Satellite-ground cooperative beam hopping scheduling method and system for mega constellation satellite network

Abstract

The application discloses a method and system for satellite-ground cooperative beam hopping scheduling for a mega-constellation satellite network, mainly solving the problems of high complexity of beam hopping scheduling, inter-satellite co-frequency interference and load imbalance under multi-satellite overlapping coverage in the prior art. The scheme comprises: dividing the mega-constellation into multiple satellite cooperative clusters; a ground control center calculates initial activation probability and isolation distance threshold of each cell according to historical business and network topology and uploads; the satellite cluster performs two-stage scheduling in orbit, that is, first, the pre-activated cell is selected by combining the iteration roulette mechanism of dynamic probability update with the isolation distance constraint, and then the pre-activated cell is allocated to the visible satellite with the minimum cumulative load in descending order of business demand by using the greedy strategy, and a beam hopping pattern matrix is output and a beam hopping schedule is generated. The complexity of the application only increases linearly with the number of satellites, and the beam hopping schedule can improve the throughput of the satellite, relieve interference and balance the load between satellites, and can be used for multi-satellite cooperative beam resource management and interference avoidance of a mega-low-orbit satellite constellation.

Description

Technical Field

[0001] This invention belongs to the field of satellite communication technology, and specifically relates to a satellite-ground cooperative beam hopping scheduling method, which can be used for multi-satellite cooperative beam resource management and interference avoidance of giant low-Earth orbit satellite constellations. Background Technology

[0002] In recent years, the rapid deployment of mega- constellations of low-Earth orbit (LEO) satellite communication, represented by Starlink and OneWeb, has marked the beginning of an accelerated development phase for mega-constellation satellite network technology. Due to their low latency and wide-area coverage capabilities, LEO satellite networks have become a key development direction for satellite communication systems. In mega-LEO constellations deploying multi-beam phased array antennas, beam-hopping technology, through dynamic scheduling of beam pointing and time slot allocation, achieves time-division coverage of a massive number of ground cells with limited beam resources, and is a key technology for improving system spectral efficiency and throughput. However, the multi-coverage characteristics of mega-constellations and the uneven distribution of ground service demands have led to increasingly serious co-channel interference and load imbalance. Among these, severe cooperative interference caused by overlapping coverage of multiple satellite beams significantly impacts the system throughput and resource utilization efficiency of satellite networks. When designing beam-hopping systems, it is necessary to assess satellite resources based on the real-time service demands of ground cells, optimize the beam-hopping schedule to avoid interference and balance inter-satellite load, and ultimately obtain a scheduling scheme that meets the requirements of system communication quality and fairness. Furthermore, traditional centralized ground planning methods suffer from excessive computational complexity when dealing with a massive number of satellite nodes. Therefore, how to quickly and efficiently perform satellite carrier beam hopping scheduling in a giant constellation satellite network is an urgent problem to be solved.

[0003] Patent application CN202310493021.0 discloses a "multi-domain resource allocation method for avoiding interference from low-Earth orbit satellites based on beam hopping," which mainly improves the system's throughput by introducing a deep reinforcement learning network and jointly optimizes time, frequency, and power domain resources. However, this method is limited to resource allocation and avoiding intra-satellite co-frequency interference for a single low-Earth orbit satellite, lacking the ability to jointly optimize multi-domain resources while avoiding inter-satellite interference, and does not guarantee the fairness of user services. Furthermore, the beam hopping resource allocation algorithm it employs has high computational complexity, while the computational capabilities of low-Earth orbit satellites are limited, making the algorithm unsuitable for direct application to low-Earth orbit satellites.

[0004] Patent document CN202311726812.X discloses "A method, device, and medium for joint allocation of multi-domain sensing and multi-dimensional resources for low-Earth orbit satellites with hopping beams." This method primarily avoids inter-satellite co-frequency interference by defining a minimum beam spacing and combines precoding and hopping beam technology to balance inter-beam interference. However, this method fails to efficiently coordinate the allocation of time and frequency resources with inter-satellite interference avoidance in the face of highly dynamic inter-system co-frequency interference and complex network topologies in low-Earth orbit constellation dense beam scenarios. Therefore, it struggles to meet the stringent requirements of low computational overhead in spaceborne environments.

[0005] In their paper "Dynamic Beam Pattern and Bandwidth Allocation Based on Multi-Agent Deep Reinforcement Learning for Beam Hopping Satellite Systems" (IEEE Transactions on Vehicular Technology, 2022, 71(4):3917-3930), Zhiyuan Lin et al. proposed a multi-satellite beam hopping scheduling algorithm that combines load balancing and interference avoidance. In a multi-satellite environment, this algorithm avoids co-channel interference and achieves load balancing by calculating the beam hopping patterns of each satellite. However, the solution complexity of this method increases exponentially with the number of coverage cells and working beams. When facing a massive number of satellite nodes in a giant constellation, this traditional centralized planning method incurs extremely high computational costs and is severely inadequate in responding to sudden service demands.

