A cross-domain task-driven giant star constellation gateway dynamic deployment and scheduling method and system

Abstract

The application discloses a kind of cross-domain task-driven giant star constellation gateway dynamic deployment scheduling method and system, belong to satellite communication field.Pass through obtaining the satellite orbit data in heterogeneous function domain and inter-domain geometric visible relation, filter candidate cross-domain window and combine task demand to carry out value evaluation, construct high-value candidate window set;Cross-domain link configuration scheme is generated using the set, check exclusive and preparation time constraint, determine the satellite node of activated gateway and link activation timing, under satisfying task constraint, gateway activation overhead is optimized and reduced;In response to the task, first search transmission path in reserved capacity to carry out direct insertion scheduling, if fail, then preferentially recover and restore flow resource, then based on preemption loss score screening controlled preemption object release resource to complete scheduling, and subsequent recovery preoccupied task.The application solves the problem that cross-domain task and gateway resource are difficult to adapt in giant star constellation dynamic environment, and realizes efficient and low-overhead gateway dynamic deployment.

Description

Technical Field

[0001] This invention relates to the field of satellite communications, and in particular to a cross-domain mission-driven dynamic deployment and scheduling method and system for the Giants Gateway. Background Technology

[0002] In existing satellite communication systems, gateways are typically considered fixed infrastructure deployed on the ground. In non-terrestrial network architectures, satellites connect to ground gateways via feeder links, with the ground gateways handling functions such as protocol access, data aggregation, routing, and network-side control. Existing technologies centered around this ground gateway primarily focus on three areas. The first is feeder link reliability enhancement and disaster recovery solutions, improving the availability and system continuity of high-frequency feeder links through site diversity, multi-gateway redundancy, or gateway switching mechanisms. The second is static site selection and deployment topology optimization schemes for ground gateways, achieving optimal ground gateway layouts under constraints such as construction cost, coverage capacity, and load balancing by constructing combinatorial optimization or mixed-integer programming models. The third is on-orbit data forwarding schemes that alleviate the pressure on ground gateways using inter-satellite links, relaying service data to satellite nodes visible to ground gateways via inter-satellite links within the same constellation, thereby reducing dependence on a single gateway location. However, a common characteristic of these existing technologies is that gateway functions are inherently dependent on ground physical facilities, or only perform inter-satellite relay forwarding within a single constellation, with cross-domain interconnection functions still primarily relying on ground gateways. Currently, there is no established deployment method for cross-domain task transmission between heterogeneous functional domains, where ordinary satellites can dynamically assume the role of cross-domain gateways and coordinate scheduling as needed. Summary of the Invention

[0003] The present invention aims to at least partially solve one of the technical problems in the related art.

[0004] To address this, this invention proposes a cross-domain task-driven dynamic deployment and scheduling method for the Giants Gateway. By acquiring satellite orbits and inter-domain geometric relationships across heterogeneous domains, candidate windows are filtered and their value is evaluated based on task requirements to construct a high-value window set. This set is then used to generate link configuration schemes, verify constraints to determine activation nodes and timing, and optimize to reduce activation overhead. In response to ad hoc tasks, direct insertion is first performed within reserved capacity; if this fails, recovery stream resources are prioritized for reclamation. Then, based on preemption loss scores, controlled preemption objects are selected to release resources, completing scheduling and subsequent recovery. This invention achieves dynamic and efficient deployment of the Giants Gateway across domains.

[0005] Another objective of this invention is to propose a cross-domain task-driven dynamic deployment and scheduling system for the Giants Gateway.

[0006] The third objective of this invention is to provide a computer device.

[0007] To achieve the above objectives, this invention proposes a cross-domain task-driven dynamic deployment and scheduling method for the Giants Gateway, comprising:

[0008] Acquire satellite orbit data and geometric visibility relationships between heterogeneous functional domains, screen candidate cross-domain windows that meet the minimum effective link establishment time constraint, evaluate the value of candidate cross-domain windows based on the needs of routine missions and the needs of hotspot areas, and construct a set of high-value candidate windows. A cross-domain link configuration scheme is generated by using a high-value candidate window set. By simulating the access order of the windows and verifying the cross-domain terminal exclusive constraints and device preparation time constraints, the satellite nodes that activate the cross-domain gateway role and the activation sequence of each cross-domain link are determined. Under the premise of meeting the constraints of normal task completion rate and cross-domain capability gap, the configuration scheme is optimized to reduce gateway activation overhead, so as to obtain the cross-domain link configuration scheme and the corresponding normal task baseline scheduling scheme. In response to online incoming ad hoc tasks, search for end-to-end transmission paths within the capacity reserved by the cross-domain link configuration scheme to perform direct insertion scheduling; if direct insertion cannot meet the transmission requirements of ad hoc tasks, prioritize reclaiming the resources occupied by the recovery stream previously generated due to preemption, and retry direct insertion. If the transmission demand is still not met after the recovery stream is recovered, then based on the preemption loss score, controlled preemption objects are selected from low-priority normal tasks that have resource conflicts with the transmission path of the ad hoc task. The resources of the corresponding controlled preemption objects are released to complete the ad hoc task scheduling, and the idle capacity is used to perform recovery scheduling for the preempted normal tasks in the subsequent time window.

[0009] In one embodiment of the present invention, a value assessment of candidate cross-domain windows is performed based on the requirements for normal task carrying and the requirements for hotspot area protection, and a set of high-value candidate windows is constructed, including: For any active routine task within any time segment of the planning period, calculate the corresponding demand intensity, which is the smaller of the task's maximum available rate and the amortized rate calculated based on the remaining time and remaining data volume. Calculate the reachability weight of the candidate window for the routine task based on the shortest access hops from the source node in the source functional domain to the satellite at one end of the window, and the shortest access hops from the satellite at the other end of the window in the target functional domain to the destination node. Accumulate the product of the task weight, demand intensity, and reachability weight over the planning period according to the time overlap duration to obtain the static value of the candidate window for the routine task. For a pre-defined set of hotspot areas, if a satellite at one end of a candidate window can be accessed by a satellite within a hotspot area within a given time segment without exceeding the hop count limit, then the physical visibility duration of the window, the hotspot area weight, and the normalized coverage indicator within the corresponding time segment are multiplied to obtain the local contribution of the window to the hotspot area in the corresponding segment; the local contributions of the window to the hotspot area in all segments within the planning period are summed to obtain the static value of the corresponding candidate window for hotspot protection. Based on the static value of routine tasks and the static value of hotspot support, windows with higher comprehensive value are extracted first to form a global master set. Then, the number of windows in the global master set within each demand time period is checked. For time periods with insufficient windows, windows are supplemented from the remaining windows according to the local value of the corresponding time period. Finally, a set of high-value candidate windows for subsequent scheduling is constructed.

