A multi-objective co-evolution resource configuration optimization method, system, device and medium based on Latin hypercube

Abstract

A multi-objective collaborative evolution resource configuration optimization method, system, device and medium based on Latin hypercube, the method is: multi-objective optimization (CMSC) algorithm initialization based on beam space division and beam coverage detection; two initial populations Pop1 and Pop2 of the same size are generated by using the Latin hypercube design (LHD) method; population cycle iteration, satellite communication transmission resource configuration strategy is optimized by collaborative evolution algorithm; the system, device and medium are used to realize the method; the population is initialized by the Latin hypercube design (LHD), the collaborative evolution framework of the original problem and the auxiliary problem and the offspring strategy selection, the multi-dimensional resources of satellite energy domain and beam domain can be efficiently distributed, the generated satellite resource configuration strategy can reduce communication energy consumption and transmission delay, and the satellite communication transmission resource configuration strategy has the advantages of rapid generation and adaptability to different ground user distribution.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology, and specifically relates to a multi-objective cooperative evolutionary resource allocation optimization method, system, device and medium based on Latin hypercube. Background Technology

[0002] With the rapid development of the Internet of Things (IoT) and 5G / 6G network technologies, achieving ubiquitous connectivity and seamless communication coverage has become one of the goals of global communication systems. Against this backdrop, Low Earth Orbit (LEO) communication satellites, as an important supplement to existing terrestrial communication networks, are receiving increasing attention. LEO satellites, with their near-Earth orbit characteristics, can provide efficient communication services to remote areas and regions with weak traditional network coverage, such as oceans. Compared to Geostationary Orbit (GEO) satellites, LEO satellites offer advantages such as low latency, high bandwidth, and flexible coverage, better meeting the needs of modern communication networks. With advancements in satellite and antenna technologies, the communication capabilities of LEO satellites have significantly improved, and the flexibility of onboard payloads has been enhanced, enabling LEO satellites to meet more diverse communication requirements.

[0003] LEO satellites possess multi-dimensional resources including time, frequency, space, energy, and wave propagation. The flexible scheduling and efficient utilization of these resources are crucial for improving their communication performance. Existing research focuses on the flexible allocation of onboard multi-dimensional resources, proposing various innovative solutions. For example, in time-frequency resource allocation, some studies have proposed dynamic resource scheduling schemes based on time-division multiplexing and frequency-division multiplexing, which can dynamically adjust resource allocation according to user needs and the communication environment. Regarding space resources, beamforming technology is widely used, intelligently adjusting the satellite antenna array to focus beam energy, achieving directional communication and improving spectrum efficiency. Furthermore, beam energy management is also a current research hotspot, optimizing satellite energy efficiency and extending satellite lifespan through adaptive power allocation. Some studies have proposed resource allocation schemes based on beam cascading, combining time, space, and frequency multi-dimensional information to optimize onboard resource scheduling.

[0004] Faced with the complex multidimensional resource allocation problem in the aforementioned scenarios, solutions can be broadly categorized into two types. The first is traditional algorithms, primarily based on classical mathematical theories such as convex optimization, game theory, and graph theory. Convex optimization methods can effectively solve partially linearized resource allocation problems; game theory models resource competition among multiple satellites and solves for the optimal solution through policy iteration. Graph theory techniques are widely used in inter-satellite link management and network topology optimization to improve communication efficiency by finding the optimal communication link. However, as the complexity of resource allocation problems increases, these problems often fall under the category of mixed-integer nonlinear programming problems (MINLP), making it difficult for traditional algorithms to find the optimal solution within a finite timeframe.

[0005] Another area of ​​research focusing on solutions based on intelligent algorithms is gradually becoming a key area of ​​research in multidimensional resource allocation problems. Intelligent algorithms, such as simulated annealing, particle swarm optimization, and reinforcement learning, can explore the solution space of large-scale complex problems through heuristic search processes. Due to the nonlinearity and complexity of the MINLP problem, traditional methods often cannot solve it independently, while intelligent algorithms provide a more flexible and efficient solution approach, adaptable to uncertain and complex environments. In summary, as the complexity of multidimensional resource allocation problems increases, intelligent algorithms, with their efficient global search capabilities and advantages in handling nonlinear problems, are gradually becoming the main research direction in this field. Intelligent methods such as simulated annealing, particle swarm optimization, and reinforcement learning can provide flexible and efficient resource allocation schemes for LEO satellite networks. In the future, the application of intelligent algorithms in satellite communications will provide stronger technical support for achieving seamless global communication coverage.