[0006] Furthermore, some existing satellite autonomous decision-making methods adopted to reduce complexity lack inter-satellite coordination, which not only reduces spectral efficiency but also fails to completely avoid severe interference caused by inter-satellite beam conflicts.

[0007] In summary, existing beam-hopping scheduling and resource allocation methods for giant low-Earth orbit satellite networks generally suffer from the following problems:

[0008] First, using traditional centralized planning or complex deep reinforcement learning algorithms results in an exponential increase in the solution space as the constellation size and the number of covered cells increase, leading to huge computational overhead and making it difficult to meet the real-time processing requirements under limited onboard computing power.

[0009] Secondly, there is a lack of efficient coordination mechanisms in multi-star environments, which are mostly limited to single-star resource allocation or simple beam spacing constraints, and cannot completely avoid severe inter-satellite co-frequency interference in scenarios where giant constellations densely overlap and cover the area.

[0010] Third, the beam resource allocation in business hotspot areas was not effectively combined with inter-satellite load balancing, resulting in local overload and severe imbalance of the overall load of the multi-satellite system. This not only reduced the resource utilization rate of the entire constellation, but also made it difficult to guarantee the fairness of basic coverage of the entire network. Summary of the Invention

[0011] The purpose of this invention is to address the shortcomings of existing technologies in the context of multiple overlapping satellites in mega-constellations, such as high computational complexity of beam hopping scheduling, severe inter-satellite co-frequency interference, and load imbalance. This invention proposes a satellite-ground collaborative beam hopping scheduling method and system for mega-constellation satellite networks, which can quickly and efficiently calculate the multi-satellite beam hopping schedule under limited onboard computing power, thereby improving network throughput and balancing inter-satellite load.

[0012] The core idea to achieve the above objectives is to construct a satellite-ground hierarchical collaborative architecture, which decouples the complex non-convex, NP-hard multi-satellite beam hopping optimization problem in terms of both time scale and algorithm logic to quickly and efficiently calculate the multi-satellite beam hopping schedule; to alleviate inter-satellite and intra-satellite interference and improve throughput by controlling the cell activation probability update mechanism; and to achieve load balancing by allocating satellites to active cells.

[0013] Based on the above ideas, the technical solution of the present invention includes:

[0014] 1. A satellite-ground coordinated beam hopping scheduling method for mega-constellation satellite networks, characterized by comprising:

[0015] (1) Divide the giant constellation into multiple satellite cooperative clusters, and use different frequency bands between adjacent clusters to ensure that the beam scheduling between clusters does not affect each other;

[0016] (2) The ground control center calculates the initial activation probability of cells in the cooperative cluster based on business statistics and satellite network topology, and sends the initial activation probability to the satellite cooperative cluster.

[0017] (3) The satellite cluster receives ground parameters, collects current service requests, and initializes the onboard system status, including the beam hopping pattern matrix, the pre-activation indication matrix, and the global cell selection probability vector;

[0018] (4) In each time slot, the satellite cluster selects candidate cells sequentially through an iterative roulette wheel betting mechanism that includes dynamic probability normalization until the number of selected cells reaches a preset threshold, based on the initial global cell selection probability vector. Finally, it generates the pre-activation indicator vector for the current time slot and updates the global cell selection probability.

[0019] (5) The satellite cluster allocates all pre-activated cells to the visible satellites with the smallest current cumulative load according to the pre-activation indication vector of each time slot and the set of visible satellites of the cell, and outputs the beam transition pattern matrix of each satellite.

[0020] Furthermore, in (2), the ground control center calculates the initial activation probability of cells within the collaborative cluster based on the average service demand intensity of each cell according to the long-term historical service data of the entire network, and determines whether each cell has service demand by combining the visibility relationship between the cell and the satellite in the network topology.

[0021] Furthermore, the initialization of the beam hopping pattern matrix, pre-activation indication matrix, and global cell selection probability vector in (3) involves converting the beam hopping pattern matrix of each satellite into a single matrix. Set to 0; reset the pre-activation indicator matrix. Set to 0; the satellite initializes the global cell selection probability vector based on the received initial activation probability. .