[0010] In one embodiment of the present invention, a cross-domain link configuration scheme is generated using a high-value candidate window set, including: The attempts of windows in the high-value candidate window set are added to the sequence encoding as chromosomes using the permutation encoding method, with one chromosome corresponding to one permutation of the corresponding set; Candidate windows are added to the configuration scheme sequentially according to the chromosome arrangement order. For each candidate window... Obtain the most recent terminal occupancy end time before the end time of the window for both satellites, and calculate the earliest feasible open-chain time using the following formula, combined with the equipment preparation time:

[0011] in, For window The physical visible start time, For window The physical visible end time, For window The A-end satellite in The last time the terminal was occupied ended. For window B-end satellite in The last time the terminal was occupied ended. Allow time for equipment preparation; if If the condition is met, the corresponding window will be accepted and the terminal occupancy status of both satellites will be updated; otherwise, the corresponding window will be rejected. The minimum effective chain establishment time threshold; During the gradual addition of windows, the routine task completion rate, routine cross-domain capability gap and average guarantee level of hotspot areas are calculated for the prefix schemes formed by the currently accepted windows. When all three indicators reach the preset threshold, the addition of windows is stopped, and the initial feasible scheme is obtained. Perform reverse redundancy pruning on the initial feasible schemes, and try to delete them sequentially starting from the last window of the scheme. If the normal task completion rate, normal cross-domain capability gap and average guarantee level of hot spot areas still meet the preset threshold after deletion, the corresponding window is permanently removed; otherwise, the corresponding window is retained. After traversal, a simplified cross-domain link configuration scheme is obtained.

[0012] In one embodiment of the present invention, searching for an end-to-end transmission path within the capacity reserved in the cross-domain link configuration scheme to perform direct insertion scheduling includes: An emergency time domain is constructed based on the arrival and end times of the emergency task, and the emergency time domain consists of all segments falling within the time interval. Without changing the committed normal business allocation, read the capacity status on each segment and generate cross-domain end-to-end candidate paths for ad hoc tasks; for each candidate path in a segment, calculate the candidate rate that can be allocated to ad hoc tasks. The candidate rate is the minimum of the ad hoc task's maximum available rate, the minimum remaining capacity on each side of the path, and the average rate of the ad hoc task's remaining data volume in this segment. The total capacity of cross-domain links is divided into a two-layer structure: the capacity available for normal tasks and the capacity reserved for ad hoc tasks. The scheme that can complete the transmission by using the reserved capacity for ad hoc tasks is given priority. If it is necessary to borrow the idle capacity of normal tasks, the scheme with the smallest amount of borrowing is selected. A multi-label dynamic programming algorithm is used to perform a segmented progressive search in the emergency time domain. Each label records the current remaining data volume, the cross-domain link used in the previous segment, and the cumulative cost. Non-dominated pruning is performed on the labels generated in each segment, and only non-dominated labels are retained to enter the next segment. If a path scheme that can complete all data transmission before the deadline is found, the path scheme is written into the current committed scheduling table to complete direct insertion scheduling.

[0013] In one embodiment of the present invention, when direct insertion search cannot complete all ad hoc task data transmission before the deadline, the method further includes: Identify that the recovery flow in the current committed scheduling table belongs to a recovery flow that was previously preempted and then reallocated, and that the resources occupied by the recovery flow conflict with the near-optimal candidate path of the current ad hoc task; Cancel the resource allocation of the conflict recovery flow on the conflict segment, put the normal task corresponding to the resource allocation back to the pending recovery state, and update the capacity view; In the updated capacity view, re-execute the direct insert search on the current ad hoc task; if the transmission completion rate after re-insertion is improved compared to before reclamation, retain the corresponding reclamation action and output the updated scheduling scheme; otherwise, roll back to the state before reclamation.

[0014] In one embodiment of the present invention, filtering controlled preemption targets based on preemption loss scores includes: Candidate corridors are constructed based on the near-optimal candidate paths generated during the direct insertion phase of the emergency task. The candidate corridors are formed by the union of intra-domain edges and cross-domain edges contained in each segment of the emergency time domain, where the transmittable amount is not less than a preset proportion of the transmittable amount of the optimal path in the corresponding segment. From routine tasks that have lower priority than ad hoc tasks and whose business time domain overlaps with the emergency time domain of ad hoc tasks, tasks whose occupied edge resources and candidate corridors have actual conflicts in time and space are selected to form a set of preemption candidate tasks. For each preemption candidate task Calculate the preemption loss score using the following formula:

[0015] in, For the task Normalized task weights For the task The normalized value of the recovery margin, For the task The recoverability normalized value, For the task The normalized value of the effective release of ad hoc tasks after being preempted. , , For preset weighting coefficients, To prevent extremely small positive numbers from being divided by zero; All candidate tasks for seizing are sorted in ascending order of their seizing loss score to form a candidate sequence for seizing. The lower the score, the higher the overall benefit of releasing the corresponding task.

[0016] In one embodiment of the present invention, after forming the preemption candidate sequence, the method further includes: According to the preemption candidate sequence, the resources occupied by individual preemption candidate tasks in the conflict segment are simulated and released in turn. Then, under the updated capacity view, the direct insertion search is re-executed for the contingent task to obtain the simulated transmission scheme. The additional transferable amount of the simulated transfer scheme for ad hoc tasks is calculated compared to the backup scheme without preemption, and the expected recoverable amount of data for released normal tasks in the remaining time domain is evaluated. Preemption of the corresponding candidate task is confirmed only when the additional transmittable amount exceeds the preset gain threshold. The resource occupation of the corresponding candidate task in the conflict segment is officially revoked, the released resources are allocated to the current ad hoc task, and the preempted normal task is marked as pending recovery and recovery information is registered. If all simulated preemption after traversing the candidate sequence does not meet the gain threshold, the non-preemption backup plan is maintained or the ad hoc task is determined to be blocked.

[0017] In one embodiment of the present invention, utilizing idle capacity within a subsequent time window to perform recovery scheduling for preempted normal tasks includes: When the system time reaches the recovery trigger point of a normal task marked as pending recovery, the normal task is woken up; Determine the remaining time window of the normal task from the recovery start point to the end time, and collect the unused free capacity in the network within the remaining time window; Within the remaining time window, the idle capacity is used to perform best-effort transmission scheduling on the awakened normal tasks to restore the remaining data volume that was preempted from the normal tasks; after the restoration is completed, the remaining capacity and task status of each link are updated, and subsequent events are responded to.

[0018] This invention also proposes a cross-domain task-driven dynamic deployment and scheduling system for the Giants Gateway, comprising: The window filtering and evaluation module is used to acquire satellite orbit data and geometric visibility relationships between heterogeneous functional domains, filter candidate cross-domain windows that meet the minimum effective link establishment time constraint, evaluate the value of candidate cross-domain windows based on the needs of routine missions and the needs of hotspot areas, and construct a set of high-value candidate windows. The link configuration and gateway activation module is used to generate cross-domain link configuration schemes by using a set of high-value candidate windows. By simulating the access order of windows and verifying the cross-domain terminal exclusive constraints and device preparation time constraints, it determines the satellite nodes that activate the cross-domain gateway role and the activation sequence of each cross-domain link. Under the premise of meeting the constraints of normal task completion rate and cross-domain capability gap, the configuration scheme is optimized to reduce gateway activation overhead, so as to obtain the cross-domain link configuration scheme and the corresponding normal task baseline scheduling scheme. The ad hoc insertion and reclamation module is used to respond to online ad hoc tasks. Within the capacity reserved by the cross-domain link configuration scheme, it searches for an end-to-end transmission path to perform direct insertion scheduling. If direct insertion cannot meet the transmission requirements of the ad hoc task, it prioritizes reclamation of the resources occupied by the recovery stream previously generated due to preemption and retryes direct insertion. The preemption and recovery scheduling module is used to select controlled preemption objects from low-priority normal tasks that conflict with the transmission path of the ad hoc task based on the preemption loss score if the transmission demand is still not met after the recovery stream is recovered. The module releases the resources of the corresponding controlled preemption objects to complete the ad hoc task scheduling, and then uses the idle capacity to perform recovery scheduling on the preempted normal tasks in the subsequent time window.

