Multi-agent automatic base station site selection method and system based on large language model driving

Through the multi-agent automated base station site selection method driven by a large language model, the multi-agent collaborative working method is used to solve the problems of time-consuming and high labor costs of traditional base station site selection, and achieve fast and flexible base station site selection, satisfying the satisfaction of High coverage and cost-effective demand.

CN119012216AInactive Publication Date: 2024-11-22SHANDONG UNIV
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411282881.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-13
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Traditional base station site selection methods are time-consuming and labor-intensive, and cannot reflect the dynamic changes in network performance in real time, resulting in a long site selection period and large data deviations.

Method used

Multi-agent automated base station site selection method driven by large language model is established through the collaborative work of Manager, Formulator, Programmer and Evaluator, mathematical modeling, programming and code verification are established, and genetic algorithms, particle swarm optimization, simulated annealing and linear programming are used. and other algorithms to quickly generate base station site selection schemes that meet constraints.

Benefits of technology

It significantly reduces the labor cost and cycle of the base station site selection process, improves the efficiency and flexibility of solving problems, can quickly adapt to changes in demand, and meets the high coverage and cost-effective base station layout.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119012216A_ABST
    Figure CN119012216A_ABST
Patent Text Reader

Abstract

The invention relates to a multi-agent automatic base station site selection method and system based on large language model driving. The method comprises the following steps: (1) mathematical modeling: converting a base station site selection problem into a mathematical model expressed by a mathematical formula according to related information provided by a user; (2) programming: screening an optimization algorithm according to the mathematical model, and generating a solution according to a Python code template provided by a user; (3) code verification: running a code generated by Programmer, judging whether a grammar error exists in the code, and further judging whether a new base station coordinate generated by the code meets a constraint condition; and (4) under overall coordination, executing the steps (2), (3) and (4) until a feasible solution is found. According to the invention, under the participation of a small amount of people, the labor cost and the site selection period of the base station site selection process are effectively reduced. According to the method provided by the invention, the construction cost of the base station can be obviously reduced on the premise of meeting the minimum traffic coverage rate.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a multi-agent automatic base station site selection method and system based on large language model driving, belonging to the technical field of communication network optimization. Background Art

[0002] As the backbone of mobile communication networks, base stations not only provide seamless connections for mobile users, ensuring the continuity and reliability of communications, but also support the growing demand for high data throughput, allowing users to enjoy high-speed network services even on the move [Cui W, Zhang Y, Wei L, et al. Optimizing base stations deployment in wireless access networks for smart grids [C] / / 2015 International Conference on Wireless Communications & Signal Processing (WCSP). IEEE, 2015: 1-5.]. With the popularity of smartphones and mobile devices, the number of mobile users has increased dramatically, and users' demand for the speed and quality of data transmission has increased accordingly. Therefore, the problem of base station site selection has become particularly critical and challenging, as it directly affects the breadth and depth of network coverage, as well as the quality of user service experience.

[0003] Traditional base station site selection methods mainly rely on road testing and user feedback to evaluate network performance and identify areas that need improvement, and use this as a standard to determine the type and location of new base stations. There are two problems: (1) Road testing collects signal quality data by deploying measurement equipment in a specific geographic area. Although this method can provide valuable insights, it is time-consuming and impractical, especially in densely populated urban environments [Tayal S, Garg PK, Vijay S. Optimization models for selecting base station sites for cellular network planning [C] / / Applications of Geomatics in Civil Engineering: Select Proceedings of ICGCE 2018. Springer Singapore, 2020: 637-647.]. In addition, road testing only captures data at a specific time and place, which cannot fully reflect the dynamic changes in network performance over time; (2) User complaints and feedback are another means of identifying coverage issues and service quality defects in traditional methods. However, this reactive strategy means that network improvements are usually not initiated until problems become severe enough to trigger user complaints [Kulin M, Kazaz T, De Poorter E, et al. A survey on machine learning-based performance improvement of wireless networks: PHY, MAC and network layer [J]. Electronics, 2021, 10 (3): 318.]. In addition, the feedback collected may not represent the entire user base, resulting in biased or incomplete data. Engineers need to analyze this feedback, model the base station site selection problem, develop solutions, deploy new base stations, and re-evaluate network performance, forming a continuous and iterative process.