[0006] Patent application CN 118709461 A discloses a "Multi-objective optimization method for cantilever beam structures based on Hall effect tension sensors". It selects samples by optimal Latin hypercube sampling and optimizes the structural parameters of cantilever beams by combining the GPR-NSGA-II algorithm, which improves the prediction efficiency of the Gaussian process regression model. However, since the algorithm depends on the combination of Gaussian process regression (GPR) and NSGA-II, its efficiency is limited for high-dimensional or nonlinear problems and it lacks versatility.

[0007] The patent application document with publication number CN 117669095A discloses "Optimization Method of Centrifugal Impeller Variable Parameters of Combined Pump Based on Agent Model". It uses Latin hypercube sampling and NSGA-II algorithm to optimize the design parameters of centrifugal impeller of fuel pump, which significantly improves head and efficiency. However, because the algorithm relies on the superposition of models such as full factor screening and high-order surface response, the algorithm has high computational cost, is difficult to handle large-scale variable problems, and has poor robustness in various scenarios. Summary of the Invention

[0008] To overcome the problems of poor global convergence and high computational complexity in existing technologies for multidimensional resource allocation, the present invention aims to provide a multi-objective co-evolutionary resource allocation optimization method, system, device, and medium based on Latin hypercube design. Through the Latin hypercube design (LHD) initialization of the population, the co-evolutionary framework of the primal problem and auxiliary problems, and the offspring strategy selection method, the present invention can efficiently allocate multidimensional resources in the satellite energy domain and beam domain. The generated satellite resource allocation strategy can reduce communication energy consumption and transmission latency, and has the technical effects of rapid generation of satellite communication transmission resource allocation strategy and adaptability to different ground user distributions.

[0009] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0010] A multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube includes the following steps:

[0011] Step 1: Initialization of the multi-objective optimization (CMSC) algorithm based on beamspace partitioning and beam coverage detection:

[0012] Input initial parameter list: Original problem f origin Auxiliary problem f help Population size N, maximum number of iterations I max ;

[0013] Among them, the original problem f origin This refers to the satellite communication optimization problem, an auxiliary problem f. help The energy consumption and latency constraints refer to the satellite communication optimization problem; the population size N refers to the number of decision variables participating in the optimization process, which determine the configuration of satellite resources such as transmit power and antenna beams; the maximum number of iterations I. max This refers to the maximum number of computation steps preset during the optimization process, used to ensure that resource allocation optimization is completed within a reasonable timeframe;

[0014] Step 2: Initialize the population:

[0015] Two initial populations, Pop1 and Pop2, of the same size were generated using the Latin hypercube design (LHD) method. The two initial populations, Pop1 and Pop2, of the same size represent different satellite communication transmission resource allocation strategies.

[0016] Step 3: Based on the initial parameters from Step 1 and the two initial populations Pop1 and Pop2 of the same size generated in Step 2, perform population cyclic iteration and optimize the satellite communication transmission resource allocation strategy through a co-evolutionary algorithm.

[0017] The specific steps of step 2 are as follows:

[0018] Step 2.1: Given the possible ranges of transmit power and antenna beamwidth for each element in the decision variables in the actual physical system, and the initial population size N, divide the intervals;

[0019] Step 2.2: For each decision variable, after dividing it into intervals to represent different resource allocation ranges, randomly select a sample point from each interval divided in Step 2.1. The sample point represents different combinations of satellite resource allocation parameters.

[0020] Step 2.3: Randomly pair the sample points generated for all variables to form a complete sample set consisting of multiple different sample points, and there is no fixed combination of sample points for different variables;

[0021] Step 2.4: For the complete sample set formed in Step 2.3, calculate the Kolmogorov-Smirnov value of the sample distribution in each dimension through statistical uniformity test to ensure uniform sample coverage; if the distribution in some areas is sparse, re-divide the intervals and supplement the sampling, adopt an adaptive interval subdivision strategy, and prioritize adding sample points in sparse areas until the distribution in all areas is uniform, thus completing the generation of the sample set.