[0019] This invention discloses a cross-domain task-driven dynamic deployment and scheduling method and system for the Giants Gateway. By acquiring satellite orbits and geometric visibility relationships within heterogeneous domains, it filters high-value candidate windows and generates cross-domain link configuration schemes, optimizing and reducing gateway activation overhead. When responding to ad hoc tasks, it sequentially executes direct insertion scheduling within reserved capacity, conflict recovery flow recycling, and controlled preemption based on preemption loss scoring, and restores preempted tasks within subsequent windows. This invention effectively solves the problems of poor flexibility and insufficient dynamic response capability in cross-domain task transmission under existing fixed terrestrial gateway architectures. It realizes on-demand dynamic deployment and cross-domain collaborative scheduling of Giants Gateway resources, significantly improving the success rate of ad hoc task insertion and the efficiency of routine task support, and enhancing the system's adaptability and engineering practical value in dynamic business scenarios.

[0020] To achieve the above objectives, a third aspect of this application provides a computer device, including a processor and a memory; wherein the processor reads executable program code stored in the memory to run a program corresponding to the executable program code, for implementing a cross-domain task-driven dynamic deployment and scheduling method for the Giants Gateway as described in the first aspect embodiment.

[0021] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0022] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a cross-domain task-driven dynamic deployment and scheduling method for the Megalogate gateway according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a dynamic deployment scenario for a cross-domain gateway according to an embodiment of the present invention; Figure 3 This is a schematic diagram of a dynamic deployment scenario for a cross-domain gateway according to an embodiment of the present invention; Figure 4 This is a flowchart of the first-stage candidate cross-domain window screening and population initialization according to an embodiment of the present invention; Figure 5This is a flowchart of the first stage of chromosome decoding, feasibility determination and genetic iteration according to an embodiment of the present invention; Figure 6 This is a flowchart of the second-stage event-driven scheduling of a generator according to an embodiment of the present invention; Figure 7 This is a flowchart of the second-stage ad hoc task insertion and controlled preemption decision-making process according to an embodiment of the present invention; Figure 8 This is a flowchart of the second-stage label dynamic programming solution according to an embodiment of the present invention; Figure 9 This is a hotspot coverage comparison chart according to an embodiment of the present invention; Figure 10 This is a comparison chart of cross-domain capability gaps according to an embodiment of the present invention; Figure 11 This is a comparison chart of the number of activations of cross-domain gateways according to an embodiment of the present invention; Figure 12 This is a graph showing the success rates of ad-hoc tasks and routine tasks according to an embodiment of the present invention; Figure 13 This is a schematic diagram of the structure of a cross-domain task-driven giant constellation gateway dynamic deployment and scheduling system according to an embodiment of the present invention; Figure 14 It is a computer device according to an embodiment of the present invention. Detailed Implementation

[0023] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0024] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0025] The following describes, with reference to the accompanying drawings, a cross-domain task-driven dynamic deployment and scheduling method and system for the Giants Gateway according to an embodiment of the present invention.

[0026] Figure 1 This is a flowchart of a cross-domain task-driven dynamic deployment and scheduling method for the Megatron Gateway, according to an embodiment of the present invention.

[0027] like Figure 1 As shown, a cross-domain task-driven dynamic deployment and scheduling method for the Giants Gateway includes the following steps: S1. Obtain satellite orbit data and geometric visibility relationships between heterogeneous functional domains, screen candidate cross-domain windows that meet the minimum effective link establishment time constraint, evaluate the value of candidate cross-domain windows based on the needs of normal missions and the needs of hotspot areas, and construct a set of high-value candidate windows. S2. A cross-domain link configuration scheme is generated using a set of high-value candidate windows. By simulating the access order of the windows and verifying the cross-domain terminal exclusive constraints and device preparation time constraints, the satellite nodes that activate the cross-domain gateway role and the activation sequence of each cross-domain link are determined. Under the premise of meeting the constraints of normal task completion rate and cross-domain capability gap, the configuration scheme is optimized to reduce the gateway activation overhead, so as to obtain the cross-domain link configuration scheme and the corresponding normal task baseline scheduling scheme. S3 responds to online incoming ad hoc tasks, searches for end-to-end transmission paths within the capacity reserved in the cross-domain link configuration scheme to perform direct insertion scheduling; if direct insertion cannot meet the transmission requirements of ad hoc tasks, it prioritizes reclaiming the resources occupied by the recovery stream previously generated due to preemption, and retryes direct insertion. S4. If the transmission requirements are still not met after the recovery stream is recovered, then based on the preemption loss score, controlled preemption objects are selected from the low-priority normal tasks that have resource conflicts with the transmission path of the ad hoc task. The resources of the corresponding controlled preemption objects are released to complete the ad hoc task scheduling, and the idle capacity is used to perform recovery scheduling for the preempted normal tasks in the subsequent time window.

[0028] Specifically, to facilitate understanding of the technical solution of this application, the application scenarios targeted by this invention will be described first.

[0029] like Figure 2 As shown, this application considers a heterogeneous converged giant constellation network scenario consisting of functional domain A and functional domain B. Each functional domain contains multiple satellite nodes performing the services of that domain, and the satellites within each domain maintain connectivity through intra-domain links. Because different functional domains differ in mission attributes, network organization, and interconnection requirements, when cross-domain tasks occur, using a ground gateway relay method can easily lead to transmission path detours and increased latency.

[0030] To this end, this application selects some satellite nodes in the two functional domains as needed, dynamically activates their cross-domain gateway roles (i.e., the dark blue and dark red satellites in the figure), and establishes cross-domain links between the corresponding gateway nodes when conditions such as cross-domain geometric visibility, link establishment preparation time, minimum effective link establishment duration, and terminal occupancy are met. Cross-domain mission data can first be forwarded to the source domain cross-domain gateway via the source functional domain intra-link, then transmitted to the target domain cross-domain gateway via the cross-domain link, and finally forwarded to the target satellite via the target domain intra-link, thereby realizing on-orbit cross-domain transmission between heterogeneous functional domains.

[0031] Because the arrival time, service intensity, and spatial distribution of cross-domain tasks change over time during the planning period, the activation objects, activation sequence, and cross-domain link configuration schemes of the cross-domain gateway all need to be dynamically adjusted. Therefore, this application proposes a two-stage collaborative method for cross-domain tasks, such as... Figure 3 As shown, the first phase completes the cross-domain link configuration and gateway deployment planning on a larger time scale. The second phase, based on the configuration scheme, performs event-driven rapid insertion for online arrival of ad hoc tasks. When direct insertion is not feasible, local correction is achieved through controlled preemption and subsequent recovery to balance the baseline of normal task carrying capacity and the rapid response capability of ad hoc tasks.

[0032] Furthermore, the following sections will provide a detailed explanation of each of the two stages: (I) First stage: Cross-domain link configuration and gateway deployment methods on a large time scale.

[0033] The goal of the first phase is to select and generate a set of cross-domain link configuration schemes from a large number of physically visible cross-domain windows within the planning period. These schemes should meet the needs of normal service operation while having low gateway activation overhead. The outputs of this phase include: the activated cross-domain gateway satellites, the formed cross-domain links, and the activation and termination times of each cross-domain link.