[0022] In step (4), an iterative roulette wheel selection mechanism with dynamic probability normalization is used to sequentially select candidate cells until the number of selected cells reaches a preset threshold, and finally, a pre-activation indicator vector for the current time slot is generated, which includes:

[0023] 4a) Each time slot within the beam-jumping time schedule period Establish a temporary set ;

[0024] 4b) Based on the current global cell selection probability vector The roulette wheel algorithm is used to select candidate cells. join in That is, calculating the cumulative probability distribution Generate random numbers Select the one that satisfies Conditions of the community Add to temporary set ;

[0025] 4c) Select the cell probability Set to 0 and normalize the remaining probabilities;

[0026] 4d) Calculate candidate cells and Minimum spatial distance of existing residential areas And based on the minimum spatial isolation distance threshold for anti-interference The temporary selection probability of the remaining cells Update the probability to obtain the updated probability. ;

[0027] 4e) Repeat steps 4b)-4d) above until the condition is met. Up to now, among them For the number of satellite beams, Number of satellites;

[0028] 4f) Set In the vector, the element at each candidate cell is set to 1, and the elements at the remaining positions are set to 0. A vector of size [size missing] is generated. Preactivation indicator vector ,in This refers to the number of communities.

[0029] 2. A satellite-ground coordinated beam-hopping scheduling system for mega-constellation satellite networks, characterized in that it comprises:

[0030] The ground preprocessing module is used to calculate the initial activation probability vector and the minimum spatial isolation distance threshold for anti-interference of each cell based on long-term historical service data and network topology of the entire network, and to send the above prior parameters to the satellite cooperative cluster.

[0031] The onboard initialization module receives ground parameters and adjusts the service demand vector based on real-time burst service requests within the current coverage area. Initialize the onboard system state, including the beam hopping pattern matrix. Pre-activation indicator matrix and global cell selection probability vector ;

[0032] The pre-activated cell selection module is used by the satellite cluster to select candidate cells sequentially in each time slot based on the initial global cell selection probability vector through an iterative roulette wheel betting mechanism that includes dynamic probability normalization until the number of selected cells reaches a preset threshold, and finally generate the pre-activated indicator vector for the current time slot.

[0033] The load balancing matching module is used to sort all pre-activated cells in descending order of service demand according to the pre-activation indication vector of each time slot, obtain the set of visible satellites that can cover each pre-activated cell in each time slot, calculate the current cumulative load of each satellite in the set, and use a greedy strategy to allocate the cell to the satellite with the lowest current load.

[0034] The output module is used to output the beam hopping pattern matrix of each satellite, which serves as a beam hopping schedule for the giant constellation satellite cluster to collaboratively avoid inter-satellite co-frequency interference and achieve inter-satellite load balancing.

[0035] Compared with the prior art, the present invention has the following advantages:

[0036] Firstly, this invention utilizes a ground control center responsible for extracting long-term macroscopic prior parameters. The satellite constellation, combined with localized real-time bursty services, performs a two-stage lightweight scheduling process on-orbit: "pre-activated cell selection" and "pre-activated cell-satellite matching." This constitutes a hierarchical space-ground collaborative architecture, avoiding the extremely high signaling overhead and transmission latency caused by relying on centralized ground computing for all scheduling decisions. This significantly reduces computational complexity, compressing the total computational complexity of the onboard algorithm to a much lower level. This complexity increases with the number of satellites in the cluster. It changes linearly, becoming This approach overcomes the shortcomings of existing centralized planning methods, where the solution space increases exponentially with constellation size and cannot meet the real-time processing requirements of onboard systems.

[0037] Secondly, this invention introduces an interference avoidance mechanism based on dynamic probability updates during the pre-activation cell selection phase. This mechanism resets the selection probability of activated cells to its initial value and adjusts the selection probability of inactive cells according to service demand driving factors. The cumulative number of inactive time slots The numbers are continuously accumulated, and the minimum spatial isolation distance threshold is used as a basis during the selection of candidate communities. By adjusting the temporary selection probability of the remaining cells proportionally, the shortcomings of existing methods in the dense overlapping coverage scenario of giant constellations cannot be completely avoided due to inter-satellite co-frequency interference. At the same time, it enables cells with high traffic or those that have not been served for a long time to get activated more quickly, ensuring the fairness of cell coverage across the entire network.

[0038] Third, this invention introduces a greedy load balancing strategy in the pre-activated cell-satellite matching stage, which allocates pre-activated cells to visible satellites with the smallest current cumulative load in descending order of service demand. This overcomes the shortcomings of existing methods that fail to effectively combine beam resource allocation in service hotspot areas with inter-satellite load balancing, resulting in local overload and overall load imbalance in multi-satellite systems. This invention can reduce the difference in normalized load between satellites while improving system throughput. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating the implementation of the satellite-ground coordinated beam skipping scheduling method for giant constellation satellite networks according to the present invention.

