud center The Euclidean distance between them For carrier frequency, At the speed of light, and These are atmospheric attenuation and rain attenuation, respectively, for satellites. The position coordinates are ( , , ), Indicates satellite The x-coordinate of the position, Indicates satellite The ordinate of the position. Indicates satellite Vertical coordinates of location, center of ground cloud The position coordinates are ( , , ), Indicates the center of the ground cloud The x-coordinate of the position, Indicates the center of the ground cloud The ordinate of the position. Indicates the center of the ground cloud The vertical coordinate of the position. It is a logarithmic function with base 2;

[0032] The channel gain on the corresponding sub-channel is expressed as:

[0033] ;

[0034] In the formula, Sub-channel Up to satellite With ground cloud center Channel gain between and Satellites With ground cloud center Antenna gain, For carrier wavelength, ;

[0035] The download speed of the satellite to the ground cloud center is expressed as:

[0036] ;

[0037] In the formula, Sub-channel Up to satellite ground cloud center Download speed, For system bandwidth, For satellite The transmission power, Sub-channel Up to satellite With ground cloud center Channel gain between For noise power, For satellite With ground cloud center Co-frequency interference, , This indicates that only the permutation number of the satellite in the satellite set is greater than the satellite number. Satellites with sequential numbers Transmission power Summation, Sub-channel Up to satellite With ground cloud center Channel gain between;

[0038] The inter-satellite transmission rate between the satellites is expressed as:

[0039] ;

[0040] ;

[0041] In the formula, For satellite and satellite Inter-satellite transmission rate, For satellite The transmission power, For satellite Antenna gain, Indicates satellite and satellite Free space path loss between , For satellite With satellite Euclidean distance between satellites The position coordinates are ,in, Indicates satellite The x-coordinate of the position, Indicates satellite The ordinate of the position. Indicates satellite The vertical coordinate of the position. Boltzmann's constant, The system noise temperature, For link margin, This represents the energy required per bit of effective noise ratio, where, For each bit of energy, This represents the additional equivalent noise term that takes into account interference and system losses.

[0042] Furthermore, tasks to be scheduled The latency cost is expressed as:

[0043] ;

[0044] In the formula, Indicates tasks to be scheduled The latency cost, Indicates tasks to be scheduled Transmission delay, Indicates tasks to be scheduled The computational delay;

[0045] Tasks to be scheduled The transmission delay is expressed as:

[0046] ;

[0047] In the formula, Sub-channel Up to satellite ground cloud center Download speed, Indicates the task forwarding path The two adjacent satellites With satellite The transmission rate of the inter-satellite link;

[0048] Tasks to be scheduled The computation delay is expressed as:

[0049] ;

[0050] In the formula, Indicates tasks to be scheduled The allocated available CPU frequency, , Indicates the total CPU frequency. This indicates the length of the calculated queue.

[0051] Furthermore, tasks to be scheduled The energy cost is expressed as:

[0052] ;

[0053] In the formula, Indicates tasks to be scheduled Energy consumption cost Indicates tasks to be scheduled Transmission energy consumption, Indicates tasks to be scheduled The computational energy consumption;

[0054] Tasks to be scheduled The transmission energy consumption is expressed as:

[0055] ;

[0056] In the formula, Indicates satellite ground cloud center Transmission power, Sub-channel Up to satellite ground cloud center Download speed, Indicates satellites on the mission forwarding path The transmission power;

[0057] Tasks to be scheduled The calculated energy consumption is expressed as:

[0058] ;

[0059] In the formula, This indicates the processor's energy efficiency coefficient.

[0060] Furthermore, the satellite-ground collaborative task scheduling optimization problem is expressed as:

[0061] ;

[0062] In the formula, Representing routing decision variables and task forwarding path As optimization variables, for the objective function Perform a minimization solution. , , Indicates arrival at satellite The set of tasks to be scheduled. , and Delay costs Energy consumption costs and experience quality fairness indicators The weighting coefficients, , This is the set of tasks to be scheduled.

[0063] Furthermore, the satellite-ground collaborative task scheduling optimization problem satisfies the following constraints:

[0064] ;

[0065] ;

[0066] ;

[0067] ;

[0068] ;

[0069] ;

[0070] In the formula, Tasks to be scheduled The latency cost constraint threshold, For satellite Transmitting tasks to be scheduled The transmission power at that time For satellite Maximum transmit power, For transmission queue, To compute the queue, This is the upper limit of the queue. For indicator functions, SC n It indicates computing power.

[0071] Furthermore, for scenarios where the constellation size is within a preset range, a mixed-integer nonlinear programming task scheduling method is used to solve the satellite-ground collaborative task scheduling optimization problem, including:

[0072] The experience quality fairness term in the satellite-ground collaborative task scheduling optimization problem is replaced with a quadratic penalty term for the average latency cost deviation, and the satellite-ground collaborative task scheduling optimization problem is reconstructed into a mixed integer quadratic constraint programming problem.

[0073] For the tasks to be scheduled and the processing nodes, the allocation decision variables for the tasks to be scheduled are defined, and the objective function is constructed based on the latency cost, energy consumption cost, average latency cost deviation and soft constraint relaxation variables of each task to be scheduled under different processing nodes, so as to solve the mixed integer quadratic constraint programming problem.

[0074] The objective function is expressed as:

[0075] ;

[0076] In the formula, Indicates the scheduling task Receiving satellites and processing nodes The corresponding task allocation variables to be scheduled are minimized. ∈{0,1} represents the task to be scheduled. In satellite The location is assigned to a processing node. Decision variables, and These represent the tasks to be scheduled. In satellite The location is assigned to a processing node. Corresponding latency cost and energy consumption costs The weighting coefficients, Represents the regularization coefficient. Represents relative to average delay deviation, , , Indicates tasks to be scheduled In satellite The actual total delay is as follows. Represents soft-constraint slack variables. This represents the relaxation penalty coefficient, used to characterize the penalty cost introduced when the delay constraint is relaxed;

[0077] In the case of remote execution where the processing node is neither a receiving satellite nor a ground cloud center, a binary link selection variable is defined to indicate whether the inter-satellite link is selected, so that the task to be scheduled is transmitted from the receiving satellite to the processing node via the inter-satellite link to form an effective transmission path.

[0078] Based on the binary link selection variables, flow conservation constraints are constructed to ensure that the task to be scheduled injects a unit flow at the receiving satellite, gathers a unit flow at the processing node, and satisfies flow conservation at the intermediate relay satellite.

