Beam hopping pattern design method of low earth orbit satellite

Abstract

A method for designing a beam hopping pattern of a low-orbit satellite belongs to the field of low-orbit satellite communication, and comprises the following steps: dynamically clustering ground cells by using a hybrid clustering cell clustering algorithm according to the service requirements of the cells in a beam hopping scene, and enabling the cells in the same cluster to approach as much as possible on the basis of load balancing; dividing user satisfaction indexes into three levels to guide three dynamic operations; selecting an optimal individual to generate a progeny by using the overall user satisfaction of all cells in the whole beam hopping period; performing individual gene crossover by using the community user satisfaction of each cluster; performing individual gene variation by using the user satisfaction of a single cell; and fine tuning of beam power is carried out when each beam of the beam hopping pattern is served in the same time slot. According to the method, load balancing and geographic position compactness are comprehensively considered, the same-frequency interference is reduced, the resource utilization efficiency is improved, the calculation overhead is reduced, and the overall calculation efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of low-orbit satellite communications, and in particular relates to a beam hopping pattern design method for a low-orbit satellite. Background Art

[0002] In low-orbit satellite communication systems, due to the high-speed movement of satellites and the limited number of beams, there is a significant mismatch between the user needs and distribution of ground cells. The number of beams generated by the onboard phased array system of low-orbit satellites is much smaller than the number of ground cells, resulting in the inability to fully cover all visible cells at the same time. Therefore, how to efficiently use limited beam resources to meet the communication needs of ground users has become an urgent problem to be solved. In order to improve the system service quality, researchers have proposed beam hopping technology, which switches beams in different time periods through time slicing to improve coverage. However, this frequent beam switching introduces additional communication overhead, affecting the user's communication experience. Therefore, designing efficient beam hopping patterns to balance user satisfaction and communication overhead is the key to current research.

[0003] The existing technology is mainly achieved through the following two methods:

[0004] (1) Clustering methods based solely on distance or load: Traditional clustering methods usually use the K-means algorithm to cluster ground cells. K-means allocates data points by minimizing the sum of squared distances within each cluster, but it is easy to fall into local optimality when dealing with complex user needs and geographical distribution, and there is no strict control over the shape and size of the cluster, which may cause co-channel interference. In addition, K-means generally does not consider load balancing issues and only focuses on geographical compactness.

[0005] (2) Beam hopping optimization scheme based on genetic algorithm: Beam hopping pattern optimization usually uses genetic algorithm. Genetic algorithm simulates natural selection and genetic mechanism and gradually evolves to obtain a beam allocation scheme close to the optimal solution. Although genetic algorithm has strong global search ability, it may face greater randomness and slow convergence speed when dealing with beam hopping problems. Especially in complex situations considering user satisfaction and communication overhead, traditional genetic algorithm may not be able to solve the problem efficiently.

[0006] Therefore, a beam hopping pattern must be designed that can efficiently allocate beams under limited resource conditions to maximize user satisfaction and minimize the additional overhead caused by beam switching. This design goal is to take beam switching into account when optimizing the allocation of beam resources, improve system performance, and ensure a good user experience in the coverage area of ​​high-speed mobile satellites. Summary of the invention

[0007] In order to solve the problems in the prior art that the K-means algorithm is prone to fall into local optimality and ignores load balancing, resulting in inefficient beam resource utilization, as well as the traditional genetic algorithm's high randomness, poor adaptability, slow convergence speed, and poor optimization effect, and the frequent clustering calculations increase the computational overhead and reduce the algorithm efficiency, the present invention provides a beam hopping pattern design method for a low-orbit satellite.

[0008] The present invention aims to comprehensively consider load balancing and geographical location compactness by improving the clustering algorithm, thereby reducing co-frequency interference and improving resource utilization efficiency. At the same time, a dynamic adjustment mechanism of user satisfaction is introduced to enhance the adaptability and efficiency of the genetic algorithm. Finally, unnecessary clustering calculations are reduced through a user distribution similarity judgment mechanism, thereby reducing computing overhead and improving overall computing efficiency.

[0009] The technical solution adopted by the present invention to solve the technical problem is as follows:

[0010] The present invention provides a method for designing a beam hopping pattern of a low-orbit satellite, which specifically comprises the following steps:

[0011] Step S1: In the beam hopping scenario, the ground cells are dynamically clustered using a hybrid clustering cell clustering algorithm according to the service requirements of the cells, so that the cells in the same cluster are as close as possible on the basis of load balancing;

[0012] Step S2: using a genetic algorithm with dynamic operation based on user satisfaction to perform time slot and beam scheduling;

[0013] The user satisfaction index is divided into three levels to guide three dynamic operations: ① selection operation: using the overall user satisfaction of all cells in the entire beam hopping period to select the best individual to generate offspring; ② crossover operation: using the user satisfaction of each cluster cell to perform individual gene crossover; ③ mutation operation: using the user satisfaction of a single cell to perform individual gene mutation;

[0014] Step S3: fine-tune the beam power of each beam of the beam hopping pattern when serving in the same time slot.