[0040] Figure 2 This is a schematic diagram of a low-Earth orbit satellite constellation hopping beam scenario in the method of the present invention;

[0041] Figure 3 This is a block diagram of the satellite-ground coordinated beam hopping scheduling system for giant constellation satellite networks according to the present invention. Detailed Implementation

[0042] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0043] Example 1: A satellite-ground coordinated beam hopping scheduling method for mega-constellation satellite networks.

[0044] This example uses Figure 2 Taking the scenario shown as an example, based on the constellation configuration, the giant constellation is divided into multiple cooperative satellite clusters, with frequency reuse between the clusters and the mutual influence between the clusters being ignored. A constellation containing... A cluster of satellites Covering specific ground areas ,common Each satellite in a single time slot can generate a maximum of [number] cells. Each service beam. The scheduling of satellite beams is controlled by a beam transition schedule, with each transition slot lasting for [duration missing]. A beam-hopping timetable cycle contains Each time slot.

[0045] Reference Figure 1 The implementation steps of this example include the following:

[0046] Step 1: Divide the giant constellation into multiple satellite cooperative clusters.

[0047] To achieve efficient satellite-ground collaborative management of mega-constellations, the mega-constellations need to be divided into multiple satellite collaborative clusters, which are specifically implemented as follows:

[0048] Will include Each orbital plane contains [number] orbital surfaces and each orbital contains [number] orbital surfaces. A giant constellation of satellites, based on the principle of minimizing the geometric distance between adjacent satellites in the same orbit and between adjacent orbits, is uniformly divided into several cooperative clusters, each cluster containing... A preset number of satellites.

[0049] After the division, satellites within the same cluster can maintain stable inter-satellite links to support subsequent collaborative computing and data aggregation.

[0050] Step 2: Ground preprocessing of the initial activation probability vector and distribution of satellite-to-ground parameters.

[0051] Based on long-term historical service data and network topology, the ground control center calculates the initial activation probability vector and minimum spatial isolation distance threshold for each cell within the coverage area of ​​the cooperative cluster, and then sends these prior parameters to the corresponding satellite cooperative cluster. This process includes:

[0052] 2.1) Based on the visibility relationship between cells and satellites in the network topology and the characteristics of antenna sidelobe interference, set a minimum spatial isolation distance threshold for anti-interference. This serves as a spatial isolation criterion for the subsequent pre-activated cell selection phase.

[0053] 2.2) Calculate the initial activation probability vector:

[0054] Based on long-term historical service data from the entire network, the average service demand intensity of each cell is statistically analyzed to obtain the cell's... Business needs ;

[0055] Based on the service needs of the network community Calculate the initial activation probability of each cell. : ,in, This represents the total number of cells covered by the collaborative cluster.

[0056] Forming the initial activation probability vector ;

[0057] 2.3) Parameter distribution:

[0058] The ground control center will include the above prior parameters, including the initial activation probability vector. Minimum spatial isolation distance threshold for interference resistance Send to the satellite collaboration cluster.

[0059] In this step, the initial activation probability vector of the ground cell is calculated by the ground control center with a global perspective based on long-term historical service data of the entire network. This provides a priori basis for short-term real-time scheduling on the satellite and avoids the high signaling interaction overhead and transmission latency caused by relying on centralized ground calculation for all scheduling decisions.

[0060] Step 3: Initialize the matrix and vector for onboard hop beam scheduling.

[0061] Set the beam hopping pattern matrix The size is Its elements ); Initialize the beam hopping pattern matrix for each satellite. =0;

[0062] Define the pre-activation indicator matrix for: ,in, Pre-activation indicator matrix The Each element is defined as: ,like The v-th element , indicating the community exist The time slot is selected as the pre-activation cell; the pre-activation indication matrix is ​​initialized. =0;

[0063] Initialize the satellite global cell selection probability vector The initial activation probability received.

[0064] Step 4, pre-activation cell selection.