[0022] Step 2.5: Use the sample set generated in Step 2.4 as the initial population. Each sample point represents an individual in the population. The characteristics of an individual are determined by its corresponding variable value. Each individual corresponds to a satellite multidimensional resource allocation scheme. The individuals form two initial populations of the same size, Pop1 and Pop2.

[0023] The specific steps of step 3 are as follows:

[0024] Step 3.1: Count and accumulate: i = i + 1;

[0025] Step 3.2: Crossover operation: Select half individuals from each of the initial populations Pop1 and Pop2 generated in Step 2 to obtain parent population 1 and parent population 2, that is, obtain two sets containing different combinations of satellite resource configurations;

[0026] Step 3.3: Mutation operation: Perform mutation operation on each individual in parent population 1 and parent population 2 obtained in step 3.2 to obtain new individuals. The new individuals form offspring population 1 and offspring population 2.

[0027] Step 3.4: Population Update: Take the intersection of the initial population Pop1 generated in Step 2, the descendant population 1 generated in Step 3.3, and the descendant population 2 as the new population Pop. new ;

[0028] Step 3.5: Population Assessment: Using the original question f from Step 1 origin and auxiliary problem f help The initial populations Pop1 and Pop2 generated in step 2, and the new population Pop obtained in step 3.4. new An evaluation was conducted, using coverage detection based on average latency and average energy consumption. Initial population Pop1 and Pop2, as well as the new population Pop, were calculated respectively. new The fitness score of each individual;

[0029] Step 3.6: Environment Selection: Based on the fitness scores calculated in Step 3.5, the performance of the satellite communication transmission resource allocation strategy in satellite communication is evaluated. The initial populations Pop1 and Pop2 generated in Step 2 and the new population Pop obtained in Step 3.4 are selected. newSelect the dominant individuals with higher fitness scores and eliminate the inferior individuals with lower fitness scores, then proceed to step 3.7;

[0030] Step 3.7: Termination condition judgment: When the number of iterations i ≤ I max When the iteration count i > I, proceed to step 3.1; max At that time, the population cycle iteration is completed;

[0031] Step 3.8: Retain step 3.7 in i = I max The dominant individual obtained at that time is output as the optimal solution, which yields a set of satellite communication transmission resource allocation strategies.

[0032] This invention also provides a multi-objective co-evolutionary resource allocation optimization system based on Latin hypercube, comprising:

[0033] The algorithm initialization module is used to initialize the multi-objective optimization (CMSC) algorithm based on beamspace partitioning and beam coverage detection.

[0034] Input initial parameter list: Original problem f origin Auxiliary problem f help Population size N, maximum number of iterations I max ;

[0035] Among them, the original problem f origin This refers to the satellite communication optimization problem, an auxiliary problem f. help The energy consumption and latency constraints refer to the satellite communication optimization problem; the population size N refers to the number of decision variables participating in the optimization process, which determine the configuration of satellite resources such as transmit power and antenna beams; the maximum number of iterations I. max This refers to the maximum number of computation steps preset during the optimization process, used to ensure that resource allocation optimization is completed within a reasonable timeframe;

[0036] The population initialization module is used to generate two initial populations of the same size, Pop1 and Pop2, using the Latin hypercube design (LHD) method. The two initial populations of the same size, Pop1 and Pop2, represent different satellite communication transmission resource allocation strategies.

[0037] The satellite communication transmission resource allocation strategy optimization module is used to optimize the satellite communication transmission resource allocation strategy by performing population cyclic iteration based on initial parameters and two initial populations Pop1 and Pop2 of the same size, and by using a co-evolutionary algorithm.

[0038] This invention also provides a multi-objective co-evolutionary resource allocation optimization device based on Latin hypercube, comprising:

[0039] Memory: A computer program that stores the above-mentioned multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube, and is a computer-readable device;

[0040] Processor: Used to implement the aforementioned multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube when executing the computer program.

[0041] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the aforementioned multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0043] 1. This invention generates a uniformly distributed initial population using Latin Hypercube Design (LHD) in the initialization phase of step 2. Simultaneously, it eliminates invalid solutions related to beam coverage constraints through an environment selection strategy in step 3.6, constructing a Co-evolutionary Strategy (CMSC) for "Beam Space Partitioning and Coverage Detection." This improves the algorithm's convergence speed and enhances the globality of the solution set. It solves the problem of slow convergence speed caused by low-quality initial solution sets in traditional intelligent algorithms (such as Particle Swarm Optimization), and also improves the scene dependence problem, enhancing the algorithm's cross-domain adaptability.