[0015] Furthermore, in step S1, the output of the K-means clustering algorithm is used as the initial solution of the simulated annealing algorithm to explore possible clustering assignments globally.

[0016] Furthermore, in step S1, the specific implementation process of the hybrid clustering cell clustering algorithm is as follows:

[0017] S1.1: Assume that the number of ground cells is X, the number of beams provided by the satellite phased array is K, and the goal is to divide X cells into K clusters; set the objective function as: value krepresents the sum of the traffic of the kth cluster, average represents the average traffic of the cluster, and max_dis k Represents the distance between the two farthest cells in the cluster after normalization;

[0018] S1.2: Use K-means clustering algorithm to find the initial solution;

[0019] S1.3: Use the output of the K-means clustering algorithm as the initial solution of the simulated annealing algorithm to explore possible clustering assignments globally;

[0020] S1.4: Determine the similarity of the user distributions of the two beam-hopping periods, that is, substitute the user distribution of the ε+1th period into the clustering result of the εth period. If the standard deviation σ is satisfied ε+1 ≤(1+ξ)σ ε , ξ represents the maximum acceptable floating value of the standard deviation, then the user distribution identity is considered to be satisfied; otherwise, it is considered not to be satisfied.

[0021] Furthermore, the acceptable maximum floating value ξ of the standard deviation does not exceed 10%.

[0022] Furthermore, in the genetic algorithm, when creating the initial population, each individual is a possible solution, so the generation of individuals is related to the services of users in the cell. At the same time, considering that each cell must be served by at least one time slot, service priority and cell polling are integrated.

[0023] Furthermore, the calculation formula for the overall user satisfaction of all cells is:

[0024]

[0025] Among them, H t represents the beam hopping pattern matrix in the tth time slot; C t represents the beam capacity that the satellite-borne phased array beam can actually serve in the cell pointed to in the tth time slot; T represents the number of time slots contained in a beam hopping cycle; β represents the weighting weight; K represents the number of beams; φ t represents the number of data packets in each cell at time t; R kt represents the element in the kth row and tth column of the matrix R. Its physical meaning is whether the kth beam in the tth time slot has changed relative to the beam pointing in the t-1th time slot. kt =0 means that the beam direction does not change, that is, the beam will not switch. kt ≠0 means that the beam direction has changed, that is, the beam has been switched.

[0026] Furthermore, assuming that the switching overhead of each beam switching is the same, the clustering matrix is ​​a K×X matrix P, which represents the result of cell clustering; assuming that the beam hopping pattern matrix is ​​an M×T matrix H, the order of serving cells of each beam in the entire beam hopping period is obtained through P×H, and P×H×(IB) is the beam switching matrix in a beam hopping period, where B is a determination matrix transformed from the unit matrix, I is the unit matrix, and (IB) indicates whether the cell pointed to by the beam in the current time slot is changed compared with the previous time slot; suppose the matrix R=P×H×(IB), then the number of switching times is the sum of the number of elements in the matrix R that are not 0:

[0027]

[0028] Where K represents the number of beams; T represents the number of time slots contained in a beam hopping cycle; R kt represents the element in the kth row and tth column of the matrix R. Its physical meaning is whether the kth beam in the tth time slot has changed relative to the beam pointing in the t-1th time slot. kt =0 means that the beam direction does not change, that is, the beam will not switch. kt ≠0 means that the beam direction has changed, that is, the beam has been switched.

[0029] Furthermore, the number of elements and cost are normalized to their maximum values, Cost = cost / (K×T), K×T beam switching means that K beams will switch once in each time slot; considering the two system indicators of user satisfaction and switching overhead, the optimization goal is:

[0030]

[0031] Among them, β represents the weighted weight; K represents the number of beams; T represents the number of time slots contained in a beam hopping cycle; φ t represents the number of data packets in each cell at time t; R kt represents the element in the kth row and tth column of the matrix R. Its physical meaning is whether the kth beam in the tth time slot has changed relative to the beam pointing in the t-1th time slot. kt =0 means that the beam direction does not change, that is, the beam will not switch. kt ≠0 means that the beam direction has changed, that is, the beam has been switched.

[0032] Furthermore, the calculation formula of the cell user satisfaction of each cluster is:

[0033]

[0034] Among them, H t,m represents the beam hopping pattern matrix of the mth cluster in the tth time slot, Ct,m represents the actual beam capacity that the satellite-borne phased array beam pair of the mth cluster can serve in the tth time slot. t,m represents the user demand of the mth cluster in the tth time slot, P k represents the cell set of the kth cluster, and T represents the number of time slots contained in a beam hopping cycle.

[0035] Furthermore, the calculation formula of the user satisfaction of a single cell is:

[0036]

[0037] Among them, H t,m represents the beam hopping pattern matrix of the mth cell in the tth time slot, C t,m represents the beam capacity that can actually be served by the satellite-borne phased array beam pair pointing to the mth cell in the tth time slot, φ t,m It represents the user demand of the mth cell in the tth time slot, and T represents the number of time slots contained in a beam hopping cycle.