[0065] Within the beam transition time period, from the time slot Time slot In each time slot, the satellite constellation selects candidate cells sequentially using an iterative roulette wheel mechanism that includes dynamic probability normalization, based on the current global cell selection probability vector, until the number of selected cells reaches a preset threshold. Finally, the pre-activation indication vector for the current time slot is generated, and the global cell selection probability is dynamically updated. This process includes:

[0066] 4.1) Each time slot within the beam-jumping time period Establish a temporary set It is initially empty;

[0067] 4.2) Based on the current global cell selection probability vector The roulette wheel algorithm is used to select candidate cells. join in :

[0068] 4.2.1) Calculate the cumulative probability distribution , of which residential area Current slot activation probability;

[0069] 4.2.2) Generate random numbers ;

[0070] 4.2.3) Select those that satisfy the following conditions Conditions of the community ;

[0071] 4.3) Select the cell probability Set to 0 and normalize the remaining probabilities;

[0072] 4.4) Calculate candidate cells and Minimum spatial distance of existing residential areas And based on the minimum spatial isolation distance threshold for anti-interference The temporary selection probability of the remaining cells Update the probability to obtain the updated probability. :

[0073] ;

[0074] 4.5) Repeat steps 4.2) to 4.4) above until the temporary set is reached. satisfy until;

[0075] 4.6) Set In the vector, the element at each candidate cell is set to 1, and the elements at the remaining positions are set to 0. A vector of size [size missing] is generated. Pre-activation indicator matrix .

[0076] Step 5: Matching pre-activated cells with satellites.

[0077] In obtaining the pre-activation indication matrix of the current beam hopping time period Next, each pre-activated cell needs to be matched with each satellite within the cluster. The goal of the matching process is to meet the requirements of the number of satellite beams. Under the premise of visibility constraints, minimize the difference between the maximum and minimum load within the satellite constellation. Its implementation includes:

[0078] 5.1) All covered cells will be allocated according to service demand. Perform descending sorting to obtain the sorted sequence. ;

[0079] 5.2) For from 1 to Every time slot Each pre-activated cell in the time slot is traversed sequentially in descending order as described in 5.1). That is, satisfying The neighborhood;

[0080] 5.3) For the current pre-activated cell Obtain the set of all visible satellites that can cover the cell at the current moment. ;

[0081] 5.4) Calculate the set The current cumulative payload of each candidate satellite :

[0082] ;

[0083] in, To ensure coverage of the community at the current moment The set of all visible satellites;

[0084] 5.5) Using a greedy strategy, find the satellite with the lowest current cumulative load:

[0085] ;

[0086] 5.6) Pre-activated cells Assigned to target satellite That is, setting the beam jump pattern variables in Set to 1 and update the satellite. The cumulative load;

[0087] 5.7) Repeat steps 5.1)-5.6) to traverse all time slots and pre-activated cells to obtain the pre-activated cell-satellite matching scheme that minimizes inter-satellite load differences, i.e., the beam hopping pattern matrix of each satellite. .

[0088] Step 6: Output beam switching timetable.

[0089] 6.1) Beam hopping pattern matrix of each satellite Calculate satellite No. Beam jump vector of time slot That is, reading the pattern matrix in time slots. No. The value in the row is Each column:

[0090] ;

[0091] 6.2) Combine the beam hopping vectors of each time slot to form a satellite. Beam jump schedule ;

[0092] 6.3) By combining the beam hopping schedules of all satellites in the cluster, a beam hopping schedule for the entire cooperative satellite cluster can be obtained. It then outputs the signal, completing the satellite-ground coordinated beam hopping scheduling.

[0093] It should be noted that the step numbers in this specification and claims are only for the purpose of clearly and completely describing the embodiments of the present invention and for ease of understanding, and their order is not limited.

[0094] Example 2: A satellite-ground coordinated beam hopping scheduling system for giant constellation satellite networks.

[0095] Reference Figure 3 This example includes: a ground preprocessing module 1, a satellite initialization module 2, a pre-activated cell selection module 3, a load balancing matching module 4, and an output module 5. The pre-activated cell selection module 3 includes a probability update submodule 31 and an isolation distance detection submodule 32; the load balancing matching module 4 includes a sorting submodule 41, a visible satellite acquisition submodule 42, a load calculation submodule 43, and a greedy allocation submodule 44.

[0096] The working principle of the entire system is as follows:

[0097] The ground preprocessing module 1 calculates the initial activation probability vector and the minimum anti-interference spatial isolation distance threshold for each cell based on long-term historical service data and network topology, and transmits the initial activation probability vector and the minimum anti-interference spatial isolation distance threshold to the spaceborne initialization module 2.

[0098] The onboard initialization module 2 initializes the beam hopping pattern matrix, pre-activation indication matrix, and global cell selection probability vector of the satellite carrier beam scheduling according to the input of module 1, and then passes them to the pre-activation cell selection module 3.