[0034] Let W be the set of all candidate cross-domain windows within the planning period, and let the configuration scheme be denoted as . The optimization objective of the first phase is to minimize the cross-domain gateway activation overhead while satisfying three constraints: normal task completion rate, normal cross-domain capability gap, and average availability of hotspot areas. Correspondingly, the optimization objective of the first phase can be expressed as:

[0035] And satisfy:

[0036]

[0037]

[0038] in, This represents the normal task completion rate under scheme S. This indicates a gap in routine cross-domain capabilities. This indicates the average coverage rate in hotspot areas of Domain A. This represents the total activation cost of the cross-domain gateway within the planning period. Since each activated cross-domain link requires the cross-domain gateway role to be activated simultaneously on both ends of its satellite, the activation cost is calculated as follows: calculate.

[0039] The overall process of the first phase is as follows: Figure 4 and Figure 5 As shown, where Figure 4 Corresponding to the candidate window selection and population initialization process, Figure 5 This corresponds to chromosome decoding, feasibility assessment, and genetic iteration processes. The specific steps of the first stage are as follows: Step 1: Network Topology and Candidate Cross-Domain Window Acquisition. Acquire orbital data, intra-domain link snapshots, and cross-domain geometric visibility relationships for satellites in Domain A and Domain B within the planning period; extract the complete set of physically visible cross-domain windows and record the satellites at both ends of each window and their visible start and end times. For any candidate window... Let the satellites at both ends be denoted as follows: and Its physically visible time interval is At the same time, invalid windows with a duration less than the minimum effective chain establishment time threshold are removed to create a complete set of candidate windows for subsequent screening.

[0040] Step 2: Construct a set of high-value candidate windows. To reduce the subsequent search space, a static value evaluation is first performed on all candidate windows. The static value includes two categories: one reflects the potential contribution of a window to routine task performance, and the other reflects the potential contribution of a window to ensuring access to hotspot areas.

[0041] For the value of routine tasks, let the set of active routine tasks within a subdivided time period r be . For any normal task Its demand intensity is defined as

[0042] in, For the maximum available rate of the task, The total amount of data for the task. and These represent the task arrival time and deadline time, respectively. Furthermore, the reachability weight of window k for task q within segment r is defined as:

[0043] in and These represent the shortest access hops from the source node in domain A within segment r to the satellite at window A, and from the satellite at window B in domain B to the destination node, respectively. Therefore, the weighted demand pressure carried by window k within segment r is defined as:

[0044] Further, by summing these values ​​over time throughout the entire planning period, we obtain the static value of window k oriented towards routine tasks:

[0045] in, Indicates the duration of interval overlap. This represents a set of subdivided time periods formed by the arrival and deadline times of routine tasks and changes in the domain topology snapshot. This indicator is used to characterize the potential support capability of candidate windows for routine tasks during the planning period.

[0046] To assess the protection value of hotspot areas, let the set of hotspot areas in domain A be denoted as . Among them, hot topics The weight is And satisfy If the A-end satellite of candidate window k Within segment r, it can not exceed the given maximum number of jumps. If, under the condition that window k is accessed by any satellite within hotspot region i, then window k is considered to cover hotspot region i within that segment. The local contribution of window k within segment r is further defined as:

[0047] in, Let k be the coverage indication of hotspot i in segment r for window k. Let r be the number of candidate windows that can cover hotspot i within segment r. It is a very small positive number. Therefore, the static value of window k for hotspot protection is obtained:

[0048] This metric is used to characterize the potential contribution of candidate windows to ensuring access to hotspot areas in Domain A.

[0049] In obtaining and Then, windows with higher overall value are extracted to form the global master set. Next, the planning period is divided into several time periods based on the arrival and deadline times of normal tasks. The number of windows in the master set during each time period with demand is checked. If the number of windows is insufficient in a certain time period, windows are supplemented from the remaining windows according to the local value of that time period, ultimately forming a set of high-value candidate windows used in subsequent genetic searches. The purpose of this step is to control the search scale while avoiding the complete omission of windows that are not of significant global value but are representative in a local time period.

[0050] Step 3: Perform permutation encoding based on the candidate window pool and initialize the population. Permutation encoding is used to represent the order in which candidate windows are attempted to be added to a scheme, i.e., one chromosome corresponds to a set of candidate windows. An arrangement:

[0051] This means that candidate windows are sequentially added to the current cross-domain link configuration scheme in this order. Therefore, the chromosome itself is not directly equivalent to the final scheme; the final scheme S needs to be generated by the decoder based on the arrangement order, device readiness constraints, and terminal occupancy constraints. During population initialization, a diverse range of individuals is generated using a combination of static value descending order templates, density descending order templates, and hierarchical random candidate lists, while retaining a small number of completely random individuals to enhance the globality of the search. To avoid convergence in decoding results due to high similarity in the initial order of different individuals, prefix comparison and deduplication are performed on individual individuals during the initialization phase.

[0052] Step 4: Block Evaluation and Decoding under Resource Constraints. Following the order given by the chromosome, candidate windows are sequentially added to the current configuration scheme. For any candidate window k, if it is accepted, the satellites at both ends are considered unusable for establishing other cross-domain links within the corresponding occupied interval.

[0053] During the decoding process, for each window, the earliest feasible link opening time is calculated by considering the cross-domain terminal occupancy status of satellites at both ends of the window and the equipment preparation requirements. Let... Preparation time for equipment and These represent the times at which the window ends. Previously, the earliest feasible open-chain time of window k was defined as the end time of the most recent terminal occupation of satellites at both ends of the window:

[0054] If the minimum effective chain establishment time condition is met:

[0055] If the window is accepted, it is accepted; otherwise, it is rejected. Therefore, this invention does not simply select cross-domain windows based on geometric visibility relationships, but simultaneously considers engineering constraints such as device preparation time, terminal exclusive occupancy, and minimum effective connection establishment time, thereby ensuring the practical feasibility of the resulting solution.

[0056] As solutions are gradually added to the window, a block-based evaluation is performed on the current prefix solution. The evaluation mainly includes three types of indicators: routine task completion rate. Regular cross-domain capability gap and average coverage in hotspot areas Among them, the routine task completion rate reflects whether routine tasks can be completed on time within the planning period; the routine cross-domain capacity gap measures the degree of cross-domain resource insufficiency of the current solution from the perspective of aggregated cross-domain supply and demand; and the average guarantee level of hotspot areas measures the average degree to which hotspot areas in Domain A are effectively accessed by active cross-domain links within the planning period. If all three indicators reach the preset threshold simultaneously, the window insertion will stop, and the prefix solution will be used as the initial feasible solution obtained from the current chromosome decoding.

[0057] Step 5: Reverse Redundancy Pruning. To further reduce the number of cross-domain gateway activations and resource consumption, a reverse deletion check is performed on the initial feasible solutions obtained in Step 4. Deletions are attempted sequentially starting from the last window of the solution; if the normal task completion rate, capacity gap, and average availability of hotspot areas still meet the predetermined thresholds after deletion, the window is permanently removed; otherwise, the window is retained. After traversal, a streamlined configuration solution corresponding to the chromosome is obtained. The purpose of this step is to eliminate redundant windows that are not essential to meeting feasibility constraints, thereby further reducing activation costs.

[0058] Step Six: Fitness Calculation and Genetic Evolution. Fitness evaluation criteria are constructed for feasible and infeasible schemes. For infeasible schemes, the evaluation is based on the normalized comprehensive results of three types of violations: insufficient completion rate of normal tasks, excessive capacity gap, and insufficient hotspot support. For feasible schemes, the number of activated cross-domain gateways is compared first, and then a hierarchical comparison is performed by combining the completion rate of normal tasks, capacity gap, and average support level of hotspot areas. Subsequently, crossover operations that preserve the initial sequence characteristics and mutation operations oriented towards the initial part of the chromosome are used to generate offspring, and the population is continuously iterated and updated until the maximum number of iterations is reached or there is no improvement for several consecutive generations. Finally, the cross-domain link configuration scheme with optimal fitness and its corresponding gateway activation schedule are output.