[0038] The beneficial effects of the present invention are:

[0039] The present invention provides a method for designing beam-hopping patterns of low-orbit satellites, which solves the problem that the number of low-orbit satellite beams is much smaller than the number of ground cells and cannot provide full coverage. The clustering technology is used to reduce the communication overhead caused by beam switching, thereby improving user satisfaction and system resource utilization efficiency.

[0040] The present invention significantly improves resource utilization efficiency under the condition of limited beam resources of low-orbit satellites by optimizing cell clustering and beam-hopping pattern design. At the same time, the present invention adopts a clustering method based on load balancing and geographic location to reduce co-channel interference, and avoids the trap of local optimal solution by combining K-means with simulated annealing algorithm, thereby improving the balance and compactness of clustering. In addition, the present invention introduces a hierarchical dynamic adjustment genetic algorithm based on user satisfaction, making the beam-hopping pattern design more adaptive and flexible, effectively reducing the signaling overhead of the system while meeting user needs.

[0041] In addition, in order to further improve the robustness of the system, the present invention also introduces a judgment criterion based on user distribution similarity. When the user distribution is similar in adjacent beam hopping periods, there is no need to re-cluster, and the existing optimization results are used for rapid iteration, reducing unnecessary computational overhead.

[0042] In summary, the present invention not only improves the overall performance of the satellite communication system, but also significantly improves user satisfaction and system stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1A flow chart of a beam hopping pattern design method for a low-orbit satellite provided by the present invention.

[0044] Figure 2 The present invention provides an application scenario of a beam-hopping pattern design method for a low-orbit satellite.

[0045] Figure 3 To compare the clustering results before and after optimization. A is the cell division map and random points within the satellite coverage; B is the cell division map and random points within the satellite coverage using the clustering result based on geographic location; C is the cell division map and random points within the satellite coverage using the result of hybrid clustering cell clustering.

[0046] Figure 4 This is the relationship curve between user satisfaction, switching cost and β.

[0047] Figure 5 The simulation results of the core beam-hopping algorithm are shown in Figure 1. A is the iterative result of the genetic algorithm; B is the beam-hopping pattern; C is the comparison of cell satisfaction of different algorithms; and D is the comparison of cluster satisfaction of different algorithms.

[0048] Figure 6 The influence of different clustering algorithms on the final satisfaction. Among them, A is the iterative result of the genetic algorithm based on distance clustering, and B is the iterative result of the genetic algorithm based on hybrid clustering cell clustering.

[0049] Figure 7 This is the simulation result of the improved NSGA-Ⅱ algorithm. DETAILED DESCRIPTION

[0050] The present invention is further described in detail below in conjunction with the accompanying drawings.

[0051] like Figure 1 As shown, the present invention mainly designs the entire process of beam hopping pattern design (only considering the downlink user link). The onboard processor of the low-orbit satellite knows the needs and distribution of ground users, and needs to make efficient use of limited satellite resources based on the satellite's own capabilities (calibrated phased array power, beam coverage capability). Therefore, the evaluation indicators are overall user satisfaction and signaling overhead (including but not limited to other costs).

[0052] In a method for designing a beam hopping pattern of a low-orbit satellite of the present invention, a cell clustering algorithm is first used to cluster ground cells. In the present invention, the number of cells is represented by X, and the number of users is represented by N. It is known that the number of cells X is much larger than the number of satellite beams K, that is, X>>K, and K≥2. If the beam can be switched between all cells in any time slice, then the search space is (C is the representation of the number of combinations in mathematics, which means the number of combinations of K elements taken from X different elements, expressed as the search space). Therefore, in order to improve the search efficiency of the algorithm, the ground cells are first clustered according to user needs and their distribution to obtain clustering results. When clustering, it can be based on the principle of cell location, cell load balancing, or a combination of the two. The final number of clusters is equal to the number of beams. When designing the beam hopping pattern, the selection of beams will be restricted, which is equivalent to each beam can only jump within its own cluster. The search space after clustering is X, which can reduce the search space and improve the search efficiency.

[0053] For the clustering results, it is necessary to determine whether the users meet the distribution identity. Assume that the minimum time slice of beam hopping is s (s is a parameter at the ms level, such as 1ms), and a beam hopping period S=M*s, which means that there are M minimum time slices in a beam hopping period. Obtain the user distribution in the ε+1th beam hopping period at least one period in advance, and compare it with the user distribution in the εth period to determine whether the user distribution in adjacent periods is the same. If they are the same, there is no need to re-cluster the ground cells, and the final solution set of the εth period can be used as the initial solution set of the ε+1th period. Under the premise that the initial solution is relatively good, the iteration time will be greatly shortened; if they are different, the ground cells need to be re-clustered, and then the core algorithm of the beam hopping pattern is solved.

