Relay satellite scheduling method considering antenna resource heterogeneity and task diversity

Abstract

The invention discloses a relay satellite scheduling method considering antenna resource heterogeneity and task diversity, and the method specifically comprises the following steps: 1, discretizing a scheduling period into equal-length time slots, and constructing a time slot set S; constructing a visibility mask and a service window mask for any task t and antenna a; 2, generating an initial feasible scheduling scheme by adopting a conflict perception initialization strategy; and step 3, performing iterative improvement on the initial scheme generated in the CAI stage by adopting a constraint local re-optimization strategy, thereby improving the task completion rate and the target function value. According to the invention, the overall task completion rate and the bandwidth utilization efficiency of the relay system can be improved.

Description

Technical Field

[0001] This invention belongs to the field of aerospace mission scheduling and space information network resource management technology, and relates to a relay satellite scheduling method that takes into account the heterogeneity of antenna resources and the diversity of missions. Background Technology

[0002] With the rapid increase in the number of low-Earth orbit spacecraft and the diversification of business needs, geostationary Earth Orbit Data Relay Satellites (GEO DRS) have become a key infrastructure for building a globally covered, highly reliable, and high-bandwidth space communication network. Next-generation DRS systems generally employ a hybrid configuration of single-access (SA) and multi-access (MA) antennas: SA antennas provide high link quality but are highly exclusive, while MA antennas can support concurrent connections but are limited by total bandwidth and maximum concurrency.

[0003] Meanwhile, user task types have expanded from traditional fixed-duration tasks to large-scale data download tasks, the latter exhibiting characteristics such as segmented execution and variable transmission rates. Furthermore, the direct transmission links supported by ground stations can provide high instantaneous rates within a short visible window, making resource coordination between relay links and direct transmission links a crucial component of the scheduling problem.

[0004] Existing research generally employs modeling approaches based on homogeneous antenna resources and a single task type, making it difficult to reflect the differences in concurrency capabilities and bandwidth allocation characteristics under conditions of coexistence of single-site and multiple-site antennas. The handling of task diversity remains insufficient, often treating all tasks as non-segmentable or uniformly assuming them to be data-driven tasks, failing to reflect the continuous occupancy requirements of fixed-duration tasks. Furthermore, there is a lack of explicit collaborative modeling for direct transmission links from ground stations, failing to fully utilize space-to-ground multi-link resources. Regarding solution strategies, existing heuristic methods often employ single-stage greedy algorithms or traditional metaheuristic structures, which are ill-suited to the complexity brought about by heterogeneous antenna resources, differences in task attributes, and multi-link coupling, easily getting trapped in local optima and resulting in limited task scheduling performance. Summary of the Invention

[0005] The purpose of this invention is to provide a relay satellite scheduling method that takes into account the heterogeneity of antenna resources and the diversity of tasks. This method can improve the overall task completion rate and bandwidth utilization efficiency of the relay system.

[0006] The technical solution adopted in this invention is a relay satellite scheduling method that considers the heterogeneity of antenna resources and the diversity of tasks, specifically including the following steps:

[0007] Step 1: Discretize the scheduling period into equal-length time slots. Construct a time slot set S For any task t With antenna a Construct a visibility mask Service window mask ; Step 2: Generate an initial feasible scheduling scheme using a conflict-aware initialization strategy; Step 3: The initial scheme generated in the CAI stage is iteratively improved by adopting a constrained local re-optimization strategy, thereby improving the task completion rate and the objective function value.

[0008] The invention is further characterized by: The specific process of step 1 is as follows: Step 1.1, determine the parameter variables; Step 1.2: Construct a mathematical programming model based on the parameter variables determined in Step 1.1, with the goal of maximizing the weighted sum of tasks completed.

[0009] The specific process of step 1.1 is as follows: Define decision variables. , indicating task t Whether it is completed, if Indicates task t Completed, if Indicates task t Incomplete; Indicates antenna a In the time slot s Is it a service task? t ,like Indicates antenna a Service task in time slot s t ,like Indicates antenna a In the time slot s Non-service tasks t ; Indicates a fixed-duration task t Is it in the time slot? s For antenna a In the beginning, if Indicates a fixed-duration task t In the time slot s For antenna a In the beginning, if Indicates a fixed-duration task t Not in time slot s For antenna a start; Indicates antenna a In the time slot s Assigned to task t The bandwidth.

[0010] The specific process of step 1.2 is as follows: The objective function of the mathematical programming model is constructed using the following formula (1): (1) In equation (1), Indicates task t The weights; (2) Equation (2) represents the slot-level visibility constraint: (3) Equation (3) represents the service window constraint: (4) Equation (4) indicates that each completed fixed-duration task has one and only one start time slot; (5) Equation (5) indicates that a fixed-duration task must occupy [a certain amount of time] to complete. Each time slot; (6) Equation (6) indicates that a fixed-duration task must be continuously occupied after the selected start time slot. Each time slot; (7) Equation (7) is used to constrain fixed-duration tasks to select at most one window from all available visible windows for execution; (8) Equation (8) establishes a consistent relationship between the task completion status and window selection. If the task is completed, its window selection variable must be 1; if the window selection variable is not activated, the task cannot be considered completed. (9) Equation (9) indicates that the cumulative amount of data transmitted for the completed quantity constraint task must meet the data volume requirement. ; (10) Equation (10) is used to ensure data volume constraints in the task. Antenna resources can only be used when the task is deemed complete; otherwise, all time slot resource allocations must be invalidated.

[0011] (11) Equation (11) indicates that each time slot of the SA antenna can serve at most one task; (12) Equation (12) represents the cooling constraint for SA antenna switching. To switch the number of cooldown slots, no new tasks can be started during the cooldown period; (13) Equation (13) indicates that the SA antenna is allocated a fixed rated bandwidth. ; (14) Equation (14) indicates that the total bandwidth allocated to each time slot of the MA antenna does not exceed its capacity. ; (15) Equation (15) indicates that the number of concurrent tasks per time slot of the MA antenna does not exceed the upper limit. ; (16) Equation (16) indicates that the antennas allocated for fixed-duration tasks must meet the minimum bandwidth requirements. ; (17) Equation (17) indicates that the same user spacecraft can only occupy one relay or direct transmission link in any time slot.