[0044] 2. This invention utilizes the original problem f input in step 1. origin Auxiliary problem f help The original problem of satellite communication performance and the auxiliary problem of energy consumption and latency constraints are constructed to form a framework of "co-evolution of the original problem and the auxiliary problem". Multi-objective joint optimization is carried out to reduce satellite energy consumption while meeting the low information freshness (AoI) requirements of 5G / 6G. It also solves the mixed integer nonlinear programming problem that is difficult to handle by traditional convex optimization methods.

[0045] 3. Through the population assessment in step 3.5 and the environment selection in step 3.6, this invention generates a dynamic beam coverage detection and adaptive resource allocation strategy. For different ground user distribution scenarios such as densely populated urban areas and remote sparse areas, it automatically adjusts the beam spatial division granularity and resource allocation weight, taking into account the robustness of different user distribution scenarios.

[0046] In summary, this invention solves the problems of slow convergence and high complexity in LEO satellite resource allocation by using a dynamic interference suppression method for LHD initialization and descendant strategy selection under a multi-objective co-evolutionary architecture. It also has high robustness and can adapt to different user distributions, providing a more universal resource optimization scheme for satellite communication systems. Attached Figure Description

[0047] Figure 1This is the flowchart of the algorithm of this invention.

[0048] Figure 2 This is a schematic diagram illustrating the coupling relationship between different parameters of an individual gene in this invention.

[0049] Figure 3 This is a schematic diagram of the individual crossover mutation operation of the present invention.

[0050] Figure 4 This is a specific logical structure diagram of the individual gene encoding method of the present invention.

[0051] Figure 5 This is a performance simulation comparison chart of the CMSC algorithm of this invention with five existing multi-objective algorithms in a rural sparse ground user scenario.

[0052] Figure 6 This is a performance simulation comparison chart of the CMSC algorithm of this invention with five existing multi-objective algorithms in a user scenario with sparse urban ground.

[0053] Figure 7 This is a performance simulation comparison chart of the CMSC algorithm of this invention with five existing multi-objective algorithms in a dense urban ground user scenario.

[0054] Figure 8 This is a performance simulation comparison chart of the CMSC algorithm of this invention with five existing multi-objective algorithms in a mixed region dense user scenario. Detailed Implementation

[0055] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0056] This invention proposes a multi-objective optimization algorithm based on beam space partitioning and beam coverage detection, abbreviated as CMSC, to solve the problem of satellite downlink communication resource allocation.

[0057] A multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube includes the following steps, and the specific algorithm flow is as follows: Figure 1 As shown:

[0058] Step 1: Initialization of the multi-objective optimization (CMSC) algorithm based on beamspace partitioning and beam coverage detection:

[0059] Input initial parameter list: Original problem f origin Auxiliary problem f help Population size N, maximum number of iterations I max ;

[0060] Among them, the original problem f origin This refers to the satellite communication optimization problem, an auxiliary problem f. helpThe energy consumption and latency constraints refer to the satellite communication optimization problem; the population size N refers to the number of decision variables participating in the optimization process, which determine the configuration of satellite resources such as transmit power and antenna beams; the maximum number of iterations I. max This refers to the maximum number of computation steps preset during the optimization process, used to ensure that resource allocation optimization is completed within a reasonable timeframe;

[0061] Step 2: Initialize the population:

[0062] Two initial populations, Pop1 and Pop2, of the same size were generated using the Latin Hypercube Design (LHD) method. These two initial populations, Pop1 and Pop2, represent different satellite communication transmission resource allocation strategies. The specific steps are as follows:

[0063] Step 2.1: Divide the range: Based on the range of the optimization decision variables (antenna beamwidth and transmit power), divide these ranges into n uniform intervals; for example, if the antenna beamwidth range is [B min B max The transmission power range is [P]. min ,P max If ], it can be divided into multiple intervals;

[0064] Step 2.2: Randomly select samples: For each decision variable (antenna beamwidth and transmit power), randomly select a sample point from each interval divided in Step 2.1. The sample point represents different combinations of satellite resource configuration parameters; ensure that the selected sample can represent the characteristics of each interval.