[0054] The following will be combined with the simulation examples of the present invention to clearly and completely describe the technical solutions in the simulation examples of the present invention. Obviously, the simulation examples described are only part of the simulation examples of the present invention, rather than all the simulation examples. Based on the simulation examples in the present invention, all other simulation examples obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0055] The present invention provides a beam hopping pattern design method for a low-orbit satellite, and its specific implementation process is as follows:

[0056] 1. Description of scenarios and models;

[0057] Beam hopping scenarios such as Figure 2 As shown, assuming that the satellite has K beams B = {b k |k=1,2…,K}, and 5 green cells are lit at the same time, indicating that the satellite has 5 beams; Assume that there are N users U={u n |n=1,2,..N} distributed in X cells C={c x|x=1,2,…,X}, where each circle on the ground represents a cell, which is usually divided into regular hexagons, but the beams are usually pointed in a circular or elliptical direction; the satellite uses K beams to cover X cells through time division multiplexing. The arrival flow of each user at time t is User n The request data packet at time t follows the parameters According to the Poisson definition and the cell where the user is located, the number of data packets in each cell at time t is obtained

[0058] In each time slice, the satellite needs to select the cell covered by the beam according to the data packet, channel conditions and payload parameters of each cell. Assuming that the cell served in time slice t is express, Indicates the serving cell c in this time slice x , Indicates that no service is provided during this time slice.

[0059] For the cell c assigned to the beam x The signal-to-noise ratio (SINR) calculation formula at time t is:

[0060]

[0061] Cell c x The beam capacity is:

[0062]

[0063] The overall satisfaction of X communities is obtained as follows:

[0064]

[0065] Among them, b k represents the kth beam, b i represents the i-th beam, Indicates that in the tth time slot, cell c x and beam b k The channel gain between Indicates that in the tth time slot, cell c x and beam b i The channel gain between (i∈C and i≠x indicates the calculation of cell c x The gain brought by other beams interferes with the channel quality), Indicates that beam b in time slot t k Point to cell c x The beam transmission power, Indicates that beam b in time slot t iPoint to cell c x The beam transmission power (i∈C and i≠x), N0 represents the Gaussian white noise power spectral density in watts per hertz (W / Hz), B w represents the channel bandwidth, Indicates the serving cell c in this time slice i , H t represents the beam hopping pattern matrix (or vector) at the tth time slot, C t It represents the beam capacity that the satellite-borne phased array beam can actually serve when pointing to the cell in the tth time slot, and T represents the number of time slots contained in a beam hopping cycle.

[0066] Each switch of each beam will cause a certain switching overhead, so if you want to reduce the switching overhead, you need to reduce the number of beam switches as much as possible. Assuming that the switching overhead of each beam switch is the same, the clustering matrix is ​​a K×X matrix P, which represents the result of cell clustering. Assuming that the beam hopping pattern matrix is ​​an M×T matrix H, the service cell order of each beam in the entire beam hopping cycle is obtained through P×H. P×H×(IB) is the beam switching matrix within a beam hopping cycle, where B is the determination matrix transformed from the unit matrix, I is the unit matrix, and (IB) indicates whether the cell pointed to by the beam in the current time slot has changed compared to the previous time slot. Assume that the matrix R = P×H×(IB), then the number of switches is the sum of the number of elements in the matrix R that are not 0:

[0067]

[0068] Among them, R kt Represents the element in the kth row and tth column of the matrix R. Its physical meaning is whether the kth beam in the tth time slot has changed relative to the beam in the t-1th time slot. kt = 0 means that the beam pointing does not change, which means that no switching will occur. kt ≠0 means that the beam direction has changed, which means that beam switching has occurred. The number of elements and cost are normalized to the maximum value, Cost = cost / (K×T), K×T times (maximum switching times) beam switching means that K beams will have a beam switching in each time slot.

[0069] Considering the two system indicators of user satisfaction and switching overhead, the optimization goal is:

[0070]

[0071] Among them, β is the weighted weight. Since the user satisfaction index is as high as possible for the beam-hopping system, and the switching overhead is as small as possible, the weights of the two system indicators are normalized and then subtracted. In addition, in the full frequency reuse beam-hopping system, there are two constraints:

[0072] ① For each satellite, the upper limit of the number of beams available in the same time slot is K, that is

[0073] ②For each satellite, the power and Does not exceed the total system power P, that is

[0074] 2. Algorithm design;

[0075] Algorithm design such as Figure 1 As shown in the figure, the whole algorithm is divided into three steps. First, the ground cells are dynamically clustered according to the cell demand. The role of clustering is to make the beam service capabilities as similar as possible, that is, load balancing between clusters, and at the same time, it can greatly reduce the search space of subsequent algorithms. Secondly, the genetic algorithm is used to flexibly schedule time slots and beams. When the initial population is generated, the cell service priority and polling method are considered. The purpose is to ensure the service priority and ensure that all cells with services can be served within the beam hopping cycle. The genetic algorithm as a whole uses three-level fitness as the guide of genetic operations, so that the algorithm can give the best beam hopping pattern in a shorter time. Finally, the beam power of each beam of the beam hopping pattern is fine-tuned when serving in the same time slot. At this time, the beam hopping pattern is no longer changed, and only the beam power is fine-tuned. The ultimate goal of fine-tuning is to match the demand with the beam capacity as much as possible, and the sum of the powers of each beam does not exceed the power provided by the satellite as a whole.