[0012] In step 1.2, the scheduling time slot length This refers to a unified minimum time granularity; time slot set. , of which elements s Indicates the first s Time slots; task set It contains all relay transmission tasks to be scheduled; a fixed-duration task set. The task must be executed completely in a continuous time slot and cannot be segmented; the data volume constrains the task set. The task can be executed in segments within the window, and it is considered complete when the cumulative transmission volume reaches the target. User spacecraft collection Each mission t corresponds to a unique user spacecraft. ; Antenna assembly It includes all available relay and ground station antenna resources; single-site antenna collection A single-address antenna can only serve one task in any given time slot; Multiple access antenna array Multiple access antennas can serve multiple tasks in parallel within the same time slot; Task weight Reflecting the task t The relative importance and target benefit are used as weighting coefficients in the objective function; Task Service Window Limited taskt The range of executable start and end time slots; Window indicator variable When the gap s The value is 1 if the task falls into the task window, otherwise it is 0. Visibility parameters Taking 1 indicates a task t With antenna a In the time slot s A link can be established; otherwise, the value is 0. Task – Antenna Service Time Slot Set , indicating task t It can be in the antenna a The set of time slot indexes executed on the current execution; Task-level serviceable time slot set For the task t Joint visible time slots on all antennas; The earliest feasible start time slot and the latest feasible end time slot of the task , respectively The minimum and maximum elements; Set of visible window indices for task t and antenna a ; No. j Visible window , indicating task t With antenna a The continuous visible time interval; The actual duration of a fixed-duration task The unit is seconds, used to calculate the corresponding number of discrete time slots; Number of consecutive time slots required for a fixed-duration task ,for Divide by the time slot length Δ and round up.

[0013] The specific process of step 2 is as follows: Step 2.1, for each task t With each antenna a Calculate all visible windows w Conflict level in the context of conflict, conflict level conf( w As shown in equation (18): (18) in, W a Indicates antenna a The set of all visible windows Representation and window w Different from other windows; Step 2.2, all fixed-duration tasks Sort the windows by conflict degree in descending order, find and assign consecutive time slots that satisfy the constraints in turn; Step 2.3, for each task t According to the earliest feasible start time slot The theoretical maximum transmission upper bound starting from this time slot is calculated as shown in equation (19): (19) in, Indicates the start time slot The initial theoretical maximum transport upper bound, This indicates utilizing the remaining bandwidth, i.e., in the time slot. r Inside, the mission t The maximum usable bandwidth is the maximum bandwidth visible among all antenna resources that is not occupied by other tasks. Step 2.4, calculate the data volume constraints for each task. Completion difficulty indicators The calculation formula is as follows: (20) in, Indicates task t Data volume requirements; Step 2.5: Sort all tasks and allocate bandwidth in sequence; Step 2.6: Record the current scheduling scheme as the initial feasible solution. .

[0014] The specific process of step 2.5 is as follows: First, based on the difficulty of completing the task. Tasks are sorted in ascending order, with easier tasks prioritized; then, if tasks are of equal difficulty, their deadlines are determined by the deadlines they submit. Sort tasks in ascending order to ensure that time-sensitive tasks are executed first; finally, if tasks have the same difficulty and deadline, then prioritize them based on their weight. The tasks are sorted in descending order, prioritizing high-value tasks. Based on the sorting results, a greedy sweep strategy is used for bandwidth allocation. In each bandwidth allocation, dynamic bandwidth allocation is performed according to equation (21): (twenty one) in, Assigned to task t Bandwidth on time slot s, It is a task t In the time slot s Initial remaining data requirements, It is an antenna. a In the time slot s Already allocated bandwidth.

[0015] The specific process of step 3 is as follows: Step 3.1: For each task that remains unfinished after the CAI phase. t calculate This constitutes the optimization interval; Step 3.2, in the interval The operation sequentially executes time-order compaction, restricted local translation, and bandwidth filling for tasks with unfinished data volume constraints. Step 3.3, if the data volume is a constraint on the task If not yet completed, tasks will be assigned according to their weight. Remove low-weight, fixed-duration tasks that have already been scheduled within the ascending order of the interval. The complete paragraph; Step 3.4: Establish an accept-rollback mechanism; Step 3.5: Repeat steps 3.1 to 3.4 until the maximum number of iterations set by the algorithm is reached.

[0016] The specific process of step 3.1 is as follows: Step 3.1.1, for each incomplete task t By examining the initial scheduling scheme of the CAI phase Find the time intervals that have not yet been scheduled; Step 3.1.2 involves calculating the earliest and latest executable time slots for task t within a given time slot set to ensure that scheduling optimization does not violate the task's time window constraints. Earliest executable time slot. The calculation formula is shown in equation (22): (twenty two) in, The starting point for the task service window, For the task t The joint set of visible time slots across all antennas, This represents the smallest time slot index among all visible time slots. Take the larger of the two values ​​to ensure that both the service window constraint (3) and the visibility constraint (2) are satisfied; Latest feasible end time slot The calculation is shown in equation (23): (twenty three) in, This is the deadline for the task. This represents the index of the largest time slot among all visible time slots. This refers to the number of consecutive time slots required for a fixed-duration task. Fixed-duration tasks need to reserve [time slots]. The duration is long, therefore the latest starting point needs to be moved forward. Each time slot.