[0065] Step 2.3: Sample Combination: Randomly pair the sample points generated from all variables to form a complete sample set consisting of multiple different sample points; in this process, ensure that there are no fixed combinations of sample points for different variables, thereby improving the diversity of the sample;

[0066] Step 2.4: Verification and Adjustment: For the complete sample set formed in Step 2.3, perform a statistical uniformity test to calculate the Kolmogorov-Smirnov values ​​of the sample distribution in each dimension to ensure uniform sample coverage. If some areas are sparsely distributed, re-divide the intervals and supplement sampling, using an adaptive interval subdivision strategy to prioritize adding sample points in sparse areas until all areas are evenly distributed, thus completing the sample set generation. For example, if the samples in a certain area are too concentrated, the number of samples in that area can be increased to ensure better coverage.

[0067] Step 2.5: Generate the initial population: Use the sample set generated in Step 2.4 as the initial population. Each sample point represents an individual in the population. The characteristics of the individual are determined by its corresponding variable values ​​(beamwidth and power). Each individual corresponds to a satellite multi-dimensional resource allocation scheme. The individuals form two initial populations of the same size, Pop1 and Pop2.

[0068] like Figure 2 As shown, the individual gene coding method generates individuals to form the initial population. Each individual includes K genes, represented as SBP(π). K Each gene is composed of a beamwidth parameter group Θ. k and power parameter group It consists of two parts, including the beamwidth parameter group Θ k Includes a beamwidth parameter Θ k =θ k Power parameter group It consists of a set of power values. Beamwidth parameter group Θ in each gene of an individual k With power parameter group There is a coupling relationship between them, beamwidth parameter group Θ k It will affect the power parameter group The number of power values ​​included, and the specific individual gene parameter correlation and coupling relationships, are as follows: Figure 3 As shown.

[0069] Step 3: Based on the initial parameters from Step 1 and the two initial populations Pop1 and Pop2 of the same size generated in Step 2, perform population cyclic iteration and optimize the satellite communication transmission resource allocation strategy through a co-evolutionary algorithm.

[0070] Perform the following operation in a loop until the maximum number of iterations is reached:

[0071] Step 3.1: Count and accumulate: i = i + 1;

[0072] Step 3.2: Crossover operation: Select half individuals from each of the initial populations Pop1 and Pop2 generated in Step 2 to obtain parent population 1 and parent population 2, that is, obtain two sets containing different combinations of satellite resource configurations, thereby increasing the exploration range of the search space;

[0073] Step 3.3: Mutation operation: Perform a mutation operation on each individual in parent population 1 and parent population 2 obtained in step 3.2 to obtain new individuals. The new individuals form offspring population 1 and offspring population 2. The mutation operation can be achieved by fine-tuning the values ​​of beamwidth and power to increase the diversity of individuals.

[0074] like Figure 4As shown, in the crossover and mutation operations, the beamwidth parameter group gene and the power parameter group gene in each individual undergo crossover and mutation operations independently.

[0075] Step 3.4: Population Update: Take the intersection of the initial population Pop1 generated in Step 2, the descendant population 1 generated in Step 3.3, and the descendant population 2 as the new population Pop. new ;

[0076] Step 3.5: Population Assessment: Using the original question f from Step 1 origin and auxiliary problem f help The initial populations Pop1 and Pop2 generated in step 2, and the new population Pop obtained in step 3.4. new An evaluation was conducted, using coverage detection based on average latency and average energy consumption. Initial population Pop1 and Pop2, as well as the new population Pop, were calculated respectively. new The fitness score of each individual;

[0077] Step 3.6: Environment Selection: Based on the fitness scores calculated in Step 3.5, the performance of the satellite communication transmission resource allocation strategy in satellite communication is evaluated. The initial populations Pop1 and Pop2 generated in Step 2 and the new population Pop obtained in Step 3.4 are selected. new Select dominant individuals with high fitness scores and eliminate inferior individuals with low fitness scores to ensure continuous improvement in the quality of the population; proceed to step 3.7.