[0079] The flow conservation constraint is expressed as:

[0080] ;

[0081] In the formula, A set of interplanetary links, Indicates from satellite Pointing to satellite Inter-satellite links, Indicates from satellite Pointing to satellite Inter-satellite links, where, when hour, =1, when hour, =-1, in other cases =0, For satellite The flow conservation flag parameter, Select variables for binary links;

[0082] By solving the mixed-integer quadratic constrained programming problem, the optimal scheduling scheme for the space-ground collaborative mission is obtained.

[0083] Furthermore, for scenarios where the constellation size is outside a preset range, a fairness-aware hybrid column generation algorithm is used to solve the satellite-ground collaborative task scheduling optimization problem, including:

[0084] For each task to be scheduled, initialize the fairness multiplier and the set of candidate scheduling paths, and generate at least one feasible initial scheduling path for each task to be scheduled. The feasible initial scheduling path is used to ensure that there is an initial feasible solution to the restricted master problem.

[0085] The experience quality fairness term in the satellite-ground collaborative mission scheduling optimization problem is replaced with an augmented Lagrangian term based on the average delay deviation, and an augmented Lagrangian function is constructed. The augmented Lagrangian function is used to introduce the experience quality fairness constraint into a decomposable solution framework.

[0086] Among them, the definition Indicates tasks to be scheduled Select candidate scheduling path Decision variables, Indicates tasks to be scheduled Select candidate scheduling path The basic cost of the task to be scheduled Select candidate scheduling path The basic cost is expressed as:

[0087] ;

[0088] In the formula, Indicates tasks to be scheduled In candidate scheduling paths The corresponding latency cost, Indicates tasks to be scheduled In candidate scheduling paths The corresponding energy consumption cost, and These represent the latency cost weighting coefficient and the energy consumption cost weighting coefficient, respectively.

[0089] The augmented Lagrange function is expressed as:

[0090] ;

[0091] In the formula, This represents the augmented Lagrange function. This represents the set of variables for task path selection. Denotes the set of fairness Lagrange multipliers. Indicates tasks to be scheduled The corresponding fairness Lagrange multipliers are used to characterize the tasks to be scheduled. The penalty weight corresponding to when the latency cost deviates from the average latency cost;

[0092] The terms related to the candidate scheduling path selection variables in the augmented Lagrangian function are merged, and a restricted master problem is constructed on the current candidate scheduling path set of each task to be scheduled, which is used to determine the optimal path selection result of each task to be scheduled within the current candidate scheduling path set.

[0093] The restricted principal problem is represented as:

[0094] ;

[0095] In the formula, Represents the set of path selection variables for tasks to be scheduled. The set of variables related to the delay cost deviation of the tasks to be scheduled To optimize the variables, the objective function is adjusted. Perform a minimization solution. Indicates tasks to be scheduled The deviation between the actual latency cost and the average latency cost. Indicates tasks to be scheduled The square of the delay cost deviation is used to characterize the task to be scheduled. The degree of punishment in terms of time delay fairness, Indicates tasks to be scheduled The set of candidate scheduling paths;

[0096] The constraints of the restricted master problem include:

[0097] The constraints on task selection (each task to be scheduled can only choose one candidate scheduling path), resource capacity constraints (each link resource usage does not exceed the capacity limit), delay cost deviation definition constraints, and candidate scheduling path selection variable value range constraints are respectively expressed as follows:

[0098] ;

[0099] ;

[0100] ;

[0101] ;

[0102] In the formula, Indicates tasks to be scheduled via candidate scheduling path During scheduling, link resources The amount of space occupied Indicates link resources Maximum capacity, The logical judgment symbols can be chosen arbitrarily;

[0103] The corresponding dual variables are obtained by solving the restricted master problem, and a pricing subproblem is constructed to search for new candidate scheduling paths that can optimize the objective function based on the current dual variables.

[0104] The path reduction cost of the pricing subproblem satisfies:

[0105] ;

[0106] In the formula, Indicates tasks to be scheduled via candidate scheduling path The reduced cost during scheduling is used to characterize the candidate scheduling path. The degree of optimization of the objective function after incorporating the current constrained principal problem. This represents the dual variable corresponding to the selection constraint of the task to be scheduled. The dual variable representing the resource capacity constraint is used to characterize the impact of the task allocation constraint and link resource occupation on the reduction cost of the candidate scheduling path;

[0107] Repeat the processing steps until the preset fairness convergence condition is met or the preset maximum number of iterations is reached. Then, use the branch pricing method to restore the integer nature of the candidate scheduling path selection variables and obtain the optimal scheduling scheme for the space-ground collaborative mission.

[0108] The processing steps include:

[0109] When a candidate scheduling path exists with a path reduction cost less than zero, the corresponding candidate scheduling path is added to the candidate scheduling path set of the corresponding task to be scheduled. And then resolve the restricted principal problem;

[0110] The fairness multiplier is updated based on the deviation between the current total delay cost and the average delay cost of each task to be scheduled, and the restricted master problem and the pricing subproblem are solved repeatedly.

[0111] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0112] 1. This invention constructs a satellite-assisted mobile edge computing architecture, obtains the set of tasks to be scheduled, the set of processing nodes, and the status of inter-satellite links and satellite-to-ground links to build a task scheduling model, calculates the latency cost, energy consumption cost, and experience quality fairness index of each task to be scheduled under different scheduling paths, and establishes a satellite-to-ground collaborative task scheduling optimization problem constrained by task latency and resource capacity. Then, for scenarios with constellation size within a preset range, a mixed integer nonlinear programming task scheduling method is used to solve the problem, and for scenarios with constellation size outside the preset range, a fairness-aware hybrid column generation algorithm is used to solve the problem. This overcomes the shortcomings of existing rule-based heuristic methods that are difficult to balance global energy consumption, task latency, and service balance between tasks, avoids the solution difficulties caused by the expansion of the combinatorial space when the constellation size increases, and solves the problem that deep reinforcement learning methods are difficult to explicitly control fairness when the action space increases and the load increases. It can reduce energy consumption and improve experience quality fairness while meeting task latency constraints, and realize the joint scheduling and offloading of satellite-to-ground collaborative tasks.

[0113] 2. This invention constructs a satellite-assisted mobile edge computing architecture to obtain the set of tasks to be scheduled, the set of processing nodes, and the status of inter-satellite links and satellite-to-ground links. It also calculates the latency cost, energy consumption cost, and experience quality fairness index of each task to be scheduled under different scheduling paths. This invention can meet the task latency constraints while taking into account energy consumption reduction and experience quality fairness improvement, overcoming the shortcomings of existing rule-based heuristic methods that cannot simultaneously take into account global energy consumption, task latency, and service balance between tasks.