[0004] In summary, the traditional method is not only time-consuming, but also requires engineers to have extensive expertise in communications, networks, and optimization, which leads to a long cycle and huge manpower expenditure in solving the base station site selection problem. Summary of the invention

[0005] In view of the shortcomings of existing base station site selection methods, the present invention provides a multi-agent automated base station site selection method driven by a large language model, which uses a large model to generate accurate, reliable, real-time and efficient site selection plans.

[0006] Specifically, the present invention proposes a "team" for solving the base station site selection problem, which specifically includes: Manager, Formulator, Programmer and Evaluator. The Manager is responsible for overall coordination, task decomposition and selection of corresponding large models to perform tasks; the Formulator is responsible for mathematical modeling of the base station site selection problem, including the mathematical representation of the objective function and various constraints; the Programmer is responsible for programming and solving the Python code based on the mathematical model established by the Formulator, and modifying the errors and unreasonable parts of the code; the Evaluator is responsible for verifying whether the code generated by the Programmer has errors and whether it meets the constraints. Through this method of cooperation among multiple large models, the large models can make full use of their respective advantages to jointly solve complex base station site selection problems. This method not only improves the efficiency of problem solving, but also significantly reduces the need for manual intervention. With the collaborative efforts of multiple large models, the system can quickly adapt to changing needs and environmental conditions, and provide flexible and efficient base station site selection solutions.

[0007] The present invention also provides a multi-agent automatic base station site selection system driven by a large language model.

[0008] Terminology explanation:

[0009] 1. Manager, task management agent;

[0010] 2. Formulator, a mathematical modeling agent;

[0011] 3. Programmer, code programming agent;

[0012] 4. Evaluator, solution verification agent;

[0013] 5. Genetic Algorithm (GA) is a search heuristic algorithm that simulates the biological evolution process. Through the simulation of natural genetic mechanisms (such as selection, crossover, mutation, etc.), the potential solutions to the problem are evolved in order to find or approach the optimal solution. The application of genetic algorithms in base station site selection can effectively balance cost and performance and find a base station layout solution that is both economical and efficient.

[0014] 6. Particle Swarm Optimization (PSO) is an optimization algorithm based on swarm intelligence. It simulates the social behavior of bird flocks or fish schools to find the optimal solution. In the base station site selection problem, the particle swarm optimization algorithm can effectively search the solution space and find a base station layout solution with high cost-effectiveness.

[0015] 7. Simulated Annealing, SA, is a probabilistic optimization algorithm that is inspired by the annealing of metals in the physical process and simulates the process of slowly cooling a substance after heating. The simulated annealing algorithm abstracts this physical process and is used to solve optimization problems. When the problem scale is large or the solution space is complex, the advantages of the simulated annealing algorithm are more obvious.

[0016] 8. Linear Programming, LP, is a mathematical method used to find the maximum or minimum value of a linear objective function under a set of linear inequality or equality constraints. Through linear programming, the location and number of base stations with the lowest cost under all constraints can be found.

[0017] The technical solution of the present invention is:

[0018] A multi-agent automatic base station site selection method based on large language model driving, comprising:

[0019] (1) Mathematical modeling: Based on the relevant information provided by the user, the base station site selection problem is converted into a mathematical model represented by a mathematical formula;

[0020] (2) Programming: Screen optimization algorithms based on mathematical models and generate solutions based on Python code templates provided by users;

[0021] (3) Code verification: Run the code generated by the Programmer to determine whether there are syntax errors in the code, and further determine whether the coordinates of the newly built base station generated by the code meet the constraints;

[0022] (4) Under overall coordination, execute steps (2), (3) and (4) until a feasible solution is found.

[0023] Preferably, in step (1), based on the relevant information provided by the user, the base station site selection problem is converted into a mathematical model represented by a mathematical formula; specifically, the step includes:

[0024] 1) Collect relevant information provided by users, including the coordinates of established base stations, the coordinates and traffic volume of users with weak coverage, and the coverage range, construction cost, distance constraints between base stations and optimization goals set by users of different types of base stations. Different types of base stations include macro base stations and micro base stations.