[0076] (1) Hybrid clustering cell clustering algorithm;

[0077] Since the satellite covers a large area, the coverage radius of a satellite at an altitude of 550km is approximately 600km. Hexagonal cells are used to divide the cells within the coverage area. However, due to the distribution of users, the service demands of each cell are extremely uneven. Since the number of cells X is much larger than the number of beams K, the cells need to be clustered.

[0078] There are two known traditional clustering methods. One is clustering based on geographical location, which can effectively reduce co-channel interference; the other is clustering based on load balancing, which can improve beam utilization. The present invention makes the cells in the same cluster as close as possible (considering spatial compactness) on the basis of load balancing, which can effectively reduce the impact of co-channel interference. This problem is a typical clustering problem, but the traditional K-means algorithm can easily fall into a local optimal solution. Therefore, the present invention uses the output of K-means as the initial solution of the simulated annealing algorithm, giving the simulated annealing algorithm with strong global search capabilities an initial optimal solution, and exploring possible clustering allocations on a global scale.

[0079] Therefore, the specific implementation process of the hybrid clustering cell clustering algorithm is as follows:

[0080] 1) Assuming that the number of ground cells is X and the number of beams provided by the satellite phased array is K, it is necessary to divide the X cells into K clusters to limit the range of beam selection cells, thus narrowing the selection space while ensuring performance.

[0081] The objective function is:

[0082] Among them, value k represents the sum of the traffic of the kth cluster, average represents the average traffic of the cluster (total traffic divided by the number of clusters), max_dis k Represents the distance between the two farthest cells in the cluster after normalization. Since the center position of each cell is known, normalization is performed by dividing the distance between the two farthest cells in the cluster by the distance between the two farthest cells in the coverage area.

[0083] In order to achieve the goal of load balancing and reducing co-channel interference, the objective function takes into account the standard deviation of the load and the sum of the maximum distances between cells within the cluster, ensuring the balance and compactness of the clustering. That is, while ensuring that the user demand of cells within each cluster is as uniform as possible, the compactness of the clustering in terms of geographical location is also guaranteed.

[0084] 2) Use the K-means clustering algorithm to find the initial solution (only considering spatial compactness, that is, the clustering criterion is to minimize the distance from the points within the cluster to the cluster center).

[0085] 3) Using the output of the K-means clustering algorithm as the initial solution of the simulated annealing algorithm, we explore possible clustering assignments globally. The simulated annealing algorithm takes into account load balancing and spatial compactness; the initial clustering results are optimized by the simulated annealing algorithm to find a better clustering configuration.

[0086] 4) Finally, the similarity of the user distributions of the two beam-hopping periods is determined by substituting the user distribution of the ε+1th period into the clustering result of the εth period. If the standard deviation σ is satisfied, ε+1 ≤(1+ξ)σ ε ξ represents the maximum acceptable floating value of the standard deviation, then the user distribution is considered to be satisfied; otherwise, it is considered not satisfied. Among them, after experimental analysis, the range of the maximum acceptable floating value ξ of the standard deviation should not exceed 10%.

[0087] The present invention divides clusters according to the needs of each cell first, which is better than distance-based clustering because the beam hopping pattern will be limited by the clustering results. And the more uneven the user distribution is, the better the algorithm performs. At the same time, the present invention adopts a method of judging the similarity of user distribution by comparing whether the standard deviation of user distribution of adjacent beam hopping periods is less than 10%. This user distribution similarity judgment criterion simplifies the transition between beam hopping periods, reduces system calculation overhead, and ensures the continuity and consistency of user services.

[0088] In the process of cell clustering, load balancing and geographical location are combined, so that cells in the same cluster are balanced in load and more compact in geographical location, effectively reducing co-frequency interference. This innovation breaks through the limitation of the traditional K-means algorithm that is prone to fall into local optimality, and provides a better global search effect by using the results of K-means as the initial solution of the simulated annealing algorithm.

[0089] (2) Beam-hopping pattern core algorithm (genetic algorithm with dynamic operation based on user satisfaction);

[0090] The goal of the core algorithm of the beam hopping pattern is to maximize overall user satisfaction and minimize signaling overhead. Since the result of the beam hopping pattern optimization is discrete, the genetic algorithm is the most suitable, but the traditional genetic algorithm is too random and cannot meet the requirements of the current problem. Therefore, the present invention adopts dynamic operations based on user satisfaction on the basis of the genetic algorithm to improve adaptability and flexibility. Specifically, user satisfaction is used as an important indicator and divided into three levels to guide three actions: using the overall user satisfaction of all cells in the entire beam hopping cycle to select the best individual to produce offspring; using the user satisfaction of each cluster of cells to perform individual gene crossover; using the user satisfaction of a single cell to perform individual gene mutation. This design not only improves the algorithm's ability to respond to user needs, but also can dynamically adjust the search direction, so that the algorithm can meet user needs while minimizing system overhead. Among them, satisfaction is limited by demand, because the overall throughput of the satellite is limited by power, phased array gain, and bandwidth, so resources determine the upper limit that can be provided. The purpose of the algorithm is to make limited resources reasonably utilized, neither wasteful nor redundant.