[0017] The beneficial effects of this invention are that by uniformly modeling heterogeneous antenna resources and multiple types of tasks, and employing a phased scheduling strategy, the scheduling process can achieve coordinated processing of task execution time, antenna occupancy, and bandwidth allocation under given constraints. This method is well-suited to application scenarios involving coexistence of single-access / multi-access antennas and a mixture of fixed-duration tasks and data-constrained tasks. It helps improve the matching of time slot resources and the feasibility of scheduling schemes, reduces scheduling failures caused by resource conflicts or window limitations, and thus enhances the overall stability and rationality of scheduling to a certain extent. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of a data relay satellite scheduling scenario in the relay satellite scheduling method of the present invention, which considers antenna resource heterogeneity and mission diversity. Figure 2 This is a time-slice modeling diagram of the relay satellite scheduling method considering antenna resource heterogeneity and task diversity in this invention. The horizontal axis represents the time slot number s within the scheduling period, the vertical axis represents different antenna resources, and the rectangles of different colors represent the execution time periods of different tasks. Figure 3 This is an example diagram of the scheduling results of a relay satellite scheduling method that considers antenna resource heterogeneity and task diversity in this invention, which includes SA antennas and MA antennas. In this scheduling result example, the horizontal axis represents the time slot number, the vertical axis shows the task allocation of SA antennas and the concurrent task allocation and bandwidth occupancy of MA antennas, and the rectangles of different colors represent the execution time periods of different tasks. Figure 4 A Gantt chart for scheduling 200 relay tasks. Detailed Implementation

[0019] The following detailed description is provided in conjunction with specific implementation methods.

[0020] Example 1 This invention provides a relay satellite scheduling method that considers antenna resource heterogeneity and mission diversity. The specific process is as follows: For single-site antennas in the relay satellite system ( SA) and multiple access antennas ( A unified resource model is used for SA antennas (SA antennas) to construct an antenna resource configuration model. The model sets the exclusive service attributes, fixed rated bandwidth, minimum handover preparation time, and number of cooling time slots for each SA antenna, and sets the total bandwidth capacity, maximum number of concurrent tasks, concurrent sub-channel structure, and handover latency for each MA antenna to support both single-task exclusive and multi-task parallel service modes.

[0021] To maximize the cumulative weight of mission completion, a data relay satellite mission scheduling model is constructed that simultaneously considers antenna heterogeneity, mission differences, and ground station coordination capabilities. Based on a time-slot discretization framework, the model divides the scheduling period into equal-length time slots. It accurately characterizes the mission-antenna usable area using service window indicators and visibility masks. Furthermore, it establishes multiple types of decision variables, including mission completion variables, antenna occupancy variables, time slot initiation variables, and bandwidth allocation variables, and sets up an objective function and feasibility constraints.

[0022] Example 2 Based on a mathematical programming model, a two-stage conflict-aware scheduling strategy with constrained local re-optimization is adopted. This generates a relay satellite scheduling scheme that includes all mission slot-level antenna allocation, bandwidth allocation, and mission completion status. The two-stage strategy includes a Conflict-Aware Initialization (CAI) phase oriented towards mission conflict intensity and a Constrained Local Re-optimization (CLR) phase that involves time-order compression, local shifting, and necessary mission replacement.

[0023] Example 3 A time-slot discretization model is adopted, introducing decision variables describing task completion status, antenna occupancy, start indication, and bandwidth allocation to construct a mathematical programming model with the objective of maximizing task weighted revenue. The model simultaneously constrains practical constraints such as the continuity of fixed-duration tasks, the cumulative transmission of data, the bandwidth and concurrency capabilities of SA / MA, antenna switching cooling, and single-link occupancy of user spacecraft. Based on this model, a two-stage conflict-aware and locally re-optimized scheduling algorithm (TSCA-CLR) is proposed: the first stage (CAI) generates an initial feasible solution based on window conflict degree and transmission upper bound; the second stage (CLR) implements time axis compression, restricted translation, bandwidth filling, and conservative replacement within the local interval of the task, and adopts an accept-rollback strategy to ensure monotonic improvement of the objective function. Simulation results verify that this method significantly improves task completion rate and cumulative weight under mixed task and heterogeneous resource conditions.

[0024] Example 4 This invention provides a relay satellite scheduling method that considers antenna resource heterogeneity and mission diversity, specifically including the following steps: Step 1: Discretize the scheduling period into equal-length time slots. Construct a time slot set S For any task t With antennaa Construct a visibility mask Service window mask ; Step 2: The conflict-aware initialization (CAI) strategy is used to solve equation (1) to generate an initial feasible scheduling scheme. The CAI phase used in Step 2 is built on task window conflict analysis, upper bound estimation of transportability, and difficulty-based scheduling priority construction. It is used to generate an initial scheduling scheme covering all schedulable tasks under the premise of satisfying all constraints.

[0025] Step 3 involves iteratively improving the initial scheme generated in the CAI stage using the Constrained Local Reoptimization (CLR) strategy to enhance task completion rate and objective function value. The core objective of the CLR stage is to explore hidden scheduling space and improve the overall task weighted benefit for tasks that are not completed or have underutilized resources in the initial scheme, within their schedulable time intervals, through operations such as timing adjustments, resource reallocation, and task replacement.

[0026] Example 5 The specific process of step 1 is as follows: Define decision variables Indicates task t Whether it is completed, if Indicates task t Completed, if Indicates task t Incomplete; Indicates antenna a In the time slot s Is it a service task? t ,like Indicates antenna a Service task in time slot s t ,like Indicates antenna a In the time slot s Non-service tasks t ; Indicates a fixed-duration task t Is it in the time slot? s For antenna a In the beginning, if Indicates a fixed-duration task t In the time slot s For antenna a In the beginning, if Indicates a fixed-duration task t Not in time slot s For antenna a start; Indicates antenna a In the time slot s Assigned to task t bandwidth; A mathematical programming model is constructed with the goal of maximizing the weighted sum of completed tasks. Equation (1) is the objective function, which aims to maximize the weighted sum of completed tasks. Equations (2)-(3) are basic feasibility constraints to ensure that tasks are executed only within the visible and permitted service window. Equations (4)-(8) are for fixed-duration tasks, which constrain their uniqueness of initiation, duration requirements, continuity, window selection, and consistency of completion. Equations (9)-(10) are for data-constrained tasks, which constrain their cumulative transmission volume and execution status. Equations (11)-(13) are for SA antennas, which constrain their exclusivity, switching cooling, and fixed bandwidth allocation. Equations (14)-(16) are for MA antennas, which constrain their total bandwidth capacity, upper limit of concurrency, and minimum bandwidth of fixed-duration tasks. Equation (17) is a single-link constraint for user spacecraft, which ensures that the same user occupies only one link in the same time slot.