[0078] Step 3.7: Termination condition judgment: When the number of iterations i ≤ I max When the iteration count i > I, proceed to step 3.1; max At that time, the population cycle iteration is completed;

[0079] Step 3.8: Retain step 3.7 in i = I max The dominant individual obtained at that time is output as the optimal solution, which yields a set of satellite communication transmission resource allocation strategies.

[0080] This satellite communication transmission resource allocation strategy has superior beamwidth and power parameters, thereby achieving lower average latency and average power consumption in satellite downlink scenarios.

[0081] Simulation conditions

[0082] 1. The satellite orbital altitude is 780km, the orbital inclination is 53°, the inter-satellite spacing is 5km, the maximum diameter of the target ground area is 500km, and the time slot length is 1s.

[0083] 2. The communication system parameters include: the satellite's maximum power is 500W, the beamwidth is 200MHz, the Ku-band carrier frequency is 12GHz, the noise power spectral density is -174dBm / Hz, the user receiving antenna gain is 30dBi, and the minimum elevation angle is 25°.

[0084] Simulation content

[0085] This study compares the performance of six multi-objective evolutionary algorithms (MOEA) in a specific region, evaluating their performance using Pareto fronts (PFs). The horizontal axis represents the average information age (AoI), and the vertical axis represents the average energy consumption. The simulation aims to evaluate the optimization effect of different algorithms on the satellite multidimensional resource allocation problem, focusing on analyzing the algorithm's ability to balance information timeliness and energy consumption. By comparing the Pareto fronts (PFs) of the six MOEA algorithms, the study verifies the optimization effect of the proposed algorithm on average AoI and average energy consumption in resource allocation, demonstrating its superiority.

[0086] A brief introduction to five existing comparison algorithms:

[0087] ToP (Tournament-based Selection): ToP is a tournament-based selection algorithm that selects the best-performing individual as the parent for crossover and mutation operations by comparing within a subset.

[0088] PPS (Pareto-based Population Selection): PPS is an algorithm based on Pareto front selection. Its core idea is to guide the evolution of the population by selecting individuals located at the Pareto front.

[0089] ARMOEA (Adaptive Reference-point-based Multi-Objective Evolutionary Algorithm): ARMOEA is an adaptive multi-objective evolutionary algorithm based on reference points. It guides the selection of individuals by dynamically adjusting the reference points, thereby promoting the diversity and convergence of solutions.

[0090] MOEA / D (Multi-Objective Evolutionary Algorithm Based on Decomposition): MOEA / D is a multi-objective optimization algorithm based on the decomposition idea. It transforms a multi-objective problem into multiple single-objective optimization subproblems and seeks the global optimal solution by optimizing the combination of these subproblems.

[0091] NSGA-II (Non-dominated Sorting Genetic Algorithm II): NSGA-II is a classic multi-objective evolutionary algorithm based on non-dominated sorting. It selects the next generation of individuals by non-dominated sorting and crowding comparison, and has strong solution set diversity and convergence.

[0092] Simulation results

[0093] Figure 5 This is a performance simulation comparison chart of the CMSC algorithm of this invention with five existing multi-objective algorithms in a rural sparse ground user scenario.

[0094] Figure 6 This is a performance simulation comparison chart of the CMSC algorithm of this invention with five existing multi-objective algorithms in a user scenario with sparse urban ground.

[0095] Figure 7 This is a performance simulation comparison chart of the CMSC algorithm of this invention with five existing multi-objective algorithms in a dense urban ground user scenario.

[0096] Figure 8 This is a performance simulation comparison chart of the CMSC algorithm of this invention with five existing multi-objective algorithms in a mixed region dense user scenario.

[0097] Figures 5 to 8 The results show a comparison between the CMSC algorithm of this invention and five existing multi-objective algorithms in four different scenarios. As can be seen from the figures, in all four scenarios, the CMSC algorithm of this invention demonstrates a significant performance improvement compared to ToP, PPS, ARMOEA, MOEA / D, and NSGA-II algorithms.