[0091] Genetic algorithms are used for flexible scheduling of time slots and beams. The cell service priority and polling method are considered when the initial population is generated. The purpose is to ensure the service priority and ensure that all cells with services can be served within the beam hopping cycle. The genetic algorithm as a whole uses three-level fitness as the guide for genetic operations, so that the algorithm can give the best beam hopping pattern in a shorter time. It specifically includes the following steps:

[0092] 1) When creating the initial population, each individual is a possible solution, so the generation of individuals is related to the services of users in the cell. At the same time, considering that each cell must be served by at least one time slot, the service priority and cell polling will be considered. Among them, whether the initial population is generated by polling depends on the user distribution. When the user distribution is extremely uneven, setting the initial population to polling will not make the effect more ideal.

[0093] 2) Fitness calculation;

[0094] The specific implementation process is as follows:

[0095] Since the user satisfaction evaluation is performed for a beam hopping cycle, and a cycle contains T time slices, K cells must be selected from X cells in each time slice to align the beam (X is much larger than K). Each selection space is very large, so the fitness needs to calculate the overall satisfaction of all cells (Formula 1.6), the user satisfaction of each cluster (Formula 1.7) and the user satisfaction of each cell (Formula 1.8) for reference in other genetic operations. The specific formula for calculating user satisfaction is as follows:

[0096]

[0097] Among them, H t,m represents the beam hopping pattern matrix of the mth cluster (or cell) in the tth time slot, C t,m represents the beam capacity that the satellite-borne phased array beam pair can actually serve in the mth cluster (or cell) at the tth time slot, φ t,m represents the user demand of the mth cluster (or cell) in the tth time slot, P k represents the cell set of the kth cluster (the maximum value of k is K, i.e., the number of beams), and T represents the number of time slots contained in a beam hopping cycle.

[0098] 3) Selection operation: This process is to select excellent individuals according to the roulette wheel method based on individual fitness and pass them on to the next generation. Therefore, the judgment criterion is the overall user satisfaction of the individual (Formula 1.6).

[0099] 4) Crossover operation: This process refers to the random exchange of some chromosomes by individuals. Since random exchange will not satisfy the requirement that only one beam serves one cell in each time slice of each cluster, each cluster performs a crossover operation as a whole. The crossover probability determines whether the individual performs a crossover operation. If a crossover operation is performed, the cluster with low fitness is exchanged as a whole in a roulette wheel manner according to the cell user satisfaction of each individual cluster (Formula 1.7), so that after the individual crossover, one individual has the opportunity to become better overall.

[0100] 5) Mutation operation: This process refers to the individual randomly changing a gene value. When performing the mutation operation, the gene value of the cell with low fitness is changed in a roulette manner according to the user satisfaction of each cell (Formula 1.8). At the same time, the gene value of the cell in the same cluster also needs to be changed, so that the gene value becomes better after the individual mutation.

[0101] 6) Finally, according to the current beam hopping pattern design, the power of each beam in each time slot is fine-tuned. When fine-tuning, the service satisfaction should be ensured first, and power saving should be considered second.

[0102] In the present invention, the core algorithm of beam hopping pattern optimizes the beam hopping pattern through genetic algorithm to maximize user satisfaction and minimize signaling overhead. The innovation lies in the introduction of dynamic operation based on user satisfaction, so that the algorithm can adaptively adjust the search direction. Specifically, user satisfaction is divided into three levels to guide the optimization operation of the algorithm at different levels, thereby improving the responsiveness to user needs.

[0103] 3. Simulation and result analysis;

[0104] The simulation parameters are shown in Table 1. The current satellite coverage area is 37 and the number of satellite beams is 4. The cells need to be clustered according to the traffic volume and the center location of each cell. The clustering results are shown in Figure 3 described.

[0105] Table 1 Simulation parameter settings

[0106] parameter value Satellite altitude 550km Number of cells covered by satellite 37 Number of satellite beams 4 Satellite beam allocatable total power 20dBw Satellite downlink operating frequency 20GHz Satellite beam available bandwidth 1GHz Total number of time slots in the time window 40 Communication minimum elevation angle 30° Center wave radius 150km Weighted weight β 0.55

[0107] Table 2 Clustering results based on geographic location

[0108] Cluster number Cell ID Set Total data demand over a period of time (bps) 1 [3011171823102516245] 96043086.82 2 [19461272381] 200110391.30 3 [21152892714132022] 160709746.26 4 [333534313229372636] 102358278.62

[0109] Table 3 Results of clustering cells using hybrid clustering

[0110] Cluster number Cell ID Set Total data demand over a period of time (bps) 1 [293231537202772218] 139733339.88 2 [2610171312163511] 139585767.28 3 [19634982433234] 139668970.30 4 [14311252825362130] 140233425.53

[0111] Combination Figure 3 From Tables 2 and 3, we can see that although the location of the clustering based on geographical location is relatively compact, the data volume requirements between clusters vary greatly, which will limit the upper limit of the satellite throughput. Performance comparison can be performed after optimization. The optimized clustering method is relatively geographically dispersed, but it also ensures that most of the compactness is met. The difference between clusters after optimization is minimal, which can improve the overall satellite throughput.