[0027] (1) In equation (1), Indicates task t The weights; (2) Equation (2) represents the slot-level visibility constraint, that is, only when the task... t With antenna a Visible time slots s Talent allocation. This constraint is based on the geometrical visibility between the task and the antenna, orbital attitude conditions, and channel reachability criteria, and is enforced through... This ensures that the mission can only obtain antenna resource allocation within time slots that meet the requirements of unobstructed line of sight and effective propagation path.

[0028] (3) Equation (3) represents the service window constraint, that is, only when the task... t Time slots in the service window s This ensures that the scheduling time slots for tasks must strictly fall within the service window range defined by the task, thereby meeting the task's business timeliness, protocol requirements, or load operation constraints.

[0029] (4) Equation (4) indicates that each completed fixed-duration task has one and only one start slot, and this constraint forces each completed task to have one start slot. There exists exactly one time slot that serves as the starting point for its execution segment. This constraint guarantees the single-segment execution property of fixed-duration tasks, preventing them from being incorrectly split into multiple independent starting segments.

[0030] (5) Equation (5) indicates that a fixed-duration task must occupy [a certain amount of time] to complete. For each time slot, the cumulative length of the constraint-enforced segment must be equal to the task duration divided by the time slot length. This ensures that the task execution time is not compressed or truncated under discrete modeling, thereby maintaining the accuracy of the task time requirements.

[0031] (6) Equation (6) indicates that a fixed-duration task must be continuously occupied after the selected start time slot. Each time slot, this constraint ensures that task execution exhibits a strictly continuous interval pattern, avoiding internal gaps or breaks, and meeting the stringent requirements for link stability in tasks such as measurement and control, and communication.

[0032] (7) Equation (7) is used to constrain fixed-duration tasks to select at most one window among all available visible windows for execution. This constraint prevents such tasks from being incorrectly scheduled in multiple window segments and ensures that tasks are executed only within a single continuous time period.

[0033] (8) Equation (8) establishes a consistent relationship between the task completion status and window selection. If the task is completed, its window selection variable must be 1; if the window selection variable is not activated, the task cannot be considered completed.

[0034] (9) Equation (9) indicates that the cumulative amount of data transmitted for the completed quantity-constrained task must meet its data volume requirements. This constraint characterizes the segmentable execution characteristic of the task, allowing the task to accumulate bandwidth resources across multiple discontinuous windows, different antennas, or different link types, but requiring the final total transmission volume to reach the task objective, thereby ensuring data integrity and task effectiveness.

[0035] (10) Equation (10) is used to ensure data volume constraints in tasks. Antenna resources can only be used when the task is deemed complete; otherwise, all time slot resource allocations must be invalidated.

[0036] (11) Equation (11) indicates that each time slot of the SA antenna can serve at most one task. Since the SA antenna physically only supports a single-pointing link, only one task is allowed to occupy the antenna in the same time slot. This constraint strictly describes the exclusive resource attribute at the device level, effectively preventing link pointing conflicts or resource overload caused by the simultaneous allocation of multiple tasks.

[0037] (12) Equation (12) represents the cooling constraint for SA antenna switching. To switch the number of slots during cooling, no new tasks can be started during the cooling period; this constraint is enforced in the near future for the antenna. If a task is being executed within a time slot, a new task cannot be started in the current time slot.

[0038] (13) Equation (13) indicates that the SA antenna is allocated a fixed rated bandwidth. This constraint forces the SA antenna to operate at its rated bandwidth when the task is occupied. When providing services, bandwidth is not allowed to be dynamically adjusted within time slots based on task or environmental conditions.

[0039] (14) Equation (14) indicates that the total bandwidth allocated to each time slot of the MA antenna does not exceed its capacity. MA antennas can support concurrent tasks through beam multiplexing or code division multiplexing, but are still limited by total bandwidth capacity.

[0040] (15) Equation (15) indicates that the number of concurrent tasks per time slot of the MA antenna does not exceed the upper limit. This constraint reflects the upper limit of the number of independent beams or despreading channels of the MA antenna. It is a discrete capacity constraint of the concurrent structure and together with Equation (11) determines the availability of the time slot of the MA.

[0041] (16) Equation (16) indicates that the antennas allocated for fixed-duration tasks must meet the minimum bandwidth requirements. This satisfies its requirements for link continuity, stability, and minimum service quality.

[0042] (17) Equation (17) indicates that a single user spacecraft can only occupy one relay or direct transmission link in any time slot. Due to the limited capabilities of the spacecraft's radio frequency link front end, pointing mechanism, and power allocation, this constraint prohibits it from using multiple antennas for parallel transmission at the same time, thereby avoiding resource conflicts and maintaining the consistency of on-board link execution.

[0043] The relevant parameters in the above formula are explained as follows: Scheduling slot length This serves as a unified minimum time granularity, used to discretize a continuous time axis into equally spaced time slots; time slot set , of which elements s Indicates the first sTime slots; task set It contains all relay transmission tasks to be scheduled, each with an independent window and data requirements; a fixed-duration task set. The task must be executed completely in a continuous time slot and cannot be segmented; the data volume constrains the task set. The task can be executed in segments within the window, and it is considered complete when the cumulative transmission volume reaches the target.

[0044] User spacecraft collection Each task t Corresponding to a unique user spacecraft .

[0045] Antenna assembly It includes all available relay and ground station antenna resources. Single-site antenna set. A single-address antenna can only serve one task in any given time slot.

[0046] Multiple access antenna array Multiple access antennas can serve multiple tasks in parallel within the same time slot.

[0047] Task weight Reflecting the task t The relative importance and target benefit are used as weighting coefficients in the objective function.