[0114] 3. This invention establishes a satellite-ground collaborative task scheduling optimization problem constrained by task latency and resource capacity, and adopts a mixed integer nonlinear programming task scheduling method to solve the problem for scenarios where the constellation size is within a preset range. This avoids the difficulty of solving the problem caused by the expansion of the combinatorial space when the number of satellites and tasks increases in the precise optimization method, and can obtain a high-quality scheduling solution under small-scale constellations.

[0115] 4. This invention employs a fairness-aware hybrid column generation algorithm to solve the satellite-ground collaborative task scheduling optimization problem for scenarios where the constellation size is outside a preset range. This solves the problem that deep reinforcement learning methods have difficulty in explicitly controlling fairness when the action space increases and the load rises. It can still achieve joint scheduling and unloading of satellite-ground collaborative tasks under large-scale constellations. Attached Figure Description

[0116] Figure 1 This is a schematic flowchart of a space-ground collaborative mission scheduling method that balances energy consumption and user experience quality fairness, provided by an embodiment of the present invention. Detailed Implementation

[0117] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0118] Example 1, as Figure 1 As shown in the figure, this embodiment introduces a space-ground collaborative mission scheduling method that balances energy consumption and user experience quality fairness, including:

[0119] Step 1: Build a satellite-assisted mobile edge computing architecture.

[0120] After constructing a satellite-assisted mobile edge computing architecture in this embodiment, satellite nodes can offload computing tasks from ground equipment to edge locations closer to the data source, reducing the detour path for tasks to be transmitted back to the ground cloud center and providing a low-latency satellite-ground collaborative basic environment for subsequent scheduling.

[0121] Step 2: Based on the mobile edge computing architecture, obtain the set of tasks to be scheduled, the set of processing nodes, and the status of inter-satellite links and satellite-to-ground links to construct a task scheduling model. The set of processing nodes includes a set of satellites and a set of ground cloud centers.

[0122] This embodiment constructs a task scheduling model by acquiring the set of tasks to be scheduled, the set of processing nodes, and the status of inter-satellite links and satellite-to-ground links. It also incorporates the set of satellites and the set of ground cloud centers into the set of processing nodes, so that scheduling decisions can utilize both on-board computing resources and ground center resources simultaneously, avoiding overload of a single node or idle links.

[0123] Step 3: Calculate the latency cost, energy consumption cost, and experience quality fairness index of each task to be scheduled under different scheduling paths based on the task scheduling model.

[0124] This embodiment calculates the latency cost, energy consumption cost, and experience quality fairness index of each task to be scheduled under different scheduling paths. It can quantify the differences of each path in three dimensions: response speed, energy consumption, and service balance among users, so as to select a path that balances efficiency and fairness.

[0125] Step 4: Based on the latency cost, energy consumption cost, and experience quality fairness index, establish a space-ground collaborative task scheduling optimization problem constrained by task latency and resource capacity.

[0126] This embodiment establishes a satellite-ground collaborative task scheduling optimization problem with task latency and resource capacity as constraints. It can transform the upper limit of task completion time and the processing capacity of each node in the actual system into a mathematical programming problem, and can prevent the optimization results from having scheduling schemes that violate physical feasibility.

[0127] Step 5: Solve the satellite-ground collaborative task scheduling optimization problem for constellation-scale differences:

[0128] For scenarios where the constellation size is within a preset range, a mixed-integer nonlinear programming task scheduling method is used to solve the satellite-ground collaborative task scheduling optimization problem;

[0129] For scenarios where the constellation size is outside the preset range, a fairness-aware hybrid column generation algorithm is used to solve the aforementioned satellite-ground collaborative task scheduling optimization problem.

[0130] This embodiment employs a mixed-integer nonlinear programming task scheduling method for scenarios where the constellation size is within a preset range, which can accurately search for the optimal solution for small-scale constellations. For scenarios where the constellation size is outside the preset range, a fairness-aware hybrid column generation algorithm is adopted, which can decompose large-scale problems through column generation and introduce fairness constraints, still obtaining near-optimal scheduling results when the number of constellation nodes increases, and can control the solution time.

[0131] Step Six: Use the solution to the aforementioned space-ground collaborative task scheduling optimization problem as the optimal scheduling scheme for space-ground collaborative tasks.

[0132] This embodiment uses the solution to the aforementioned satellite-ground collaborative task scheduling optimization problem as the optimal scheduling scheme. It can directly output which satellite or ground cloud center each task should be assigned to and which link it should be transmitted through, so that the satellite-ground collaborative system can reduce the average latency and total energy consumption in actual operation, while keeping the difference in experience quality between different users within a set range.

[0133] Example 2: This example introduces a space-ground collaborative task scheduling method that balances energy consumption and user experience quality fairness. The implementation steps include:

[0134] Step 1: Build a satellite-assisted mobile edge computing architecture.

[0135] In the mobile edge computing architecture, each satellite is equipped with computing resources to maintain a receiving queue, a transmission queue, and a computing queue. After the satellite stores the ground tasks uploaded via satellite links or base station links into the receiving queue, it makes routing decisions. The ground tasks are routed to be executed locally, forwarded via inter-satellite relays, or offloaded to any ground cloud center.

[0136] Define the task to be scheduled The routing decision variables are , ,in, This represents the total number of satellites, of which, Indicates tasks to be scheduled It was forwarded to the ground cloud center for execution. Indicates tasks to be scheduled In satellite Execute locally. Indicates tasks to be scheduled Forwarded to satellite implement, and They represent satellites and satellite The arrangement and numbering within the satellite set.

[0137] Define the task to be scheduled One of the task forwarding paths is , ,in, This indicates the number of hops in the task forwarding path. These represent the first intermediate forwarding node and the second intermediate forwarding node on the task forwarding path, respectively. One intermediate forwarding node.

[0138] Define the task to be scheduled for ,in, Indicates tasks to be scheduled The amount of input data, Indicates tasks to be scheduled Delay constraints, Indicates tasks to be scheduled The number of CPU cycles required to input each bit of data.