[0099] The pre-activated cell selection module 3, based on the input from module 2, sequentially selects candidate cells using an iterative roulette wheel selection mechanism incorporating dynamic probability normalization until the selected number reaches a preset threshold. In each round of selection, the isolation distance detection submodule 31 detects the minimum spatial distance between the candidate cell and each cell in the selected cell set, and passes this minimum spatial distance to the probability update submodule 32. After completing the pre-activated cell selection for one time slot, the probability update submodule 32 updates the selection probability of the remaining cells based on the input from the isolation distance detection submodule 31. Then, based on the updated probability distribution, it performs a roulette wheel selection to select a new cell and add it to the selected set. This process is repeated until a predetermined number of cells are selected, and a pre-activated indication vector for the current time slot is generated based on the pre-selected cell set and passed to module 4.

[0100] The load balancing matching module 4, based on the input from module 3, matches each pre-activated cell to each satellite within the cluster. Specifically, the sorting submodule 41 sorts all covered cells in descending order according to service demand and passes the sorting result to the greedy allocation submodule 44; the visible satellite acquisition submodule 42 acquires a set of all visible satellites capable of covering each pre-activated cell and passes the set of visible satellites to the greedy allocation submodule 44; the load calculation submodule 43 calculates the current cumulative load of each candidate satellite in the set of visible satellites and passes the cumulative satellite load to the greedy allocation submodule 44; the greedy allocation submodule 44 assigns the current pre-activated cell to the satellite with the smallest cumulative load in the set of visible satellites and updates the cumulative load of that satellite. Finally, it obtains a pre-activated cell-satellite matching scheme that minimizes inter-satellite load differences and passes it to module 5.

[0101] The output module 5, based on the input from module 4, traverses the pre-activated cell-satellite matching scheme and calculates the beam hopping table for the entire cooperative satellite cluster.

[0102] It should be noted that the above functional modules can be implemented, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, they can be implemented, in whole or in part, as a program instruction product. A program instruction product includes one or a set of program instructions. When the program instructions are loaded and executed on a computer, the described process or function is generated, in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The program instructions can be stored in a computer-readable and writable storage medium, or transferred from one computer's readable and writable storage medium to another.

[0103] In this embodiment, the direct coupling or communication connection between the modules can be achieved through indirect coupling or communication connection via interfaces, devices, or modules. Furthermore, each functional module and submodule can dynamically reside within a single processing unit, or each module can exist physically independently, or two or more modules can dynamically reside within a single processing unit. When these dynamic components are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable and writable storage medium. This storage medium can be a memory, disk, or optical disc, etc.

[0104] The effects of this invention can be further illustrated by the following simulation results:

[0105] I. Simulation Conditions

[0106] A collaborative satellite constellation consisting of 9 satellites will be established, with an orbital altitude of 508 km, a ground cell coverage radius of 35.5 km, a total coverage area of ​​1325 cells, 16 working beams per satellite, a total of 22 time slots for beam hopping schedules, a system bandwidth of 500 MHz, and a total satellite power of 500 W.

[0107] II. Simulation Content and Results

[0108] Under the above simulation conditions, the two-stage beam hopping scheduling method of the present invention and the six existing baseline algorithms R-BH, P-BH, G-BH, IM-BH, SAO-BH and UCO-BH were used to perform beam hopping scheduling on the same satellite constellation, respectively. The system throughput and the number of inactive cells were counted. The results are shown in Table 1.

[0109] Table 1. Comparison of throughput and number of inactive cells for different beam-hopping scheduling methods

[0110]

[0111] In Table 1, R-BH is the random beam hopping algorithm, P-BH is the polling beam hopping algorithm, G-BH is the service-greedy beam hopping algorithm, IM-BH is the interference avoidance beam hopping algorithm, SAO-BH is the satellite autonomous optimization algorithm, and UCO-BH is the unified centralized optimization algorithm.

[0112] As shown in Table 1, the performance of the present invention is significantly better than the six existing baseline algorithms, wherein:

[0113] In terms of system throughput, the method of this invention achieves a system throughput of 396061.0 Mb, which is 15.8%, 16.34%, 111.69%, 10.25%, 13.01%, and 1.16% higher than R-BH, P-BH, G-BH, IM-BH, SAO-BH, and UCO-BH, respectively.

[0114] In terms of coverage fairness, the method of this invention has only 61 cells that are not activated, and the cell coverage rate reaches 95.40%, which is significantly better than other baseline algorithms except P-BH in terms of coverage fairness.

[0115] Regarding inter-satellite load balancing, traditional scheduling methods often lead to significant load differences between satellites due to uneven service demand from ground cells and imbalanced service allocation among satellites. Simulation results show that before the second-stage pre-activated cell-satellite matching optimization proposed in this invention, the normalized load difference between satellites is as high as 0.1459; however, after optimization using the greedy matching strategy proposed in this invention, the maximum normalized load difference is significantly reduced to 0.1074, a reduction of 26.38% in inter-satellite load difference. This indicates that this invention not only improves network capacity but also effectively alleviates the load imbalance problem between satellites.