[0048] Task Service Window Limited task t The range of executable start and end time slots.

[0049] Window indicator variable When the gap s If the value falls into the task window, use 1; otherwise, use 0.

[0050] Visibility parameters Taking 1 indicates a task t With antenna a In the time slot s A link can be established; otherwise, the value is 0.

[0051] Task – Antenna Service Time Slot Set , indicating task t It can be in the antenna a The set of time slot indexes executed on the above.

[0052] Task-level serviceable time slot set For the task t Jointly visible time slots on all antennas.

[0053] The earliest feasible start time slot and the latest feasible end time slot of the task , respectively The minimum and maximum elements.

[0054] Task t With antenna a Visible window index set This is used to enumerate consecutive visible time periods.

[0055] No. j Visible window , indicating task t With antenna a The continuous visible time interval.

[0056] The actual duration of a fixed-duration task The unit is seconds, used to calculate the corresponding number of discrete time slots.

[0057] Number of consecutive time slots required for a fixed-duration task ,for Divide by the time slot length Δ and round up.

[0058] Minimum data volume for data volume tasks When this value is reached, the task is considered complete.

[0059] Minimum data rate required for fixed-duration tasks under multiple access antennas This is used to ensure link quality.

[0060] antenna a Total bandwidth capacity This indicates the maximum bandwidth it can provide within any given time slot.

[0061] Multiple access antenna a Maximum number of concurrent tasks This limits the number of tasks that can be served simultaneously in the same time slot.

[0062] antenna a Switchover preparation time This refers to the shortest configuration time to switch from one task to the next.

[0063] antenna a number of cooling time slots This indicates the minimum time interval that must be retained after a task switch.

[0064] Example 6 The specific process of step 2 is as follows: Step 2.1, for each task t With antenna a The conflict degree for each visible window w is calculated using the following formula. First, for each task... t With each antenna a Calculate its value across all visible windows. wConflict level in the context of conflict, conflict level conf( w As shown in equation (18): (18) in, W a Indicates antenna a The set of all visible windows Representation and window w This is a different window from others. The conflict level metric reflects the temporal overlap between the task and antenna windows; a higher value indicates stronger resource contention for the task-antenna pair within that time interval. In practical applications, high-conflict task pairs often require priority scheduling to avoid failing to meet their execution needs in later scheduling phases.

[0065] Step 2.2, all fixed-duration tasks Sort tasks by conflict level in descending order based on their highest conflict level window, and then find and assign consecutive time slots that satisfy the constraints. All fixed-duration tasks According to its collision degree value conf( ) in the maximum collision degree window of different antennas w The tasks are sorted in descending order. The priority of the sorted tasks determines their processing order in subsequent scheduling. The specific scheduling strategy is as follows: (1) For each task t First, find the conflict value of the task in the maximum conflict window, and then sort all tasks in descending order of this value.

[0066] (2) For each sorted task, the algorithm sequentially searches for consecutive time slots that satisfy the following constraints: A) The time continuity of fixed-duration tasks, that is, the task must be executed continuously on the selected time slot; B) Time slot visibility constraints, i.e., task t With antenna a In a certain time slot s It must be visible; C) Service window constraint, meaning that the task must be executed within its predefined service window; D) Antenna switching cooling constraint, i.e., at least [amount missing] must be maintained after task switching. Cooling time for each time slot.

[0067] E) Concurrency limit: Within the same time slot, the MA antenna can only have a maximum of [number missing]. Each task provides services.

[0068] The algorithm finds antennas and consecutive time slots that satisfy all constraints, including time continuity, time slot visibility, service window, handover cooling, and concurrency limits, for each fixed-duration task, and completes resource allocation. The scheduling process prioritizes tasks with higher conflict levels to ensure the timely execution of high-priority tasks.

[0069] Step 2.3, calculate the time slot for all data volume constraints using the following formula. s The algorithm sets an upper bound on the theoretical maximum transmittable data volume. The algorithm constrains the task for all data volumes. Process it. For each task. t Based on its earliest feasible start time slot The theoretical maximum transmission upper bound starting from this time slot is calculated as shown in equation (19): (19) in, Indicates the start time slot The initial theoretical maximum transmission upper bound, This indicates utilizing the remaining bandwidth, i.e., in the time slot. r Inside, the mission t The maximum available bandwidth is the maximum bandwidth visible among all antenna resources that is not occupied by other tasks.

[0070] Calculated The upper bound of the transmission potential of each data volume-constrained task within its visible time slots is a key indicator for resource allocation. This upper bound provides a theoretical basis for subsequent task prioritization and bandwidth allocation, ensuring that high-demand tasks can obtain resources preferentially when bandwidth is insufficient.

[0071] Step 2.4, calculate the data volume constraints for each task. Completion difficulty indicators This metric is used to evaluate the task. t The level of difficulty in completing the task given time slots and resources; completion difficulty index. As shown below: (20) in, Indicates task t Data volume requirements. Difficulty of completion. The size is directly related to the task's resource requirements and available bandwidth. The smaller the value, the greater the likelihood of task completion. Therefore, tasks will be prioritized based on this metric, with more difficult tasks scheduled in time slots with more abundant resources.

[0072] Step 2.5: Sort all tasks and allocate bandwidth accordingly. The sorting rule is: first, based on the difficulty of completing the tasks. Tasks are sorted in ascending order, with easier tasks prioritized; then, if tasks are of equal difficulty, their deadlines are determined by the deadlines they submit. Sort tasks in ascending order to ensure that time-sensitive tasks are executed first; finally, if tasks have the same difficulty and deadline, then prioritize them based on their weight. The tasks are sorted in descending order, prioritizing high-value tasks. Based on the sorting results, the algorithm employs a greedy sweep strategy for bandwidth allocation. In each bandwidth allocation, dynamic bandwidth allocation is performed according to equation (21): (twenty one) in, Assigned to task t Bandwidth on time slot s, It is a task t In the time slot s Initial remaining data requirements, It is an antenna. a In the time slot s Allocated bandwidth. Dynamic bandwidth allocation ensures that tasks are completed to the maximum extent possible within resource constraints.