[0139] Define satellite and ground cloud center They are respectively and ,in:

[0140] ;

[0141] In the formula, Indicates satellite The x-coordinate of the position, Indicates satellite The ordinate of the position. Indicates satellite The vertical coordinate of the position. Indicates the center of the ground cloud The x-coordinate of the position, Indicates the center of the ground cloud The ordinate of the position. Indicates the center of the ground cloud The vertical coordinate of the position. Represents the cosine function. Represents the sine function. Represents the Earth's radius. Indicates the satellite's orbital altitude. Indicates the satellite's orbital inclination. Indicates the satellite phase angle, Indicates the latitude of the cloud center on the ground. Indicates the longitude of the center of the cloud on the ground.

[0142] Step 2: Based on the mobile edge computing architecture, obtain the set of tasks to be scheduled, the set of processing nodes, and the status of inter-satellite links and satellite-to-ground links to construct a task scheduling model. The set of processing nodes includes a set of satellites and a set of ground cloud centers.

[0143] Step 2.1: For satellite-to-ground links:

[0144] Step 2.1.1: Based on the location of the satellite and the ground cloud center, the carrier frequency, and the propagation environment parameters, establish a satellite-to-ground link transmission model between the satellite and the ground cloud center, and calculate the link path loss between the satellite and the ground cloud center.

[0145] In this embodiment, the link path loss between the satellite and the ground cloud center is expressed as:

[0146] ;

[0147] ;

[0148] In the formula, For satellite With ground cloud center Link path loss between It is a logarithmic function with base 10. For satellite With ground cloud center The Euclidean distance between them For carrier frequency, At the speed of light, and These are atmospheric attenuation and rain attenuation, respectively, for satellites. The position coordinates are ( , , ), Indicates satellite The x-coordinate of the position, Indicates satellite The ordinate of the position. Indicates satellite Vertical coordinates of location, center of ground cloud The position coordinates are ( , , ), Indicates the center of the ground cloud The x-coordinate of the position, Indicates the center of the ground cloud The ordinate of the position. Indicates the center of the ground cloud The vertical coordinate of the position. It is a logarithmic function with base 2.

[0149] Step 2.1.2: Calculate the channel gain on the corresponding sub-channel based on the link path loss between the satellite and the ground cloud center, the satellite antenna gain, the ground cloud center antenna gain, and the carrier wavelength.

[0150] In this embodiment, the channel gain on the corresponding sub-channel is expressed as:

[0151] ;

[0152] In the formula, Sub-channel Up to satellite With ground cloud center Channel gain between and Satellites With ground cloud center Antenna gain, For carrier wavelength, .

[0153] Step 2.1.3: Calculate the satellite's download rate to the ground cloud center based on the channel gain, satellite transmit power, system bandwidth, noise power, and co-channel interference on the corresponding sub-channel.

[0154] In this embodiment, the download rate of the satellite to the ground cloud center is expressed as:

[0155] ;

[0156] In the formula, Sub-channel Up to satellite ground cloud center Download speed, For system bandwidth, For satellite The transmission power, Sub-channel Up to satellite With ground cloud center Channel gain between For noise power, For satellite With ground cloud center Co-frequency interference, , This indicates that only the permutation number of the satellite in the satellite set is greater than the satellite number. Satellites with sequential numbers Transmission power Summation, Sub-channel Up to satellite With ground cloud center Channel gain between.

[0157] Step 2.2: For inter-satellite links:

[0158] Based on the node positions, carrier frequencies, antenna gains, Euclidean distances, system noise temperatures, link margins, and energy required per bit of effective noise ratio of the satellites at both ends of the inter-satellite link, an inter-satellite link transmission model is established, and the inter-satellite transmission rate between satellites is calculated.

[0159] In this embodiment, the inter-satellite transmission rate between the satellites is expressed as:

[0160] ;

[0161] ;

[0162] In the formula, For satellite and satellite Inter-satellite transmission rate, For satellite The transmission power, For satellite Antenna gain, Indicates satellite and satellite Free space path loss between , For satellite With satellite Euclidean distance between satellites The position coordinates are ,in, Indicates satellite The x-coordinate of the position, Indicates satellite The ordinate of the position. Indicates satellite The vertical coordinate of the position. Boltzmann's constant, The system noise temperature, For link margin, This represents the energy required per bit of effective noise ratio, where, For each bit of energy, This represents the additional equivalent noise term that takes into account interference and system losses.

[0163] Step 3: Calculate the latency cost, energy consumption cost, and experience quality fairness index of each task to be scheduled under different scheduling paths based on the task scheduling model.

[0164] In this embodiment, the task to be scheduled The latency cost is expressed as:

[0165] ;

[0166] In the formula, Indicates tasks to be scheduled The latency cost, Indicates tasks to be scheduled Transmission delay, Indicates tasks to be scheduled The computational delay.

[0167] In this embodiment, the task to be scheduled The transmission delay is expressed as:

[0168] ;

[0169] In the formula, Sub-channel Up to satellite ground cloud center Download speed, Indicates the task forwarding path The two adjacent satellites With satellite The transmission rate of the inter-satellite link.

[0170] In this embodiment, the task to be scheduled The computation delay is expressed as:

[0171] ;

[0172] In the formula, Indicates tasks to be scheduled The allocated available CPU frequency, , Indicates the total CPU frequency. This indicates the length of the calculated queue.

[0173] In this embodiment, the task to be scheduled The energy cost is expressed as:

[0174] ;

[0175] In the formula, Indicates tasks to be scheduled Energy consumption cost Indicates tasks to be scheduled Transmission energy consumption, Indicates tasks to be scheduled The computational energy consumption;

[0176] In this embodiment, the task to be scheduled The transmission energy consumption is expressed as:

[0177] ;

[0178] In the formula, Indicates satellite ground cloud center Transmission power, Sub-channel Up to satellite ground cloud center Download speed, Indicates satellites on the mission forwarding path The transmission power;

[0179] In this embodiment, the task to be scheduled The calculated energy consumption is expressed as:

[0180] ;

[0181] In the formula, This indicates the processor's energy efficiency coefficient.

[0182] Step 4: Based on the latency cost, energy consumption cost, and experience quality fairness index, establish a space-ground collaborative task scheduling optimization problem constrained by task latency and resource capacity.

[0183] In this embodiment, the satellite-ground collaborative task scheduling optimization problem is expressed as:

[0184] ;

[0185] In the formula, Representing routing decision variables and task forwarding path As optimization variables, for the objective function Perform a minimization solution. , , Indicates arrival at satellite The set of tasks to be scheduled. , and Delay costs Energy consumption costs and experience quality fairness indicators The weighting coefficients, , This is the set of tasks to be scheduled.