[0112] Before solving the genetic algorithm for dynamic operation based on user satisfaction, it is necessary to determine the weighted weight β value in the individual fitness calculation, because the individual fitness is composed of two parts: overall user satisfaction and beam switching overhead. Therefore, the appropriate weighted weight β value has a great impact on the result. Figure 4 The overall user satisfaction, beam switching overhead and the influence of the weighted weight β are shown, and the weighted weight β=0.55 is determined by comprehensive consideration.

[0113] Figure 5 Simulation results of a genetic algorithm based on dynamic operation of user satisfaction are presented, where Figure 5 Figure A is the result of 100 iterations of the algorithm. It can be seen that the final user satisfaction is 0.9856 and the switching overhead is 0.5688. The final satisfaction of the traditional genetic algorithm after 100 iterations is 0.3742 and the switching overhead is 0.8438. The running time for the beam hopping pattern design of 100 generations of 40 time slots in the first cycle is 1 minute. If the clustering does not need to be recalculated in the second cycle and the iteration is based on the beam hopping pattern results of the first 40 time slots, the calculation results can be obtained in less than 30 generations, and the running time is only a few milliseconds, which is highly consistent with the requirements of real-time computing. Figure 5 B in the figure is the beam hopping result, which can be directly sent to the digital beamforming controller to complete beam switching and service. Figure 5 C and D in the figure are the comparison of the calculation results of the algorithm of the present invention and the traditional genetic algorithm after 100 iterations. The dark gray line is the traditional genetic algorithm, and the light gray line is the satisfaction of the algorithm of the present invention. From the perspective of community satisfaction, there is a phenomenon that the satisfaction of individual communities will decrease, but from the perspective of the satisfaction of each cluster, the overall satisfaction has increased.

[0114] Figure 6 The influence of different cell clustering results on user satisfaction in the final beam hopping period is shown. Figure 6 A is the iterative result of the genetic algorithm based on distance clustering, and B is the iterative result of the genetic algorithm based on hybrid clustering. The core algorithm is the same, so the different satisfaction results are caused by the influence of different clustering results, which also verifies that the hybrid clustering algorithm is better.

[0115] When dealing with multi-objective optimization, in addition to weighting, you can also use the idea of ​​Pareto predicate to find the optimal solution set. Figure 7The results of satisfaction and signaling overhead of the improved NSGA-Ⅱ after 100 iterations show that the algorithm cannot accurately give the optimal beam hopping pattern in the current scenario (a feasible solution set is given), and the result of the Pareto preamble is not as good as the result of the weighted method. For example, in the current scenario, the weighted method can achieve a satisfaction of 0.9856 with a signaling overhead of 0.5688, while the NSGA-Ⅱ algorithm can only give the optimal satisfaction of 0.775 and a signaling overhead of 0.73 in the same 100 generations. If the NSGA-Ⅱ algorithm is iterated more times, it may be able to obtain better results than the weighted algorithm, but the purpose of the algorithm of the present invention is to give the satellite onboard processor an optimal decision in a relatively shorter time. Then the NSGA-Ⅱ algorithm is not suitable for the current scenario.

[0116] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A beam hopping pattern design method for a low-orbit satellite, characterized in that: The following steps are involved: Step S1: In the beam hopping scenario, the ground cells are dynamically clustered using a hybrid clustering cell clustering algorithm according to the service requirements of the cells, so that the cells in the same cluster are as close as possible on the basis of load balancing; Step S2: using a genetic algorithm with dynamic operation based on user satisfaction to perform time slot and beam scheduling; The user satisfaction index is divided into three levels to guide three dynamic operations: ① selection operation: using the overall user satisfaction of all cells in the entire beam hopping period to select the best individual to generate offspring; ② crossover operation: using the user satisfaction of each cluster cell to perform individual gene crossover; ③ mutation operation: using the user satisfaction of a single cell to perform individual gene mutation; Step S3: fine-tune the beam power of each beam of the beam hopping pattern when serving in the same time slot.

2. The method for designing a beam hopping pattern of a low-orbit satellite according to claim 1, characterized in that: In step S1, the output of the K-means clustering algorithm is used as the initial solution of the simulated annealing algorithm to explore possible clustering assignments globally.