[0073] Step 2.6: Record the current scheduling scheme as the initial feasible solution. This solution is a complete scheduling plan derived from the resource allocation and scheduling results of the CAI phase, including information such as the start time slot, duration, bandwidth allocation, antenna allocation, and task completion status for all tasks. This initial feasible solution will be locally optimized in the subsequent CLR phase to improve task completion rate and system throughput.

[0074] Example 7 The specific process of step 3 is as follows: Step 3.1: For each task that remains unfinished after the CAI phase. t calculate This constitutes the optimization interval, as detailed below: First, for each task that was not completed after the CAI phase... t Calculate its earliest feasible start time slot. Gap with the latest feasible end And based on this, define its optimization interval. The optimization interval is the task. t The range of time slots that can still be scheduled within a given time frame. The specific calculation process is as follows: (1) For each unfinished task t By examining the initial scheduling scheme of the CAI phase Find the time interval in which it has not yet been scheduled; (2) Through calculation tasks tGiven the earliest and latest executable time slots in the time slot set, ensure that scheduling optimization does not violate the task's time window constraints. Earliest executable time slot The calculation formula is shown in equation (22): (twenty two) in, The starting point for the task service window, For the task t The set of joint visible time slots across all antennas, This represents the smallest time slot index among all visible time slots. Take the larger of the two values ​​to ensure that both the service window constraint (3) and the visibility constraint (2) are satisfied.

[0075] Latest feasible end time slot The calculation is shown in equation (23): (twenty three) in, This is the deadline for the task. This represents the index of the largest time slot among all visible time slots. This refers to the number of consecutive time slots required for a fixed-duration task. Fixed-duration tasks need to reserve [time slots]. The duration is long, therefore the latest starting point needs to be moved forward. Each time slot. The interval calculated using equations (22)-(23) Strictly limited to the intersection of the task service window and the antenna visibility window, this ensures that all local optimization operations do not violate the time window constraint, thereby avoiding redundant calculations and improving scheduling efficiency.

[0076] Step 3.2, in the interval The process sequentially executes time-order compaction, restricted local translation, and bandwidth padding for tasks with unfinished data volume constraints: (1) Timing compression operation, which improves resource utilization efficiency by moving the scheduled task segments to the left within the interval, reducing the time slot gap. The compression operation ensures that the scheduling time of the task is shortened as much as possible without violating the task service window and other constraints, and reducing idle time slots.

[0077] (2) Restricted local translation operation, which allows task segments to be translated to a limited extent left and right between time slots within an interval in order to free up more contiguous resources and avoid excessive fragmentation. The restrictions on local translation ensure that the task does not exceed its available time window and does not affect the scheduling of other tasks.

[0078] (3) Bandwidth compensation operation, for data volume constrained tasks Bandwidth compensation ensures that during task execution, especially with the remaining data volume... In cases where data transmission is incomplete, bandwidth resources are dynamically supplemented to accelerate task completion. The supplementation operation calculates the required bandwidth increment based on the remaining data volume of the task and the available bandwidth, and allocates sufficient bandwidth to the task until it is completed.

[0079] Step 3.3, if the data volume is a constraint on the task If not yet completed, tasks will be assigned according to their weight. Remove low-weight, fixed-duration tasks that have already been scheduled within the ascending order of the interval. The complete segment. The goal of this operation is to free up sufficient time and bandwidth resources for unfinished, high-weight tasks. The specific steps are as follows: (1) By task weight Sort all tasks in ascending order, and first remove the fixed-duration task segment with the lowest weight.

[0080] (2) After each task segment is removed, try to assign the current high-weight tasks to idle or released resources to ensure that these tasks are completed as soon as possible.

[0081] Step 3.4: Establish an accept-rollback mechanism. After each timing compaction, restricted shift, or bandwidth compensation operation in Step 3.2, or each task replacement operation in Step 3.3, immediately check whether the operation has improved the scheduling scheme. The specific judgment process is as follows: Before executing any local operation, save the scheduling status and objective function value of all tasks in the current interval, including whether each task is completed, which antennas are used in which time slots, how much bandwidth is allocated, and all other information. After executing the operation, recalculate the total weight of all completed tasks and compare the new total weight with the value before the operation. If the new value is greater than the old value, it means that the objective value has been improved. At this time, accept the modification, retain all scheduling changes after the operation, and use this as the basis for the next optimization. If the new value is less than or equal to the old value, it means that the operation has failed to improve the scheduling effect. At this time, roll back the modification, that is, restore the scheduling status of all tasks in the current interval to the state saved before the operation was executed, discard all changes. Rollback is not re-executing the operation but undoing the result of the operation. The interval referred to here is the range of the earliest and latest feasible execution time slots calculated in step 3.1 for the task currently being optimized. All modifications refer to the adjustments made in step 3.2 or 3.3 to decisions regarding antenna allocation, time slot occupancy, bandwidth allocation, etc., for tasks within this interval. This accept-rollback mechanism ensures that each retained modification improves the quality of the scheduling scheme through a sequential operation and judgment approach, thereby guaranteeing that the objective function value monotonically increases during the optimization process and avoiding performance degradation due to invalid or inappropriate adjustments.

[0082] Step 3.5: Repeat steps 3.1–3.4 until the maximum number of iterations set by the algorithm is reached. Each round of optimization aims to improve task completion rate and bandwidth utilization, and prioritizes critical tasks based on their urgency and weight. The algorithm's ultimate goal is to output the optimal scheduling scheme. The scheme includes the final scheduling time, antenna allocation, bandwidth allocation, and task completion status for all tasks.

[0083] The application process of the data relay satellite scheduling method described in this invention will be explained below in a specific scenario. This embodiment includes 200 relay missions, 100 user spacecraft, 1 first-generation data relay satellite, 2 second-generation data relay satellites, and 2 ground stations, with a scheduling cycle of 24 hours. A partial relay mission list is shown in Table 1.