[0059] (ii) Second stage: Methods for inserting and resuming ad hoc tasks on a small time scale.

[0060] The second phase operates on top of the cross-domain resource baseline output in the first phase. Its goal is not to re-plan all routine tasks, but rather to quickly insert ad hoc tasks that arrive randomly online, while minimizing disruption to existing routine business commitments. When direct insertion cannot meet the demand, local adjustments are made through recovery stream recycling, controlled preemption, and subsequent recovery mechanisms. Unlike the first phase, which focuses on overall configuration throughout the planning cycle, the second phase better reflects the actual operational characteristics of immediate response upon the arrival of ad hoc tasks.

[0061] Let S be the cross-domain link configuration scheme output in the first phase, and let S be the baseline scheduling scheme for normal tasks corresponding to phase one. The inputs for the second phase include: Scheme S and the baseline scheduling scheme. The set of routine tasks and the set of ad hoc tasks, various link capacity parameters, and the proportion of cross-domain capacity reserved for ad hoc tasks in Phase 1. The output of the second phase includes: the actual segment allocation results, the completion rate of normal tasks and ad hoc tasks, the number of preemptions, the utilization rate of cross-domain links and the entire network links, as well as the recovery record of each ad hoc insertion event and each preempted task.

[0062] To ensure the real-time processing of ad hoc tasks and minimal disruption to routine operations, the second phase follows three principles: First, incremental rather than starting from scratch, always using the Phase 1 baseline as the starting point and making only local adjustments near ad hoc events; Second, tiered resource utilization, prioritizing the use of cross-domain capacity reserved for ad hoc tasks, borrowing idle capacity from routine tasks when necessary, and only considering preemption when still insufficient; Third, minimal disruption, prioritizing the recovery flow and then preempting low-priority routine tasks that are truly obstructing the flow and still have the potential for recovery.

[0063] The overall operation process of the second phase is as follows: Figure 6 As shown. The specific steps are as follows: Step 1: Construct and dynamically segment event time periods. The system first initializes event time periods based on the normal business baseline scheduling scheme formed in the first phase. This time period sequence is determined by the arrival and end times of normal tasks, as well as the start and end times of cross-domain links in scheme S. When a new ad hoc task arrives during online operation, or a preempted task recovery scheduling event is triggered, if the event time falls within an existing time period, the time period is segmented in real time at that moment. This ensures that resource status changes, task arrivals, and end times are aligned with the time period boundaries, thereby improving subsequent scheduling accuracy.

[0064] Step Two: Direct Insertion Search for Impromptu Tasks. The decision-making process for inserting, preempting, and guaranteeing tasks for impromptu tasks is as follows: Figure 7 As shown. When the ad hoc task k is at time... Upon arrival, the second phase first constructs its emergency time domain:

[0065] in, This indicates the start time of segment r. This indicates the deadline for task k. In other words, the planning scope of ad hoc tasks is limited to all valid segments from the deadline to the deadline; other segments are not included in this processing.

[0066] Without altering any committed routine business allocations, the system reads the current committed scheduling table. The capacity status of each segment is determined, and cross-domain end-to-end candidate paths are generated for ad hoc tasks. For any candidate path p within a segment r, the candidate rate that this segment can be allocated to task k is determined by three upper limits: the maximum available rate of the task. The minimum remaining capacity of each edge on the path. And the amortized rate of the remaining data volume of the task within this segment. Therefore, the definition is:

[0067] The corresponding transmittable amount for this segment is:

[0068] in, Let r be the duration of segment r. This formula shows that the allocatable rate of an ad hoc task within a certain segment is simultaneously constrained by the task rate cap, the remaining capacity of the path bottleneck, and the remaining amount of data in the task.

[0069] In terms of resource allocation, the second phase adopts a two-tier capacity structure for cross-domain links. Let the total capacity of the cross-domain links be... Then the available capacity for normal tasks is:

[0070] The reserved capacity for ad-hoc tasks is:

[0071] If the required speed for the ad hoc task falls entirely within If the allocation exceeds the reserved share and borrows a portion not fully occupied by regular tasks, it is called borrowing from the regular idle share. During the direct insertion phase, the option that falls entirely within the reserved share is preferred; when borrowing is necessary, the smaller the borrowing amount, the better.

[0072] Path search employs multi-label dynamic programming. Each label records at least the current remaining data volume, the cross-domain link used in the previous segment, and the corresponding cumulative cost; the cumulative cost comprehensively characterizes factors such as cross-domain switching, the degree of borrowing from normal idle shares, and path load. Labels significantly inferior to other candidate states are deleted using non-dominated pruning. If a direct insertion scheme exists that can complete all data transmission before the deadline, the scheme is directly written into the current committed scheduling table, and the insertion ends; if only partial data transmission can be completed, it is retained as a best-effort fallback scheme for this insertion event, and steps three and four are performed to seek further improvement; if no feasible transmission is possible, the insertion is considered blocked. The label dynamic programming solution process is as follows: Figure 8 As shown.

[0073] Step 3: Establish the conflict scope and filter preemptible normal tasks. If direct insertion fails, or only partial transmission can be completed, the system will construct candidate corridors based on the near-optimal candidate paths of the ad hoc tasks to define the resource range that can actually be used in the subsequent processing of the current ad hoc task. Specifically, in each segment... In the process, only near-optimal paths with a transmittance not less than a certain proportion of the optimal path for that segment are retained. Then, the intra-domain edges and cross-domain edges involved in these paths are combined to form the corridor edge set on segment r. This yields candidate corridors across the entire emergency response time domain:

[0074] Candidate corridors serve three purposes: first, to define the range of resources that may actually be occupied in the subsequent processing of ad hoc tasks; second, to screen regular tasks that intersect with the corridor in terms of time and edge resources as preemption candidates; and third, to calculate the effective release amount of the committed resources that can truly help the current ad hoc task when evaluating whether a regular task is worth preempting.

[0075] Before transitioning to preemption, the system first checks if any recovery flows in the currently committed scheduling table are occupying critical resources in the candidate corridor. A recovery flow refers to a segmented allocation that was previously preempted to handle other ad hoc tasks and subsequently regained through the recovery mechanism; these are not part of the original Phase 1 baseline but rather remedial allocations generated in Phase 2. If a recovery flow exists and... In case of a conflict, the allocation of the conflicting segment is first revoked, the remaining data of the corresponding task is rolled back, and it is placed back into the recovery queue. The arrival time of the current ad hoc task is used as the new recovery starting point. Then, the direct insertion in step two is re-executed under the updated capacity view. If the result after re-insertion is better than the original plan, the recovery action is retained; otherwise, the recovery stream recovery is rolled back, and step four is initiated. This design conforms to the principle of minimum disturbance because prioritizing the recovery stream recovery does not introduce new disturbed normal tasks.

[0076] Step 4: Controlled Preemption Decision Based on Preemption Loss Scoring. If direct insertion and recovery flow recovery fail to complete the ad hoc task, the process transitions to controlled preemption. Controlled preemption does not indiscriminately release all normal tasks; it only considers low-priority normal tasks that are indeed blocking the current ad hoc task's path. For the current ad hoc task k, from normal tasks with lower priority and overlapping business time domains with the current ad hoc task's emergency time domain, tasks that actually conflict with the current candidate corridor in terms of time and resources are further selected as preemption candidate tasks.