[0116] Regarding computational complexity, the solution complexity of existing centralized scheduling methods increases exponentially with the number of satellites and coverage cells. This invention, however, reduces the total computational complexity of the algorithm to [a much lower level] through satellite-ground hierarchical coordination and two-stage decoupling. This complexity increases with the number of satellites in the cluster. It changes linearly, that is This significantly reduces computational overhead and is able to adapt to the limited computing power of giant constellation-borne equipment.

[0117] It should be noted that the above description is merely a specific example of the present invention and does not constitute any limitation on the present invention. Obviously, those skilled in the art, after understanding the content and principles of the present invention, may make various modifications and changes in form and detail without departing from the principles and structure of the present invention. For example, in addition to being applicable to a low-Earth orbit satellite network with 9 satellites and an orbital altitude of 508km in this embodiment, the present invention can also be applied to mega-satellite networks with other constellation configurations and parameter settings; besides adopting a two-stage sequential execution scheme of first selecting pre-activated cells and then matching pre-activated cells with satellites, the present invention can also iteratively optimize the two stages according to actual system requirements. However, these modifications and changes based on the ideas of the present invention are still within the scope of protection of the claims of the present invention.

Claims

1. A satellite-ground cooperative beam hopping scheduling method for mega-constellation satellite networks, characterized in that, include: (1) Divide the giant constellation into multiple satellite cooperative clusters, and use different frequency bands between adjacent clusters to ensure that the beam scheduling between clusters does not affect each other; (2) The ground control center calculates the initial activation probability of cells in the cooperative cluster based on business statistics and satellite network topology, and sends the initial activation probability to the satellite cooperative cluster. (3) The satellite cluster receives ground parameters, collects current service requests, and initializes the onboard system status, including the beam hopping pattern matrix, the pre-activation indication matrix, and the global cell selection probability vector; (4) In each time slot, the satellite cluster selects candidate cells sequentially through an iterative roulette wheel betting mechanism that includes dynamic probability normalization until the number of selected cells reaches a preset threshold, based on the initial global cell selection probability vector. Finally, it generates the pre-activation indicator vector for the current time slot and updates the global cell selection probability. (5) The satellite cluster allocates all pre-activated cells to the visible satellites with the smallest current cumulative load according to the pre-activation indication vector of each time slot and the set of visible satellites of the cell, and outputs the beam transition pattern matrix of each satellite.

2. The method according to claim 1, characterized in that, The division of the mega-constellation into multiple satellite cooperative clusters, as described in (1), involves... Each orbital plane contains [number] orbital surfaces and each orbital contains [number] orbital surfaces. A giant constellation of satellites, based on the principle of minimizing the geometric distance between adjacent satellites in the same orbit and between adjacent orbits, is uniformly divided into several cooperative clusters, each cluster containing... A preset number of satellites.

3. The method according to claim 1, characterized in that, In step (2), the ground control center calculates the initial activation probability of cells within the collaborative cluster. This is based on the average service demand intensity of each cell according to long-term historical service data of the entire network, and combined with the visibility relationship between cells and satellites in the network topology to determine whether each cell has service demand. The calculation formula is as follows: ; in, For the community The initial activation probability, This represents the total number of cells covered by the collaborative cluster. For the community The business requirement is to assign an equal initial activation probability to cells with business requirements and to set the initial activation probability to zero for cells without business requirements.

4. The method according to claim 1, characterized in that, The initialization of the beam hopping pattern matrix, pre-activation indication matrix, and global cell selection probability vector in (3) involves setting the beam hopping pattern matrix of each satellite. Set to 0; reset the pre-activation indicator matrix. Set to 0; the satellite initializes the global cell selection probability vector based on the received initial activation probability. .

5. The method according to claim 1, characterized in that, In step (4), an iterative roulette wheel selection mechanism with dynamic probability normalization is used to sequentially select candidate cells until the number of selected cells reaches a preset threshold, and finally, a pre-excitation indication vector for the current time slot is generated, which includes: 4a) Each time slot within the beam-jumping time schedule period Establish a temporary set ; 4b) Based on the current global cell selection probability vector The roulette wheel algorithm is used to select candidate cells. join in : 4b1) Calculate the cumulative probability distribution , of which residential area Current slot activation probability; 4b2) Generate random numbers ; 4b3) Select the option that satisfies Conditions of the community ; 4c) Select the cell probability Set to 0 and normalize the remaining probabilities; 4d) Calculate candidate cells and Minimum spatial distance of existing residential areas And based on the minimum spatial isolation distance threshold for anti-interference The temporary selection probability of the remaining cells Update the probability to obtain the updated probability. Its update method is as follows: ； 4e) Repeat steps 3b)-3d) above until the condition is met. Up to now, among them Satellite beam count Number of satellites; 4f) Set In the vector, the element corresponding to each candidate cell is set to 1, and the elements at the other positions are set to 0, generating a vector of size . Preactivation indicator vector ,in This refers to the number of communities.