[0186] In this embodiment, the satellite-ground collaborative mission scheduling optimization problem satisfies the following constraints:

[0187] ;

[0188] ;

[0189] ;

[0190] ;

[0191] ;

[0192] ;

[0193] In the formula, Tasks to be scheduled The latency cost constraint threshold, For satellite Transmitting tasks to be scheduled The transmission power at that time For satellite Maximum transmit power, For transmission queue, To compute the queue, This is the upper limit of the queue. This is an indicator function.

[0194] Step 5: Solve the satellite-ground collaborative task scheduling optimization problem for constellation-scale differences:

[0195] Step 5.1: For scenarios where the constellation size is within a preset range, the satellite-ground collaborative task scheduling optimization problem is solved using a mixed-integer nonlinear programming task scheduling method.

[0196] Step 5.1.1: Replace the experience quality fairness term in the satellite-ground collaborative task scheduling optimization problem with a quadratic penalty term for the average latency cost deviation, and reconstruct the satellite-ground collaborative task scheduling optimization problem into a mixed integer quadratic constraint programming problem;

[0197] Step 5.1.2: For the tasks to be scheduled and the processing nodes, define the allocation decision variables for the tasks to be scheduled, and construct the objective function based on the latency cost, energy consumption cost, average latency cost deviation and soft constraint relaxation variables of each task to be scheduled under different processing nodes, so as to solve the mixed integer quadratic constrained programming problem.

[0198] In this embodiment, the objective function is expressed as:

[0199] ;

[0200] In the formula, Indicates the scheduling task Receiving satellites and processing nodes The corresponding task allocation variables to be scheduled are minimized. ∈{0,1} represents the task to be scheduled. In satellite The location is assigned to a processing node. Decision variables, and These represent the tasks to be scheduled. In satellite The location is assigned to a processing node. Corresponding latency cost and energy consumption costs The weighting coefficients, Represents the regularization coefficient. Represents relative to average delay deviation, , , Indicates tasks to be scheduled In satellite The actual total delay is as follows. Represents soft-constraint slack variables. This represents the relaxation penalty coefficient, used to characterize the penalty cost introduced when the delay constraint is relaxed.

[0201] In this embodiment, for remote execution scenarios where the processing node is neither a receiving satellite nor a ground cloud center, a binary link selection variable is defined to indicate whether an inter-satellite link is selected, so that the task to be scheduled is transmitted from the receiving satellite to the processing node via the inter-satellite link to form an effective transmission path; based on the binary link selection variable, a flow conservation constraint is constructed so that the task to be scheduled is injected with a unit flow at the receiving satellite, aggregated with a unit flow at the processing node, and satisfies flow conservation at the intermediate relay satellite.

[0202] In this embodiment, the flow conservation constraint is expressed as:

[0203] ;

[0204] In the formula, A set of interplanetary links, Indicates from satellite Pointing to satellite Inter-satellite links, Indicates from satellite Pointing to satellite Inter-satellite links, where, when hour, =1, when hour, =-1, in other cases =0, For satellite The flow conservation flag parameter, Select variables for binary links.

[0205] Step 5.1.3: By solving the mixed integer quadratic constrained programming problem, the optimal scheduling scheme for the space-ground collaborative mission is obtained.

[0206] Step 5.2: For scenarios where the constellation size is outside the preset range, a fairness-aware hybrid column generation algorithm is used to solve the satellite-ground collaborative task scheduling optimization problem:

[0207] Step 5.2.1: For each task to be scheduled, initialize the fairness multiplier and the set of candidate scheduling paths, and generate at least one feasible initial scheduling path for each task to be scheduled. The feasible initial scheduling path is used to ensure that there is an initial feasible solution to the restricted principal problem.

[0208] Step 5.2.2: Replace the experience quality fairness term in the satellite-ground cooperative mission scheduling optimization problem with an augmented Lagrangian term based on the average delay deviation, and construct an augmented Lagrangian function. The augmented Lagrangian function is used to introduce the experience quality fairness constraint into a decomposable solution framework.

[0209] Step 5.2.3: Definition Indicates tasks to be scheduled Select candidate scheduling path Decision variables, Indicates tasks to be scheduled Select candidate scheduling path The basic cost.

[0210] In this embodiment, the task to be scheduled Select candidate scheduling path The basic cost is expressed as:

[0211] ;

[0212] In the formula, Indicates tasks to be scheduled In candidate scheduling paths The corresponding latency cost, Indicates tasks to be scheduled In candidate scheduling paths The corresponding energy consumption cost, and These represent the delay cost weighting coefficient and the energy consumption cost weighting coefficient, respectively.

[0213] In this embodiment, the augmented Lagrange function is expressed as:

[0214] ;

[0215] In the formula, This represents the augmented Lagrange function. This represents the set of variables for task path selection. Denotes the set of fairness Lagrange multipliers. Indicates tasks to be scheduled The corresponding fairness Lagrange multipliers are used to characterize the tasks to be scheduled. The penalty weight corresponding to when the latency cost deviates from the average latency cost.

[0216] The terms related to the candidate scheduling path selection variables in the augmented Lagrangian function are merged, and a restricted principal problem is constructed on the current candidate scheduling path set of each task to be scheduled, which is used to determine the optimal path selection result of each task to be scheduled within the current candidate scheduling path set.

[0217] In this embodiment, the restricted principal problem is represented as:

[0218] ;

[0219] In the formula, Represents the set of path selection variables for tasks to be scheduled. The set of variables related to the delay cost deviation of the tasks to be scheduled To optimize the variables, the objective function is adjusted. Perform a minimization solution. Indicates tasks to be scheduled The deviation between the actual latency cost and the average latency cost. Indicates tasks to be scheduled The square of the delay cost deviation is used to characterize the task to be scheduled. The degree of punishment in terms of time delay fairness.

[0220] In this embodiment, the constraints of the restricted principal problem include:

[0221] The constraints on task selection (each task to be scheduled can only choose one candidate scheduling path), resource capacity constraints (each link resource usage does not exceed the capacity limit), delay cost deviation definition constraints, and candidate scheduling path selection variable value range constraints are respectively expressed as follows:

[0222] ;

[0223] ;

[0224] ;

[0225] ;

[0226] In the formula, Indicates tasks to be scheduled via candidate scheduling path During scheduling, link resources The amount of space occupied Indicates link resources Maximum capacity, The logical judgment symbol can be chosen arbitrarily.