[0077] For each preemption candidate task m, the system further calculates its preemption loss score. This scoring system comprehensively reflects four factors: the more important the preempted task itself, the greater the loss; the smaller the recovery margin after preemption, the greater the loss; the more difficult it is expected to recover after preemption, the greater the loss; conversely, if releasing the committed resources of the task in the conflict segment can significantly help the current contingent task, then its loss should be reduced accordingly. Therefore, a task weight normalization value is defined. , Restoration margin normalized value Recoverability normalized value and the normalized value of effective release And construct:

[0078] in, These are the weighting coefficients. To prevent dividing by zero from yielding a very small positive number, the numerator represents the cost of releasing the value, and the denominator represents the benefit of releasing the value. Therefore... The smaller the value, the more valuable the task is to be prioritized. The system sorts all candidate tasks by... Sort them from smallest to largest to form a preemptive candidate sequence.

[0079] Step 5: Single-task preemption simulation and result distribution based on gain threshold. Following the candidate sequence obtained in Step 4, the system sequentially simulates releasing resources already occupied by individual normal tasks within the conflict segment. Then, under the updated capacity view, it re-executes the direct insertion search from Step 2 to calculate a new transmission scheme for the ad hoc task. Subsequently, it calculates the additional transmittable capacity and additional completion percentage brought to the ad hoc task by this preemption compared to the safety net scheme without preemption. Simultaneously, it performs a strict recoverability estimate on the released normal task to assess its expected recoverable data volume in the remaining time domain. Preemption is only confirmed if the additional benefit from this preemption exceeds a preset gain threshold; otherwise, it is rejected, and the system continues to try the next task in the candidate sequence. This threshold is used to avoid significantly disrupting the existing normal scheduling in order to increase the transmission capacity of the ad hoc task by only a tiny fraction.

[0080] Once the optimal controlled preemption scheme is determined, the system revokes some or all of the resources occupied by the affected normal task within the corresponding conflict segment and reallocates the released resources to the ad hoc task. Simultaneously, the normal task is marked as "pending recovery" and its recovery information is recorded. If the ad hoc task completes before the deadline after preemption, it is recorded as fully completed. If only partial transmission can be completed, it is compared with the backup scheme without preemption, and the better one is retained as the final scheme. If all candidate preemptions are rejected by the gain threshold, or none bring substantial improvement, the backup scheme obtained in step two is maintained, or the system is judged as blocked.

[0081] Step Six: Best-effort recovery mechanism for preempted normal tasks. When the simulation time progresses to the recovery event trigger point, the system reawakens the normal tasks in the pending recovery state and performs best-effort recovery using the remaining time window between the recovery start point and the task's deadline, utilizing the currently unused idle capacity in the network. After recovery is complete, the system updates the remaining capacity of each link, the remaining data volume of the task, and its running status, and continues to respond to subsequent events until the planned cycle ends.

[0082] The method of this invention optimizes cross-domain window configuration through genetic evolution, reducing gateway activation overhead. It responds to ad hoc tasks based on a controlled preemption mechanism driven by direct insertion, recovery flow recycling, and preemption loss scoring, supplemented by subsequent recovery scheduling. Simulation results show that, compared to the static greedy method, this method achieves lower capacity penalties and better hotspot coverage under light load, baseline, and heavy load normal task sets, with a significant reduction in activation counts. In light to heavy ad hoc disturbance scenarios, ad hoc tasks receive priority protection, and the completion rate of normal tasks remains at a reasonable level under resource constraints. This invention effectively solves the problem of poor flexibility in static gateway deployment during the Giant Star cross-domain task scheduling, achieving dynamic deployment and scheduling with high resource utilization and strong robustness.

[0083] To make the objectives, technical solutions, and technical effects of this invention clearer, the invention will be further explained below with reference to specific simulation examples. This example considers a cross-domain transmission scenario in the constellation Gigantopithecus, consisting of two heterogeneous functional domains. Functional domain A uses a sun-synchronous orbit constellation at an altitude of 500km, containing 5 orbital planes with 9 satellites per orbit, totaling 45 satellites; functional domain B uses a circular orbit constellation at an altitude of 1000km, containing 8 orbital planes with 10 satellites per orbit, totaling 80 satellites. The maximum distance for establishing inter-domain inter-satellite links is set to 5000km. The network topology is sampled at 60s time steps, and the minimum length threshold for cross-domain links is set to 300s. Furthermore, four hotspot areas are set within domain A, with weights of 0.35, 0.25, 0.20, and 0.20, respectively, to characterize the access guarantee requirements for hotspot areas in Phase 1.

[0084] In terms of task set design, to verify the applicability of the proposed method under different normal load intensities, three normal task sets—Light60, Normal72, and Heavy84—were constructed, respectively. Light60 represents a relatively light load scenario, Normal72 a baseline load scenario, and Heavy84 a relatively heavy load scenario. In the ad hoc task simulation, Normal72 was used as the normal baseline, and three ad hoc task sets—smoke, adversarial, and stress—were superimposed to characterize light, moderate, and heavy online perturbation scenarios. The number of ad hoc tasks corresponding to the three ad hoc scenarios were 8, 12, and 20, respectively, with a total number of tasks of 80, 84, and 92, respectively.

[0085] In the above scenarios, the static greedy configuration method and the cross-domain gateway window configuration method based on genetic evolution proposed in this invention are compared. The static greedy method only sorts and selects windows based on their static value, without crossover, mutation, or iterative evolution; the method of this invention, under the same set of candidate windows and feasibility constraints, further optimizes the sparsity of window combinations, capacity matching quality, and hotspot coverage performance through genetic search.

[0086] In the simulation, the thresholds for normal task fulfillment rate were set to 1, hotspot coverage rate to 0.8, and cross-domain capability gap to 0.1. From... Figure 9 , Figure 10 and Figure 11 As can be seen, compared with the static greedy method, the cross-domain gateway window configuration method based on genetic evolution proposed in this invention achieves superior overall performance on all three normal task sets. Specifically, the method of this invention has significantly lower capacity penalty indicators than the comparative methods, fewer activations in Light60 and Heavy84 scenarios, and higher hotspot coverage in Normal72 and Heavy84 scenarios. This indicates that the method of this invention can achieve better capacity matching and higher resource configuration quality while ensuring the feasibility of the solution.

[0087] from Figure 12 As can be seen, the Phase 2 method maintained good task completion capabilities in all three ad hoc scenarios, demonstrating relatively stable overall performance. With the increase in ad hoc task concurrency, ad hoc tasks were consistently well-supported, while the completion rate of regular tasks declined. This indicates that the method prioritizes ad hoc task requirements when they arise, and maintains regular task services as much as possible under resource constraints.

[0088] In summary, the Phase 2 approach effectively reflects the design principle of prioritizing responses to ad hoc tasks while also considering routine tasks, demonstrating its good adaptability and robustness under dynamic disturbance scenarios of varying intensities.

[0089] The simulation examples of this invention optimize cross-domain window configuration through genetic evolution and perform hierarchical scheduling and controlled preemption when ad hoc tasks arrive. Simulation results show that, compared with the static greedy method, this invention achieves lower capacity penalties and better hotspot coverage under light load, baseline, and heavy load normal task sets, with a significant reduction in activation counts. In light to heavy ad hoc disturbance scenarios, ad hoc tasks are prioritized, and the completion rate of normal tasks remains at a reasonable level when resources are limited. This invention effectively solves the problem of poor flexibility in the static deployment of gateways in the Giant Star cross-domain task scheduling, achieving dynamic deployment and scheduling with high resource utilization and strong robustness.