[0098] To facilitate a direct comparison of the performance of the six algorithms in balancing convergence and diversity, the figure presents the solution distribution in the objective space under four different scenarios. The approximate Pareto fronts obtained by NSGA-II, MOEA / D, ARMOEA, PPS, ToP, and CMSC are labeled with sky blue, green, purple, yellow, orange, and navy blue, respectively. In each subplot, the x-axis represents the average information freshness (AoI) (in milliseconds), and the y-axis represents the average transmission energy consumption (in joules). Depending on the optimization problem, a smaller objective function value indicates better algorithm performance. Through intuitive observation, the CMSC algorithm proposed in this invention outperforms the other five existing multi-objective algorithms in terms of solution distribution and convergence in multi-objective problems, demonstrating the best optimization results. Specifically, the CMSC algorithm of this invention consistently obtains the non-dominated solution set with the minimum Pareto front in all four scenarios, indicating that it achieves the best trade-off between average AoI and average energy.

[0099] This invention also provides a multi-objective co-evolutionary resource allocation optimization system based on Latin hypercube, comprising:

[0100] The algorithm initialization module is used to initialize the multi-objective optimization (CMSC) algorithm based on beam space partitioning and beam coverage detection in step 1:

[0101] Input initial parameter list: Original problem f origin Auxiliary problem f help Population size N, maximum number of iterations I max ;

[0102] Among them, the original problem f origin This refers to the satellite communication optimization problem, an auxiliary problem f. help The energy consumption and latency constraints refer to the satellite communication optimization problem; the population size N refers to the number of decision variables participating in the optimization process, which determine the configuration of satellite resources such as transmit power and antenna beams; the maximum number of iterations I. max This refers to the maximum number of computation steps preset during the optimization process, used to ensure that resource allocation optimization is completed within a reasonable timeframe;

[0103] The population initialization module is used to generate two initial populations Pop1 and Pop2 of the same size in step 2 using the Latin hypercube design (LHD) method. The two initial populations Pop1 and Pop2 of the same size represent different satellite communication transmission resource allocation strategies.

[0104] The satellite communication transmission resource allocation strategy optimization module is used to implement the population cyclic iteration based on the initial parameters in step 1 and the two initial populations Pop1 and Pop2 of the same size generated in step 2 in step 3, and optimize the satellite communication transmission resource allocation strategy through a co-evolutionary algorithm.

[0105] This invention also provides a multi-objective co-evolutionary resource allocation optimization device based on Latin hypercube, comprising:

[0106] Memory: A computer program that stores the above-mentioned multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube, and is a computer-readable device;

[0107] Processor: Used to implement the aforementioned multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube when executing the computer program.

[0108] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the aforementioned multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube.

Claims

1. A multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube, characterized in that, Includes the following steps: Step 1: Initialization of the multi-objective optimization (CMSC) algorithm based on beamspace partitioning and beam coverage detection: Input initial parameter list: Original problem f origin Auxiliary problem f help Population size N, maximum number of iterations I max ; Among them, the original problem f origin This refers to the satellite communication optimization problem; auxiliary problem f help The energy consumption and latency constraints refer to the satellite communication optimization problem; the population size N refers to the number of decision variables participating in the optimization process, which determine the configuration of satellite resources such as transmit power and antenna beams; the maximum number of iterations I. max This refers to the maximum number of computation steps preset during the optimization process, used to ensure that resource allocation optimization is completed within a reasonable timeframe; Step 2: Initialize the population: Two initial populations, Pop1 and Pop2, of the same size were generated using the Latin hypercube design (LHD) method. The two initial populations, Pop1 and Pop2, of the same size represent different satellite communication transmission resource allocation strategies. Step 3: Based on the initial parameters from Step 1 and the two initial populations Pop1 and Pop2 of the same size generated in Step 2, perform population cyclic iteration and optimize the satellite communication transmission resource allocation strategy through a co-evolutionary algorithm.

2. The multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube according to claim 1, characterized in that, The specific steps of step 2 are as follows: Step 2.1: Given the possible ranges of transmit power and antenna beamwidth for each element in the decision variables in the actual physical system, and the initial population size N, divide the intervals; Step 2.2: For each decision variable, after dividing it into intervals to represent different resource allocation ranges, randomly select a sample point from each interval divided in Step 2.