3. The method for designing a beam hopping pattern of a low-orbit satellite according to claim 1, characterized in that: In step S1, the specific implementation process of the hybrid clustering cell clustering algorithm is as follows: S1.1: Assume that the number of ground cells is X, the number of beams provided by the satellite phased array is K, and the goal is to divide X cells into K clusters; set the objective function as: value k represents the sum of the traffic of the kth cluster, average represents the average traffic of the cluster, and max_dis k Represents the distance between the two farthest cells in the cluster after normalization; S1.2: Use K-means clustering algorithm to find the initial solution; S1.3: Use the output of the K-means clustering algorithm as the initial solution of the simulated annealing algorithm to explore possible clustering assignments globally; S1.4: Determine the similarity of the user distributions of the two beam-hopping periods, that is, substitute the user distribution of the ε+1th period into the clustering result of the εth period. If the standard deviation σ is satisfied ε+1 ≤(1+ξ)σ ε , ξ represents the maximum acceptable floating value of the standard deviation, then the user distribution identity is considered to be satisfied; otherwise, it is considered not to be satisfied.

4. The method for designing a beam hopping pattern for a low-orbit satellite according to claim 3, characterized in that: The maximum acceptable floating value ξ of the standard deviation does not exceed 10%.

5. The method for designing a beam hopping pattern of a low-orbit satellite according to claim 1, characterized in that: In the genetic algorithm, when creating the initial population, each individual is a possible solution, so the generation of individuals is related to the services of users in the cell. At the same time, considering that each cell must be served by at least one time slot, service priority and cell polling are integrated.

6. The method for designing a beam hopping pattern for a low-orbit satellite according to claim 1, characterized in that: The calculation formula for the overall user satisfaction of all cells is: Among them, H t represents the beam hopping pattern matrix in the tth time slot; C t represents the beam capacity that the satellite-borne phased array beam can actually serve in the cell pointed to in the tth time slot; T represents the number of time slots contained in a beam hopping cycle; β represents the weighting weight; K represents the number of beams; φ t represents the number of data packets in each cell at time t; R kt represents the element in the kth row and tth column of the matrix R. Its physical meaning is whether the kth beam in the tth time slot has changed relative to the beam pointing in the t-1th time slot. kt =0 means that the beam direction does not change, that is, the beam will not switch. kt ≠0 means that the beam direction has changed, that is, the beam has been switched.

7. The method for designing a beam hopping pattern of a low-orbit satellite according to claim 6, characterized in that: Assuming that the switching overhead of each beam switching is the same, the clustering matrix is ​​a K×X matrix P, which represents the result of cell clustering; assuming that the beam hopping pattern matrix is ​​an M×T matrix H, the order of serving cells of each beam in the entire beam hopping cycle is obtained through P×H, and P×H×(IB) is the beam switching matrix in a beam hopping cycle, where B is the determination matrix transformed from the unit matrix, I is the unit matrix, and (IB) indicates whether the cell pointed to by the beam in the current time slot is changed compared with the previous time slot; suppose the matrix R=P×H×(IB), then the number of switching times is the sum of the number of elements in the matrix R that are not 0: Where K represents the number of beams; T represents the number of time slots contained in a beam hopping cycle; R kt represents the element in the kth row and tth column of the matrix R. Its physical meaning is whether the kth beam in the tth time slot has changed relative to the beam pointing in the t-1th time slot. kt =0 means that the beam direction does not change, that is, the beam will not switch. kt ≠0 means that the beam direction has changed, that is, the beam has been switched.

8. The method for designing a beam hopping pattern of a low-orbit satellite according to claim 7, characterized in that: The number of elements and cost are normalized to their maximum values, Cost = cost / (K×T). K×T beam switching means that K beams will switch once in each time slot. Combining the two system indicators of user satisfaction and switching overhead, the optimization goal is: Among them, β represents the weighted weight; K represents the number of beams; T represents the number of time slots contained in a beam hopping cycle; φ t represents the number of data packets in each cell at time t; R kt represents the element in the kth row and tth column of the matrix R. Its physical meaning is whether the kth beam in the tth time slot has changed relative to the beam pointing in the t-1th time slot. kt =0 means that the beam direction does not change, that is, the beam will not switch. kt ≠0 means that the beam direction has changed, that is, the beam has been switched.

9. The method for designing a beam hopping pattern of a low-orbit satellite according to claim 1, characterized in that: The calculation formula of the cell user satisfaction of each cluster is: Among them, H t,m represents the beam hopping pattern matrix of the mth cluster in the tth time slot, C t,m represents the actual beam capacity that the satellite-borne phased array beam pair of the mth cluster can serve in the tth time slot. t,m represents the user demand of the mth cluster in the tth time slot, P k represents the cell set of the kth cluster, and T represents the number of time slots contained in a beam hopping cycle.

10. The method for designing a beam hopping pattern of a low-orbit satellite according to claim 1, characterized in that: The calculation formula of the user satisfaction of a single cell is: Among them, H t,m represents the beam hopping pattern matrix of the mth cell in the tth time slot, C t,m represents the beam capacity that can actually be served by the satellite-borne phased array beam pair pointing to the mth cell in the tth time slot, φ t,m It represents the user demand of the mth cell in the tth time slot, and T represents the number of time slots contained in a beam hopping cycle.

Images