[0227] Step 5.2.4: Obtain the corresponding dual variables by solving the restricted master problem, and construct a pricing subproblem for searching for new candidate scheduling paths that can optimize the objective function based on the current dual variables.

[0228] In this embodiment, the path reduction cost of the pricing subproblem satisfies:

[0229] ;

[0230] In the formula, Indicates tasks to be scheduled via candidate scheduling path The reduced cost during scheduling is used to characterize the candidate scheduling path. The degree of optimization of the objective function after incorporating the current constrained principal problem. This represents the dual variable corresponding to the selection constraint of the task to be scheduled. This represents the dual variable corresponding to the resource capacity constraint, used to characterize the impact of the task allocation constraint and link resource occupation on the reduction cost of the candidate scheduling path.

[0231] Step 5.2.5: Repeat the processing steps until the preset fairness convergence condition is met or the preset maximum number of iterations is reached. Use the branch pricing method to restore the integer nature of the candidate scheduling path selection variables and obtain the optimal scheduling scheme for the space-ground collaborative mission.

[0232] In this embodiment, the processing steps include:

[0233] When a candidate scheduling path exists with a path reduction cost less than zero, the corresponding candidate scheduling path is added to the candidate scheduling path set of the corresponding task to be scheduled. And then resolve the restricted principal problem;

[0234] The fairness multiplier is updated based on the deviation between the current total delay cost and the average delay cost of each task to be scheduled, and the restricted master problem and the pricing subproblem are solved repeatedly.

[0235] Step 6: Use the solution to the satellite-ground collaborative task scheduling optimization problem as the optimal scheduling scheme for the satellite-ground collaborative task.

[0236] In summary, this invention constructs a satellite-assisted mobile edge computing architecture, acquires the set of tasks to be scheduled, the set of processing nodes, and the status of inter-satellite links and satellite-to-ground links to build a task scheduling model, calculates the latency cost, energy consumption cost, and experience quality fairness index of each task to be scheduled under different scheduling paths, and establishes a satellite-to-ground collaborative task scheduling optimization problem constrained by task latency and resource capacity. Then, for scenarios with constellation size within a preset range, a mixed integer nonlinear programming task scheduling method is used to solve the problem; for scenarios with constellation size outside the preset range, a fairness-aware hybrid column generation algorithm is used to solve the problem. This overcomes the shortcomings of existing rule-based heuristic methods in balancing global energy consumption, task latency, and service balance between tasks, avoids the solution difficulties caused by the expansion of the combinatorial space when the constellation size increases, and solves the problem that deep reinforcement learning methods are difficult to explicitly control fairness when the action space increases and the load increases. It can reduce energy consumption and improve experience quality fairness while meeting task latency constraints, and realize the joint scheduling and offloading of satellite-to-ground collaborative tasks.

[0237] This invention constructs a satellite-assisted mobile edge computing architecture to obtain the set of tasks to be scheduled, the set of processing nodes, and the status of inter-satellite links and satellite-to-ground links. It also calculates the latency cost, energy consumption cost, and experience quality fairness index of each task under different scheduling paths. This invention can meet the task latency constraints while taking into account energy consumption reduction and experience quality fairness improvement, overcoming the shortcomings of existing rule-based heuristic methods that cannot simultaneously take into account global energy consumption, task latency, and service balance between tasks.

[0238] This invention establishes a satellite-ground collaborative task scheduling optimization problem constrained by task latency and resource capacity, and adopts a mixed-integer nonlinear programming task scheduling method to solve the problem for scenarios where the constellation size is within a preset range. This avoids the difficulty of solving the problem caused by the expansion of the combinatorial space when the number of satellites and tasks increases in the precise optimization method, and can obtain a high-quality scheduling solution under small-scale constellations.

[0239] This invention employs a fairness-aware hybrid column generation algorithm to solve the satellite-ground collaborative task scheduling optimization problem for scenarios where the constellation size is outside a preset range. It solves the problem that deep reinforcement learning methods have difficulty in explicitly controlling fairness when the action space increases and the load rises. It can still achieve joint scheduling and offloading of satellite-ground collaborative tasks under large-scale constellations.

[0240] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention 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.

[0241] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. 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 illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0242] 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.

[0243] 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.

[0244] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method for satellite-ground cooperative task scheduling with energy consumption and quality of experience fairness, characterized in that, include: Construct a satellite-assisted mobile edge computing architecture; Based on the mobile edge computing architecture, a task scheduling model is constructed by acquiring a set of tasks to be scheduled, a set of processing nodes, and the status of inter-satellite links and satellite-to-ground links. The set of processing nodes includes a set of satellites and a set of ground cloud centers. Based on the task scheduling model, the latency cost, energy consumption cost, and experience quality fairness index of each task to be scheduled are calculated under different scheduling paths. Based on the latency cost, energy consumption cost, and experience quality fairness index, a space-ground collaborative task scheduling optimization problem is established with task latency and resource capacity as constraints. The satellite-ground collaborative task scheduling optimization problem is expressed as follows: ； In the formula, Representing routing decision variables and task forwarding path As optimization variables, for the objective function Perform a minimization solution. , , Indicates arrival at satellite The set of tasks to be scheduled. , and Delay costs Energy consumption costs And experience quality fairness indicators The weighting coefficients, , Let be the set of tasks to be scheduled, where, Indicates the total number of satellites. Indicates tasks to be scheduled The latency cost, Indicates tasks to be scheduled Energy consumption cost; The satellite-ground collaborative mission scheduling optimization problem satisfies the following constraints: ； ； ； ； ； ； In the formula, Tasks to be scheduled The latency cost constraint threshold, For satellite Transmitting tasks to be scheduled The transmission power at that time For satellite Maximum transmission power, For transmission queue, To compute the queue, This is the upper limit of the queue. For indicator functions, SC n Indicates computing power. Indicates tasks to be scheduled Transmission delay, Indicates tasks to be scheduled The computational delay, Indicates tasks to be scheduled The allocated available CPU frequency, Indicates tasks to be scheduled Routing decision variables, Indicates the processing node; For scenarios where the constellation size is within a preset range, a mixed-integer nonlinear programming task scheduling method is used to solve the satellite-ground collaborative task scheduling optimization problem; For scenarios where the constellation size is outside the preset range, a fairness-aware hybrid column generation algorithm is used to solve the satellite-ground collaborative task scheduling optimization problem; The solution to the aforementioned satellite-ground collaborative task scheduling optimization problem is adopted as the optimal scheduling scheme for satellite-ground collaborative tasks.