1. The sample point represents different combinations of satellite resource allocation parameters. Step 2.3: Randomly pair the sample points generated for all variables to form a complete sample set consisting of multiple different sample points, and there is no fixed combination of sample points for different variables; Step 2.4: For the complete sample set formed in Step 2.3, calculate the Kolmogorov-Smirnov value of the sample distribution in each dimension through statistical uniformity test to ensure uniform sample coverage; if the distribution in some areas is sparse, re-divide the intervals and supplement the sampling, adopt an adaptive interval subdivision strategy, and prioritize adding sample points in sparse areas until the distribution in all areas is uniform, thus completing the generation of the sample set. Step 2.5: Use the sample set generated in Step 2.4 as the initial population. Each sample point represents an individual in the population. The characteristics of an individual are determined by its corresponding variable value. Each individual corresponds to a satellite multidimensional resource allocation scheme. The individuals form two initial populations of the same size, Pop1 and Pop2.

3. The multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube according to claim 1, characterized in that, The specific steps of step 3 are as follows: Step 3.1: Count and accumulate: i = i + 1; Step 3.2: Crossover operation: Select half individuals from each of the initial populations Pop1 and Pop2 generated in Step 2 to obtain parent population 1 and parent population 2, that is, obtain two sets containing different combinations of satellite resource configurations; Step 3.3: Mutation operation: Perform mutation operation on each individual in parent population 1 and parent population 2 obtained in step 3.2 to obtain new individuals. The new individuals form offspring population 1 and offspring population 2. Step 3.4: Population Update: Take the intersection of the initial population Pop1 generated in Step 2, the descendant population 1 generated in Step 3.3, and the descendant population 2 as the new population Pop. new ; Step 3.5: Population Assessment: Using the original question f from Step 1 origin and auxiliary problem f help The initial populations Pop1 and Pop2 generated in step 2, and the new population Pop obtained in step 3.

4. new An evaluation was conducted, using coverage detection based on average latency and average energy consumption. Initial population Pop1 and Pop2, as well as the new population Pop, were calculated respectively. new The fitness score of each individual; Step 3.6: Environment Selection: Based on the fitness scores calculated in Step 3.5, the performance of the satellite communication transmission resource allocation strategy in satellite communication is evaluated. The initial populations Pop1 and Pop2 generated in Step 2 and the new population Pop obtained in Step 3.4 are selected. new Select the dominant individuals with higher fitness scores and eliminate the inferior individuals with lower fitness scores, then proceed to step 3.7; Step 3.7: Termination condition judgment: When the number of iterations i ≤ I max At that time, proceed to step 3.1; When the iteration number i > I max At that time, the population cycle iteration is completed; Step 3.8: Retain step 3.7 in i = I max The dominant individual obtained at that time is output as the optimal solution, which yields a set of satellite communication transmission resource allocation strategies.

4. A multi-objective co-evolutionary resource allocation optimization system based on Latin hypercube according to any one of claims 1 to 3, characterized in that, include: The algorithm initialization module is used to initialize the multi-objective optimization (CMSC) algorithm based on beamspace partitioning and beam coverage detection. Input initial parameter list: Original problem f origin Auxiliary problem f help Population size N, maximum number of iterations I max ; Among them, the original problem f origin This refers to the satellite communication optimization problem, an auxiliary problem f. help The energy consumption and latency constraints refer to the satellite communication optimization problem; the population size N refers to the number of decision variables participating in the optimization process, which determine the configuration of satellite resources such as transmit power and antenna beams; the maximum number of iterations I. max This refers to the maximum number of computation steps preset during the optimization process, used to ensure that resource allocation optimization is completed within a reasonable timeframe; The population initialization module is used to generate two initial populations of the same size, Pop1 and Pop2, using the Latin hypercube design (LHD) method. The two initial populations of the same size, Pop1 and Pop2, represent different satellite communication transmission resource allocation strategies. The satellite communication transmission resource allocation strategy optimization module is used to optimize the satellite communication transmission resource allocation strategy by performing population cyclic iteration based on initial parameters and two initial populations Pop1 and Pop2 of the same size, and by using a co-evolutionary algorithm.

5. A multi-objective co-evolutionary resource allocation optimization device based on Latin hypercube, characterized in that, include: Memory: A computer program for a multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube as described in any one of claims 1-3, and is a computer-readable device; Processor: Used to implement the multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube as described in any one of claims 1-3 when executing the computer program.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, enables the implementation of the multi-objective co-evolutionary resource allocation optimization method based on Latin hypercube as described in any one of claims 1-3.

Images