6. The method according to claim 1, characterized in that, The dynamic update of the global cell selection probability in (4) involves resetting the selection probability of activated cells to its initial value and continuously accumulating the selection probability of inactive cells. The update formula is expressed as: ， in, For the community In the The activation probability of a time slot For the community In the Pre-activation indication of time slots, For the community Business demand drivers For the community business needs, This represents the cumulative number of time slots that have not been activated in this cell.

7. The method according to claim 1, characterized in that, In (5), the satellite cluster allocates all pre-activated cells to the visible satellites with the lowest current cumulative load based on the pre-activation indicator vector of each time slot and the set of visible satellites in each cell, according to service demand. This includes: (5a) All covered cells are arranged according to service demand. Sort the cells in descending order to obtain the cell sorting sequence. , To be ranked in The community ID is located in the specified position. (5b) For each time slot Iterate through each satisfying condition in descending order. pre-activated cells ; (5c) For the current pre-activated cell Obtain the set of all visible satellites that can cover the cell. Calculate the current cumulative payload of each candidate satellite in the set using the following formula. : , in The number of time slots, For the number of cells, Indicates time slot Lower Community With satellite The correlation variable, when it is 1, indicates that the two are related; (5d) according to The satellite with the lowest cumulative load is identified, and the currently pre-activated cell is assigned to that satellite. And update the satellite's cumulative payload; (5e) Traverse all time slots and pre-activated cells to obtain a pre-activated cell-satellite matching scheme that minimizes inter-satellite load differences.

8. A satellite-ground coordinated beam-hopping scheduling system for mega-constellation satellite networks, characterized in that, include: The ground preprocessing module is used to calculate the initial activation probability vector and the minimum spatial isolation distance threshold for anti-interference of each cell based on long-term historical service data and network topology of the entire network, and to send the above prior parameters to the satellite cooperative cluster. The onboard initialization module receives ground parameters and adjusts the service demand vector based on real-time burst service requests within the current coverage area. Initialize the onboard system state, including the beam hopping pattern matrix. Pre-activation indicator matrix and global cell selection probability vector ; The pre-activated cell selection module is used by the satellite cluster to select candidate cells sequentially in each time slot based on the initial global cell selection probability vector through an iterative roulette wheel betting mechanism that includes dynamic probability normalization until the number of selected cells reaches a preset threshold, and finally generate the pre-activated indicator vector for the current time slot. The load balancing matching module is used to sort all pre-activated cells in descending order of service demand according to the pre-activation indication vector of each time slot, obtain the set of visible satellites that can cover each pre-activated cell in each time slot, calculate the current cumulative load of each satellite in the set, and use a greedy strategy to allocate the cell to the satellite with the lowest current load. The output module is used to output the beam hopping pattern matrix of each satellite, which serves as a beam hopping schedule for the giant constellation satellite cluster to collaboratively avoid inter-satellite co-frequency interference and achieve inter-satellite load balancing.

9. The system according to claim 8, characterized in that: The pre-activated cell selection module includes: The isolation distance detection submodule is used to calculate the minimum spatial distance between a candidate cell and each cell in the selected cell set during the candidate cell selection process. , and according to Update the provisional selection probability of the remaining cells, where, Minimum spatial isolation distance threshold for interference resistance The original probability of the neighborhood being selected. The updated alternative probabilities; The probability update submodule is used to reset the selection probability of the activated cell to its initial value after the pre-activated cell selection for a time slot is completed. For cells that have not been activated, Accumulate their selection probabilities, where, For the community In the The activation probability of a time slot For the community Business demand drivers For the community business needs, This represents the cumulative number of inactive time slots in the cell. The load balancing matching module includes: The sorting submodule is used to sort all coverage cells in descending order according to service demand. The visible satellite acquisition submodule is used to acquire the set of all visible satellites that can cover each pre-activated cell. The load calculation submodule is used to calculate the current cumulative load of each candidate satellite in the visible satellite set; The greedy allocation submodule is used to allocate the currently pre-activated cell to the satellite with the lowest cumulative load and update the cumulative load of that satellite.

Images