2. The space-ground collaborative task scheduling method that balances energy consumption and user experience quality fairness according to claim 1, characterized in that, In the mobile edge computing architecture, each satellite is equipped with computing resources for maintaining a receiving queue, a transmission queue, and a computing queue. After the satellite stores the ground tasks uploaded via satellite links or base station links into the receiving queue, it makes routing decisions. The ground tasks are routed to be executed locally, forwarded via inter-satellite relays, or offloaded to any ground cloud center. Define the task to be scheduled The routing decision variables are , , Indicates tasks to be scheduled It was forwarded to the ground cloud center for execution. Indicates tasks to be scheduled In satellite Execute locally. Indicates tasks to be scheduled Forwarded to satellite implement, and They represent satellites and satellite The arrangement and numbering within the satellite set; Define the task to be scheduled One of the task forwarding paths is , ,in, This indicates the number of hops in the task forwarding path. These represent the first intermediate forwarding node and the second intermediate forwarding node on the task forwarding path, respectively. One intermediate forwarding node; Define the task to be scheduled for ,in, Indicates tasks to be scheduled The amount of input data, Indicates tasks to be scheduled Delay constraints, Indicates tasks to be scheduled The number of CPU cycles required to input each bit of data; Define satellite and ground cloud center They are respectively and ,in: ； In the formula, Indicates satellite The x-coordinate of the position, Indicates satellite The ordinate of the position. Indicates satellite The vertical coordinate of the position. Indicates the center of the ground cloud The x-coordinate of the position, Indicates the center of the ground cloud The ordinate of the position. Indicates the center of the ground cloud The vertical coordinate of the position. Represents the cosine function. Represents the sine function. Represents the Earth's radius. Indicates the satellite's orbital altitude. Indicates the satellite's orbital inclination. Indicates the satellite phase angle, Indicates the latitude of the cloud center on the ground. Indicates the longitude of the center of the cloud on the ground.

3. The space-ground collaborative task scheduling method that balances energy consumption and user experience quality fairness according to claim 2, characterized in that, Based on the aforementioned mobile edge computing architecture, a task scheduling model is constructed by acquiring the set of tasks to be scheduled, the set of processing nodes, and the status of inter-satellite links and satellite-to-ground links. For satellite-to-ground links: Based on the location of the satellite and the ground cloud center, the carrier frequency, and the propagation environment parameters, a satellite-to-ground link transmission model is established, and the link path loss between the satellite and the ground cloud center is calculated. Based on the link path loss between the satellite and the ground cloud center, the satellite antenna gain, the ground cloud center antenna gain, and the carrier wavelength, calculate the channel gain on the corresponding sub-channel. Calculate the satellite's download rate to the ground cloud center based on the channel gain, satellite transmit power, system bandwidth, noise power, and co-channel interference on the corresponding sub-channel. For inter-satellite links: Based on the node positions, carrier frequencies, antenna gains, Euclidean distances, system noise temperatures, link margins, and energy required per bit of effective noise ratio of the satellites at both ends of the inter-satellite link, an inter-satellite link transmission model is established, and the inter-satellite transmission rate between satellites is calculated.

4. The space-ground collaborative task scheduling method that balances energy consumption and user experience quality fairness according to claim 3, characterized in that, The link path loss between the satellite and the ground cloud center is expressed as: ； ； In the formula, For satellite With ground cloud center Link path loss between It is a logarithmic function with base 10. For satellite With ground cloud center The Euclidean distance between them For carrier frequency, At the speed of light, and These are atmospheric attenuation and rain attenuation, respectively, for satellites. The position coordinates are ( , , ), Indicates satellite The x-coordinate of the position, Indicates satellite The ordinate of the position. Indicates satellite Vertical coordinates of location, center of ground cloud The position coordinates are ( , , ), Indicates the center of the ground cloud The x-coordinate of the position, Indicates the center of the ground cloud The ordinate of the position. Indicates the center of the ground cloud The vertical coordinate of the position. It is a logarithmic function with base 2; The channel gain on the corresponding sub-channel is expressed as: ； In the formula, Sub-channel Up to satellite With ground cloud center Channel gain between and Satellites and ground cloud center Antenna gain, For carrier wavelength, ; The download speed of the satellite to the ground cloud center is expressed as: ； In the formula, Sub-channel Up to satellite ground cloud center Download speed, For system bandwidth, For satellite The transmission power, Sub-channel Up to satellite With ground cloud center Channel gain between For noise power, For satellite With ground cloud center Co-frequency interference, , This indicates that only the permutation number of the satellite in the satellite set is greater than the satellite number. Satellites with sequential numbers Transmission power Summation, Sub-channel Up to satellite With ground cloud center Channel gain between; The inter-satellite transmission rate between the satellites is expressed as: ； ； In the formula, For satellite and satellite Inter-satellite transmission rate, Indicates satellite and satellite Free space path loss between , For satellite With satellite Euclidean distance between satellites The position coordinates are , Indicates satellite The x-coordinate of the position, Indicates satellite The ordinate of the position. Indicates satellite The vertical coordinate of the position. Boltzmann's constant, The system noise temperature, For link margin, This represents the energy required to represent the effective noise ratio per bit. For each bit of energy, This represents the additional equivalent noise term that takes into account interference and system losses.

5. The space-ground collaborative mission scheduling method that balances energy consumption and user experience quality fairness according to claim 4, characterized in that, Tasks to be scheduled The latency cost is expressed as: ； Tasks to be scheduled The transmission delay is expressed as: ； In the formula, Sub-channel Up to satellite ground cloud center Download speed, Indicates the task forwarding path The two adjacent satellites With satellite The transmission rate of the inter-satellite link; Tasks to be scheduled The computation delay is expressed as: ； In the formula, , Indicates the total CPU frequency. This indicates the length of the calculated queue.

6. The space-ground collaborative task scheduling method that balances energy consumption and user experience quality fairness according to claim 5, characterized in that, Tasks to be scheduled The energy cost is expressed as: ； In the formula, Indicates tasks to be scheduled Transmission energy consumption, Indicates tasks to be scheduled The computational energy consumption; Tasks to be scheduled The transmission energy consumption is expressed as: ； In the formula, Indicates satellite ground cloud center Transmission power, Sub-channel Up to satellite ground cloud center Download speed, Indicates satellites on the mission forwarding path The transmission power; Tasks to be scheduled The calculated energy consumption is expressed as: ； In the formula, This indicates the processor's energy efficiency coefficient.