[0084] In this scenario, the first-generation relay satellite is equipped with one single-site antenna with a rated bandwidth of 40 Mbps and a handover preparation time of 240 s. Each second-generation relay satellite is equipped with two single-site antennas (40 Mbps, 240 s) and one multiple-site antenna (total bandwidth 20 Mbps, concurrent connection limit of 5, handover time of 5 s). Two ground stations are each equipped with two single-site receiving antennas with a bandwidth of 40 Mbps and a handover time of 120 s.

[0085] Table 1 Relay Task Table

[0086] According to the method of the present invention, the entire scheduling cycle is first discretized with a time slot length of 20 s to construct the visibility mask and service window mask for all tasks, and to establish the capacity, handover, and concurrency constraints of single-access / multi-access antennas. The 200 tasks are then classified by task type, where fixed-duration tasks must be completed on consecutive time slots, and data-constrained tasks can be dynamically transmitted according to bandwidth within their respective windows.

[0087] Subsequently, collision-aware initialization (CAI) is performed. The collision degree is calculated for each task within the antenna's visibility window. Fixed-duration tasks are sorted in descending order of maximum window collision degree and time slots are allocated sequentially while satisfying visibility, continuity, bandwidth, and handover constraints. After the fixed-duration tasks are assigned, the theoretical maximum transmission capacity upper bound from the earliest starting time slot is calculated for all data-constrained tasks, and they are sorted based on completion difficulty, deadline, and weight. The algorithm performs a bandwidth sweep allocation for these tasks within each antenna's visible time slot and generates an initial feasible scheduling scheme.

[0088] Based on the initial scheme, Constrained Local Re-optimization (CLR) is further performed. For tasks not completed in the CAI phase, their respective schedulable intervals are calculated, and timeline compaction, constrained shifting, and bandwidth padding operations are implemented within these intervals. For data-heavy tasks that still cannot be completed, some consecutive segments of fixed-duration tasks are removed according to the principle of prioritizing low-weight tasks to attempt to release resources, and an accept-rollback mechanism is used to ensure improvement of the target value. After processing all incomplete tasks, the final scheduling scheme of this embodiment is obtained.

[0089] Figure 4 To schedule the Gantt chart, this scheme enables the effective scheduling of most tasks in scenario T01, simultaneously meeting the continuous execution requirements of fixed-duration tasks and the multi-link segmented transmission of data-volume tasks. It also rationally allocates resources between single-address / multi-address antennas, demonstrating the adaptability of the method of this invention to heterogeneous antennas and diverse task conditions.

Claims

1. A relay satellite scheduling method considering antenna resource heterogeneity and mission diversity, characterized by: Specifically, the steps include the following: Step 1: Discretize the scheduling period into equal-length time slots. Construct a time slot set S For any task t With antenna a Construct a visibility mask Service window mask ; Step 2: Generate an initial feasible scheduling scheme using a conflict-aware initialization strategy; Step 3: The initial scheme generated in the CAI stage is iteratively improved by adopting a constrained local re-optimization strategy, thereby improving the task completion rate and the objective function value.

2. The relay satellite scheduling method considering antenna resource heterogeneity and mission diversity according to claim 1, characterized in that: The specific process of step 1 is as follows: Step 1.1, determine the parameter variables; Step 1.2: Construct a mathematical programming model based on the parameter variables determined in Step 1.1, with the goal of maximizing the weighted sum of tasks completed.

3. The relay satellite scheduling method considering antenna resource heterogeneity and mission diversity according to claim 2, characterized in that: The specific process of step 1.1 is as follows: Define decision variables. , indicating task t Whether it is completed, if Indicates task t Completed, if Indicates task t Incomplete; Indicates antenna a In the time slot s Is it a service task? t ,like Indicates antenna a Service task in time slot s t ,like Indicates antenna a In the time slot s Non-service tasks t ; Indicates a fixed-duration task t Is it in the time slot? s For antenna a In the beginning, if Indicates a fixed-duration task t In the time slot s For antenna a In the beginning, if Indicates a fixed-duration task t Not in time slot s For antenna a start; Indicates antenna a In the time slot s Assigned to task t The bandwidth.

4. The relay satellite scheduling method considering antenna resource heterogeneity and mission diversity according to claim 3, characterized in that: The specific process of step 1.2 is as follows: The objective function of the mathematical programming model is constructed using the following formula (1): （1） In equation (1), Indicates task t The weights; （2） Equation (2) represents the slot-level visibility constraint: （3） Equation (3) represents the service window constraint: （4） Equation (4) indicates that each completed fixed-duration task has one and only one start time slot; （5） Equation (5) indicates that a fixed-duration task must occupy [a certain amount of time] to complete. Each time slot; （6） Equation (6) indicates that a fixed-duration task must be continuously occupied after the selected start time slot. Each time slot; （7） Equation (7) is used to constrain fixed-duration tasks to select at most one window from all available visible windows for execution; （8） Equation (8) establishes a consistent relationship between the task completion status and window selection. If the task is completed, its window selection variable must be 1; if the window selection variable is not activated, the task cannot be considered completed. （9） Equation (9) indicates that the cumulative amount of data transmitted for the completed quantity constraint task must meet the data volume requirement. ; （10） Equation (10) is used to ensure data volume constraints in the task. Antenna resources can only be used when the task is determined to be completed; otherwise, all time slot resource allocations must be invalidated. （11） Equation (11) indicates that each time slot of the SA antenna can serve at most one task; （12） Equation (12) represents the cooling constraint for SA antenna switching. To switch the number of cooldown slots, no new tasks can be started during the cooldown period; （13） Equation (13) indicates that the SA antenna is allocated a fixed rated bandwidth. ; （14） Equation (14) indicates that the total bandwidth allocated to each time slot of the MA antenna does not exceed its capacity. ; （15） Equation (15) indicates that the number of concurrent tasks per time slot of the MA antenna does not exceed the upper limit. ; （16） Equation (16) indicates that the antennas allocated for fixed-duration tasks must meet the minimum bandwidth requirements. ; （17） Equation (17) indicates that the same user spacecraft can only occupy one relay or direct transmission link in any time slot.