[0090] To implement the method of the above embodiments, such as Figure 13 As shown, this invention proposes a cross-domain task-driven dynamic deployment and scheduling system 10 for the Giants Gateway, comprising: The window screening and evaluation module 100 is used to acquire satellite orbit data and geometric visibility relationships between heterogeneous functional domains, screen candidate cross-domain windows that meet the minimum effective link establishment time constraint, evaluate the value of candidate cross-domain windows based on the needs of normal missions and the needs of hotspot areas, and construct a set of high-value candidate windows.

[0091] The link configuration and gateway activation module 200 is used to generate a cross-domain link configuration scheme by using a set of high-value candidate windows. By simulating the access order of the windows and verifying the cross-domain terminal exclusive constraints and device preparation time constraints, it determines the satellite nodes that activate the cross-domain gateway role and the activation sequence of each cross-domain link. Under the premise of meeting the constraints of normal task completion rate and cross-domain capability gap, the configuration scheme is optimized to reduce gateway activation overhead, so as to obtain the cross-domain link configuration scheme and the corresponding normal task baseline scheduling scheme.

[0092] The ad hoc insertion and reclamation module 300 is used to respond to online ad hoc tasks. Within the capacity reserved by the cross-domain link configuration scheme, it searches for an end-to-end transmission path to perform direct insertion scheduling. If direct insertion cannot meet the transmission requirements of the ad hoc task, it prioritizes reclamation of the resources occupied by the recovery stream previously generated due to preemption and retryes direct insertion.

[0093] The preemption and recovery scheduling module 400 is used to select controlled preemption objects from low-priority normal tasks that conflict with the transmission path of the ad hoc task based on the preemption loss score if the transmission demand is still not met after the recovery stream is recovered. The module releases the resources of the corresponding controlled preemption objects to complete the ad hoc task scheduling, and performs recovery scheduling on the preempted normal tasks using the idle capacity in the subsequent time window.

[0094] The system of this invention constructs a set of high-value candidate windows through a window filtering and evaluation module, generates a cross-domain link configuration scheme and optimizes it to reduce gateway activation overhead through a link configuration and gateway activation module, performs direct insertion and recovery flow recycling through an ad hoc insertion and recycling module, and implements controlled preemption and subsequent recovery based on preemption loss scoring through a preemption and recovery scheduling module. This invention effectively solves the problems of poor static deployment flexibility and insufficient ad hoc response capability of gateways in cross-domain task transmission between heterogeneous domains in the Giant Star constellation. It achieves a collaborative solution throughout the entire process from window filtering and link configuration to ad hoc scheduling and recovery, significantly improving the system's resource utilization efficiency and scheduling robustness in dynamic business scenarios, and enhancing its practical engineering value.

[0095] To implement the methods of the above embodiments, the present invention also provides a computer device, such as... Figure 14As shown, the computer device 600 includes a memory 601 and a processor 602; wherein, the processor 602 reads the executable program code stored in the memory 601 to run a program corresponding to the executable program code, so as to implement the various steps of the cross-domain task-driven dynamic deployment and scheduling method for the Giants Gateway described above.

[0096] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0097] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A cross-domain task-driven dynamic deployment and scheduling method for the Giants Gateway, characterized in that, include: Acquire satellite orbit data and geometric visibility relationships between heterogeneous functional domains, screen candidate cross-domain windows that meet the minimum effective link establishment time constraint, evaluate the value of candidate cross-domain windows based on the needs of routine missions and the needs of hotspot areas, and construct a set of high-value candidate windows. A cross-domain link configuration scheme is generated by using a high-value candidate window set. By simulating the access order of the windows and verifying the cross-domain terminal exclusive constraints and device preparation time constraints, the satellite nodes that activate the cross-domain gateway role and the activation sequence of each cross-domain link are determined. Under the premise of meeting the constraints of normal task completion rate and cross-domain capability gap, the configuration scheme is optimized to reduce gateway activation overhead, so as to obtain the cross-domain link configuration scheme and the corresponding normal task baseline scheduling scheme. In response to online incoming ad hoc tasks, search for end-to-end transmission paths within the capacity reserved by the cross-domain link configuration scheme to perform direct insertion scheduling; if direct insertion cannot meet the transmission requirements of ad hoc tasks, prioritize reclaiming the resources occupied by the recovery stream previously generated due to preemption, and retry direct insertion. If the transmission demand is still not met after the recovery stream is recovered, then based on the preemption loss score, controlled preemption objects are selected from low-priority normal tasks that have resource conflicts with the transmission path of the ad hoc task. The resources of the corresponding controlled preemption objects are released to complete the ad hoc task scheduling, and the idle capacity is used to perform recovery scheduling for the preempted normal tasks in the subsequent time window.

2. The method as described in claim 1, characterized in that, Based on the needs of routine tasks and the needs of hotspot areas, the value of candidate cross-domain windows is evaluated, and a set of high-value candidate windows is constructed, including: For any active routine task within any time segment of the planning period, calculate the corresponding demand intensity, which is the smaller of the task's maximum available rate and the amortized rate calculated based on the remaining time and remaining data volume. Calculate the reachability weight of the candidate window for the routine task based on the shortest access hops from the source node in the source functional domain to the satellite at one end of the window, and the shortest access hops from the satellite at the other end of the window in the target functional domain to the destination node. Accumulate the product of the task weight, demand intensity, and reachability weight over the planning period according to the time overlap duration to obtain the static value of the candidate window for the routine task. For a pre-defined set of hotspot areas, if a satellite at one end of a candidate window can be accessed by a satellite within a hotspot area within a given time segment without exceeding the hop count limit, then the physical visibility duration of the window, the hotspot area weight, and the normalized coverage indicator within the corresponding time segment are multiplied to obtain the local contribution of the window to the hotspot area in the corresponding segment; the local contributions of the window to the hotspot area in all segments within the planning period are summed to obtain the static value of the corresponding candidate window for hotspot protection. Based on the static value of routine tasks and the static value of hotspot support, windows with higher comprehensive value are extracted first to form a global master set. Then, the number of windows in the global master set within each demand time period is checked. For time periods with insufficient windows, windows are supplemented from the remaining windows according to the local value of the corresponding time period. Finally, a set of high-value candidate windows for subsequent scheduling is constructed.

3. The method as described in claim 1, characterized in that, Generate cross-domain link configuration schemes using a high-value candidate window set, including: The attempts of windows in the high-value candidate window set are added to the sequence encoding as chromosomes using the permutation encoding method, with one chromosome corresponding to one permutation of the corresponding set; Candidate windows are added to the configuration scheme sequentially according to the chromosome arrangement order. For each candidate window... Obtain the most recent terminal occupancy end time before the end time of the window for both satellites, and calculate the earliest feasible open-chain time using the following formula, combined with the equipment preparation time: in, For window The physical visible start time, For window The physical visible end time, For window The A-end satellite in The last time the terminal was occupied ended. For window B-end satellite in The last time the terminal was occupied ended. Allow time for equipment preparation; if If the condition is met, the corresponding window will be accepted and the terminal occupancy status of both satellites will be updated; otherwise, the corresponding window will be rejected. The minimum effective chain establishment time threshold; During the gradual addition of windows, the routine task completion rate, routine cross-domain capability gap and average guarantee level of hotspot areas are calculated for the prefix schemes formed by the currently accepted windows. When all three indicators reach the preset threshold, the addition of windows is stopped, and the initial feasible scheme is obtained. Perform reverse redundancy pruning on the initial feasible schemes, and try to delete them sequentially starting from the last window of the scheme. If the normal task completion rate, normal cross-domain capability gap and average guarantee level of hot spot areas still meet the preset threshold after deletion, the corresponding window is permanently removed; otherwise, the corresponding window is retained. After traversal, a simplified cross-domain link configuration scheme is obtained.