7. The space-ground collaborative task scheduling method that balances energy consumption and user experience quality fairness according to claim 6, characterized in that, For scenarios where the constellation size is within a preset range, a mixed-integer nonlinear programming task scheduling method is used to solve the satellite-ground collaborative task scheduling optimization problem, including: The experience quality fairness term in the satellite-ground collaborative task scheduling optimization problem is replaced with a quadratic penalty term for the average latency cost deviation, and the satellite-ground collaborative task scheduling optimization problem is reconstructed into a mixed integer quadratic constraint programming problem. For the tasks to be scheduled and the processing nodes, the allocation decision variables for the tasks to be scheduled are defined, and the objective function is constructed based on the latency cost, energy consumption cost, average latency cost deviation and soft constraint relaxation variables of each task to be scheduled under different processing nodes, so as to solve the mixed integer quadratic constraint programming problem. The objective function is expressed as: ； In the formula, Indicates the scheduling task Receive satellite and processing nodes The corresponding task allocation variables to be scheduled are minimized. Indicates tasks to be scheduled In satellite The location is assigned to a processing node. Decision variables, ∈{0,1}, and These represent the tasks to be scheduled. In satellite The location is assigned to a processing node. Corresponding latency cost and energy consumption costs The weighting coefficients, Represents the regularization coefficient. Represents relative to average delay deviation, , , Indicates tasks to be scheduled In satellite The actual total delay is as follows. Represents soft-constraint slack variables. Indicates the relaxation penalty coefficient; In the case of remote execution where the processing node is neither a receiving satellite nor a ground cloud center, a binary link selection variable is defined to indicate whether the inter-satellite link is selected, so that the task to be scheduled is transmitted from the receiving satellite to the processing node via the inter-satellite link to form an effective transmission path. Based on the binary link selection variables, flow conservation constraints are constructed to ensure that the task to be scheduled injects a unit flow at the receiving satellite, gathers a unit flow at the processing node, and satisfies flow conservation at the intermediate relay satellite. The flow conservation constraint is expressed as: ； In the formula, A set of interplanetary links, Indicates from satellite Pointing to satellite Inter-satellite links, Indicates from satellite Pointing to satellite Inter-satellite links, where, when hour, =1, when hour, =-1, in other cases =0, For satellite The flow conservation flag parameter, Select variables for binary links; By solving the mixed-integer quadratic constrained programming problem, the optimal scheduling scheme for the space-ground collaborative mission is obtained.

8. The space-ground collaborative mission scheduling method that balances energy consumption and user experience quality fairness according to claim 7, characterized in that, For scenarios where the constellation size is outside a preset range, a fairness-aware hybrid column generation algorithm is used to solve the satellite-ground collaborative task scheduling optimization problem, including: For each task to be scheduled, initialize the fairness multiplier and the set of candidate scheduling paths, and generate at least one feasible initial scheduling path for each task to be scheduled. The feasible initial scheduling path is used to ensure that there is an initial feasible solution to the restricted master problem. The experience quality fairness term in the satellite-ground collaborative mission scheduling optimization problem is replaced with an augmented Lagrangian term based on the average delay deviation, and an augmented Lagrangian function is constructed. The augmented Lagrangian function is used to introduce the experience quality fairness constraint into a decomposable solution framework. Among them, the definition Indicates tasks to be scheduled Select candidate scheduling path Decision variables, Indicates tasks to be scheduled Select candidate scheduling path The basic cost of the task to be scheduled Select candidate scheduling path The basic cost is expressed as: ； In the formula, Indicates tasks to be scheduled In candidate scheduling paths The corresponding latency cost, Indicates tasks to be scheduled In candidate scheduling paths The corresponding energy consumption cost, and These represent the latency cost weighting coefficient and the energy consumption cost weighting coefficient, respectively. The augmented Lagrange function is expressed as: ； In the formula, This represents the augmented Lagrange function. This represents the set of variables for task path selection. Denotes the set of fairness Lagrange multipliers. Indicates tasks to be scheduled The corresponding fairness Lagrange multipliers; The terms related to the candidate scheduling path selection variables in the augmented Lagrangian function are merged, and a restricted master problem is constructed on the current candidate scheduling path set of each task to be scheduled, which is used to determine the optimal path selection result of each task to be scheduled within the current candidate scheduling path set. The restricted principal problem is represented as: ； In the formula, Represents the set of path selection variables for tasks to be scheduled. The set of variables related to the delay cost deviation of the tasks to be scheduled To optimize the variables, the objective function is adjusted. Perform a minimization solution. Indicates tasks to be scheduled The deviation between the actual latency cost and the average latency cost. Indicates tasks to be scheduled The square of the delay cost deviation, Indicates tasks to be scheduled The set of candidate scheduling paths; The constraints of the restricted master problem include: The constraints on task selection (each task to be scheduled can only choose one candidate scheduling path), resource capacity constraints (each link resource usage does not exceed the capacity limit), delay cost deviation definition constraints, and candidate scheduling path selection variable value range constraints are respectively expressed as follows: ； ； ； ； In the formula, Indicates tasks to be scheduled via candidate scheduling path During scheduling, link resources The amount of space occupied Indicates link resources Maximum capacity The logical judgment symbols can be chosen arbitrarily; The corresponding dual variables are obtained by solving the restricted master problem, and a pricing subproblem is constructed to search for new candidate scheduling paths that can optimize the objective function based on the current dual variables. The path reduction cost of the pricing subproblem satisfies: ； In the formula, Indicates tasks to be scheduled via candidate scheduling path The reduced cost during scheduling is used to characterize the candidate scheduling path. The degree of optimization of the objective function after incorporating the current constrained principal problem. This represents the dual variable corresponding to the selection constraint of the task to be scheduled. Represents the dual variable corresponding to the resource capacity constraint; Repeat the processing steps until the preset fairness convergence condition is met or the preset maximum number of iterations is reached. Then, use the branch pricing method to restore the integer nature of the candidate scheduling path selection variables and obtain the optimal scheduling scheme for the space-ground collaborative mission. The processing steps include: When a candidate scheduling path exists with a path reduction cost less than zero, the corresponding candidate scheduling path is added to the corresponding task to be scheduled. Candidate scheduling path set And then resolve the restricted principal problem; The fairness multiplier is updated based on the deviation between the current total delay cost and the average delay cost of each task to be scheduled, and the restricted master problem and the pricing subproblem are solved repeatedly.

Images