5. The relay satellite scheduling method considering antenna resource heterogeneity and mission diversity according to claim 4, characterized in that: In step 1.2, the scheduling time slot length This refers to a unified minimum time granularity; time slot set. , of which elements s Indicates the first s Time slots; task set It contains all relay transmission tasks to be scheduled; a fixed-duration task set. The task must be executed completely in a continuous time slot and cannot be segmented; the data volume constrains the task set. The task can be executed in segments within the window, and it is considered complete when the cumulative transmission volume reaches the target. User spacecraft collection Each mission t corresponds to a unique user spacecraft. ; Antenna assembly It includes all available relay and ground station antenna resources; single-site antenna collection A single-address antenna can only serve one task in any given time slot; Multiple access antenna array Multiple access antennas can serve multiple tasks in parallel within the same time slot; Task weight Reflecting the task t The relative importance and target benefit are used as weighting coefficients in the objective function; Task Service Window Limited task t The range of executable start and end time slots; Window indicator variable When the gap s The value is 1 if the task falls into the task window, otherwise it is 0. Visibility parameters Taking 1 indicates a task t With antenna a In the time slot s A link can be established; otherwise, the value is 0. Task – Antenna Service Time Slot Set , indicating task t It can be in the antenna a The set of time slot indexes executed on the current execution; Task-level serviceable time slot set For the task t Joint visible time slots on all antennas; The earliest feasible start time slot and the latest feasible end time slot of the task , respectively The minimum and maximum elements; Set of visible window indices for task t and antenna a ; No. j Visible window , indicating task t With antenna a The continuous visible time interval; The actual duration of a fixed-duration task The unit is seconds, used to calculate the corresponding number of discrete time slots; Number of consecutive time slots required for a fixed-duration task ,for Divide by the time slot length Δ and round up.

6. The relay satellite scheduling method considering antenna resource heterogeneity and mission diversity according to claim 5, characterized in that: The specific process of step 2 is as follows: Step 2.1, for each task t With each antenna a Calculate all visible windows w Conflict level in the context of conflict, conflict level conf( w As shown in equation (18): （18） in, W a Indicates antenna a The set of all visible windows Representation and window w Different from other windows; Step 2.2, all fixed-duration tasks Sort the windows by conflict degree in descending order, find and assign consecutive time slots that satisfy the constraints in turn; Step 2.3, for each task t According to the earliest feasible start time slot The theoretical maximum transmission upper bound starting from this time slot is calculated as shown in equation (19): （19） in, Indicates the start time slot The initial theoretical maximum transport upper bound, This indicates utilizing the remaining bandwidth, i.e., in the time slot. r Inside, the mission t The maximum usable bandwidth is the maximum bandwidth visible among all antenna resources that is not occupied by other tasks. Step 2.4, calculate the data volume constraints for each task. Completion difficulty indicators The calculation formula is as follows: （20） in, Indicates task t Data volume requirements; Step 2.5: Sort all tasks and allocate bandwidth in sequence; Step 2.6: Record the current scheduling scheme as the initial feasible solution. .

7. The relay satellite scheduling method considering antenna resource heterogeneity and mission diversity according to claim 6, characterized in that: The specific process of step 2.5 is as follows: First, based on the difficulty of completing the task. Tasks are sorted in ascending order, with easier tasks prioritized; then, if tasks are of equal difficulty, their deadlines are determined by the deadlines they submit. Sort tasks in ascending order to ensure that time-sensitive tasks are executed first; finally, if tasks have the same difficulty and deadline, then prioritize them based on their weight. The tasks are sorted in descending order, prioritizing high-value tasks. Based on the sorting results, a greedy sweep strategy is used for bandwidth allocation. In each bandwidth allocation, dynamic bandwidth allocation is performed according to equation (21): （21） in, Assigned to task t Bandwidth on time slot s, It is a task t In the time slot s Initial remaining data requirements, It is an antenna. a In the time slot s Already allocated bandwidth.

8. The relay satellite scheduling method considering antenna resource heterogeneity and mission diversity according to claim 7, characterized in that: The specific process of step 3 is as follows: Step 3.1: For each task that remains unfinished after the CAI phase. t calculate This constitutes the optimization interval; Step 3.2, in the interval The operation sequentially executes time-order compaction, restricted local translation, and bandwidth filling for tasks with unfinished data volume constraints. Step 3.3, if the data volume is a constraint on the task If not yet completed, tasks will be assigned according to their weight. Remove low-weight, fixed-duration tasks that have already been scheduled within the ascending order of the interval. The complete paragraph; Step 3.4: Establish an accept-rollback mechanism; Step 3.5: Repeat steps 3.1 to 3.4 until the maximum number of iterations set by the algorithm is reached.

9. The relay satellite scheduling method considering antenna resource heterogeneity and mission diversity according to claim 8, characterized in that: The specific process of step 3.1 is as follows: Step 3.1.1, for each incomplete task t By examining the initial scheduling scheme of the CAI phase Find the time intervals that have not yet been scheduled; Step 3.1.2: By calculating the earliest and latest executable time slots of task t within the given time slot set, ensure that scheduling optimization does not violate the task's time window constraints, and determine the earliest executable time slot. The calculation formula is shown in equation (22): （22） in, The starting point for the task service window, For the task t The set of joint visible time slots across all antennas, This represents the smallest time slot index among all visible time slots. Take the larger of the two values ​​to ensure that both the service window constraint (3) and the visibility constraint (2) are satisfied; Latest feasible end time slot The calculation is shown in equation (23): （23） in, This is the deadline for the task. This represents the index of the largest time slot among all visible time slots. To determine the number of consecutive time slots required for fixed-duration tasks, fixed-duration tasks need to reserve [reservations / slots]. The duration is long, therefore the latest starting point needs to be moved forward. Each time slot.

Images