4. The method as described in claim 1, characterized in that, Searching for end-to-end transmission paths within the capacity reserved in the cross-domain link configuration scheme to perform direct insertion scheduling includes: An emergency time domain is constructed based on the arrival and end times of the emergency task, and the emergency time domain consists of all segments falling within the time interval. Without changing the committed normal business allocation, read the capacity status on each segment and generate cross-domain end-to-end candidate paths for ad hoc tasks; for each candidate path in a segment, calculate the candidate rate that can be allocated to ad hoc tasks. The candidate rate is the minimum of the ad hoc task's maximum available rate, the minimum remaining capacity on each side of the path, and the average rate of the ad hoc task's remaining data volume in this segment. The total capacity of cross-domain links is divided into a two-layer structure: the capacity available for normal tasks and the capacity reserved for ad hoc tasks. The scheme that can complete the transmission by using the reserved capacity for ad hoc tasks is given priority. If it is necessary to borrow the idle capacity of normal tasks, the scheme with the smallest amount of borrowing is selected. A multi-label dynamic programming algorithm is used to perform a segmented progressive search in the emergency time domain. Each label records the current remaining data volume, the cross-domain link used in the previous segment, and the cumulative cost. Non-dominated pruning is performed on the labels generated in each segment, and only non-dominated labels are retained to enter the next segment. If a path scheme that can complete all data transmission before the deadline is found, the path scheme is written into the current committed scheduling table to complete direct insertion scheduling.

5. The method as described in claim 1, characterized in that, When a direct insert search cannot complete all ad hoc task data transfers before the deadline, it also includes: Identify that the recovery flow in the current committed scheduling table belongs to a recovery flow that was previously preempted and then reallocated, and that the resources occupied by the recovery flow conflict with the near-optimal candidate path of the current ad hoc task; Cancel the resource allocation of the conflict recovery flow on the conflict segment, put the normal task corresponding to the resource allocation back to the pending recovery state, and update the capacity view; In the updated capacity view, re-execute the direct insert search on the current ad hoc task; if the transmission completion rate after re-insertion is improved compared to before reclamation, retain the corresponding reclamation action and output the updated scheduling scheme; otherwise, roll back to the state before reclamation.

6. The method as described in claim 1, characterized in that, Controlled preemption targets are selected based on preemption loss scores, including: Candidate corridors are constructed based on the near-optimal candidate paths generated during the direct insertion phase of the emergency task. The candidate corridors are formed by the union of intra-domain edges and cross-domain edges contained in each segment of the emergency time domain, where the transmittable amount is not less than a preset proportion of the transmittable amount of the optimal path in the corresponding segment. From routine tasks that have lower priority than ad hoc tasks and whose business time domain overlaps with the emergency time domain of ad hoc tasks, tasks whose occupied edge resources and candidate corridors have actual conflicts in time and space are selected to form a set of preemption candidate tasks. For each preemption candidate task Calculate the preemption loss score using the following formula: in, For the task Normalized task weights For the task The normalized value of the recovery margin, For the task The recoverability normalized value, For the task The normalized value of the effective release of ad hoc tasks after being preempted. , , For preset weighting coefficients, To prevent extremely small positive numbers from being divided by zero; All candidate tasks for seizing are sorted in ascending order of their seizing loss score to form a candidate sequence for seizing. The lower the score, the higher the overall benefit of releasing the corresponding task.

7. The method as described in claim 6, characterized in that, After forming the preemption candidate sequence, it also includes: According to the preemption candidate sequence, the resources occupied by individual preemption candidate tasks in the conflict segment are simulated and released in turn. Then, under the updated capacity view, the direct insertion search is re-executed for the contingent task to obtain the simulated transmission scheme. The additional transferable amount of the simulated transfer scheme for ad hoc tasks is calculated compared to the backup scheme without preemption, and the expected recoverable amount of data for released normal tasks in the remaining time domain is evaluated. Preemption of the corresponding candidate task is confirmed only when the additional transmittable amount exceeds the preset gain threshold. The resource occupation of the corresponding candidate task in the conflict segment is officially revoked, the released resources are allocated to the current ad hoc task, and the preempted normal task is marked as pending recovery and recovery information is registered. If all simulated preemption after traversing the candidate sequence does not meet the gain threshold, the non-preemption backup plan is maintained or the ad hoc task is determined to be blocked.

8. The method as described in claim 7, characterized in that, In subsequent time windows, idle capacity is used to perform recovery scheduling for preempted routine tasks, including: When the system time reaches the recovery trigger point of a normal task marked as pending recovery, the normal task is woken up; Determine the remaining time window of the normal task from the recovery start point to the end time, and collect the unused free capacity in the network within the remaining time window; Within the remaining time window, the idle capacity is used to perform best-effort transmission scheduling on the awakened normal tasks to restore the remaining data volume that was preempted from the normal tasks; after the restoration is completed, the remaining capacity and task status of each link are updated, and subsequent events are responded to.

9. A cross-domain task-driven dynamic deployment and scheduling system for the Giants Gateway, characterized in that, include: The window filtering and evaluation module is used to acquire satellite orbit data and geometric visibility relationships between heterogeneous functional domains, filter candidate cross-domain windows that meet the minimum effective link establishment time constraint, evaluate the value of candidate cross-domain windows based on the needs of routine missions and the needs of hotspot areas, and construct a set of high-value candidate windows. The link configuration and gateway activation module is used to generate cross-domain link configuration schemes by using a set of high-value candidate windows. By simulating the access order of windows and verifying the cross-domain terminal exclusive constraints and device preparation time constraints, it determines the satellite nodes that activate the cross-domain gateway role and the activation sequence of each cross-domain link. Under the premise of meeting the constraints of normal task completion rate and cross-domain capability gap, the configuration scheme is optimized to reduce gateway activation overhead, so as to obtain the cross-domain link configuration scheme and the corresponding normal task baseline scheduling scheme. The ad hoc insertion and reclamation module is used to respond to online ad hoc tasks. Within the capacity reserved by the cross-domain link configuration scheme, it searches for an end-to-end transmission path to perform direct insertion scheduling. If direct insertion cannot meet the transmission requirements of the ad hoc task, it prioritizes reclamation of the resources occupied by the recovery stream previously generated due to preemption and retryes direct insertion. The preemption and recovery scheduling module is used to select controlled preemption objects from low-priority normal tasks that conflict with the transmission path of the ad hoc task based on the preemption loss score if the transmission demand is still not met after the recovery stream is recovered. The module releases the resources of the corresponding controlled preemption objects to complete the ad hoc task scheduling, and then uses the idle capacity to perform recovery scheduling on the preempted normal tasks in the subsequent time window.

10. An electronic device, comprising: processor; The memory stores executable instructions; when the processor executes the instructions, it implements the cross-domain task-driven dynamic deployment and scheduling method for the Giants Gateway as described in any one of claims 1-8.

Images