[0025] 2) Analyze the data and clarify the objective function and constraints of the base station site selection problem, including identifying the objectives to be optimized and the constraints to be met;

[0026] The goal is to rationally plan the location and type of base stations, minimize costs, and ensure reasonable layout and network optimization while covering at least 90% of the traffic. The constraint is that the distance between new base stations and between new base stations and existing base stations must be greater than a number of grids.

[0027] 3) Use LaTeX to define the mathematical expressions of the objective function and constraints, and define the necessary symbols to represent parameters and variables.

[0028] Further preferably, the mathematical model expressed in mathematical formula is as follows:

[0029]

[0030] in,

[0031]

[0032] In formulas (I) and (II), i represents the i-th existing base station; the subscript h represents a macro base station, and C h represents the cost of the macro base station, d h Indicates the coverage of the macro base station; x n represents the horizontal coordinate of the nth candidate base station; y n represents the vertical coordinate of the nth candidate base station; P i,h,t and P i,d,t They represent the probability that the device at position t∈Gi (coordinates (xt, yt)) is covered by the macro base station or micro base station located at (xi, yi); G i represents the entire area in grid i; w t Represents the position (x t ,y t ) related business volume; represents the coordinate set of all candidate new base station locations; N represents the total number of grid points; represents the coordinate set of all existing base stations; d h and d d Respectively represent the coverage radius of the macro base station and micro base station; C h and C d Denote the cost of macro base station and micro base station respectively; D min represents the minimum distance between any two base stations; p i and q i Boolean variables representing whether a macro base station or a micro base station is deployed at the center point of grid i; (x i ,y i ) are the coordinates of the center point of grid i; are the coordinates of the existing base station j.

[0033] Preferably, in step (2), the optimization algorithm is selected according to the mathematical model, and a solution is generated according to the Python code template provided by the user; including:

[0034] 1) Determine the type of mathematical model according to the characteristics of the generated mathematical model; the characteristics of the mathematical model include the form of the objective function, the complexity of the constraints, the number and type of variables; the type of mathematical model includes linear problems or nonlinear problems;

[0035] 2) Access Python's built-in optimization algorithm library, which includes a variety of algorithms suitable for different types of optimization problems;

[0036] 3) Evaluate the applicability of candidate algorithms, consider the algorithm's convergence speed, computational complexity and solution quality, algorithm versatility, stability and scalability factors, and finally select the best optimization algorithm.

[0037] Further preferably, in step (2), the user provides a Python code template, including: outputting the type and coordinates of the newly built base station, calling a verification function generated based on the large model to verify whether the constraints are met, and printing the results including printing the final coordinates and type of the newly built base station, and drawing a base station site selection map; in addition, the code and path for loading the data set are added to the Python code template.

[0038] Preferably, in step (3), the code generated by the Programmer is run to determine whether there is a syntax error in the code, and further to determine whether the coordinates of the newly-built base station generated by the code meet the constraint conditions; including:

[0039] 4) First check whether there are compilation errors in the code; if there are errors, return the error code and the error information of the programmer's code, modify it according to the feedback information, and check the code again until there are no errors in the code;

[0040] 5) Run the code and check whether the type and coordinates of the newly created base station output by the code meet the constraints; if not, feedback the unsatisfied constraint information and make targeted modifications to the code; until the maximum number of iterations is reached or the generated base station type and coordinates meet all constraints.

[0041] Preferably, in step (4), task decomposition and coordinated management, information exchange and cooperation are performed, specifically including:

[0042] 6) Divide the task into three stages: mathematical modeling, programming, and code verification based on the information provided by the user;

[0043] 7) Information interaction: Analyze the progress and status of solving the current mathematical problem based on the information returned by each large model, and select the appropriate large model to perform the task based on the characteristics of each large model provided by the user.

[0044] A multi-agent automatic base station site selection system driven by a large language model, comprising:

[0045] The mathematical modeling agent is configured to: convert the base station site selection problem into a mathematical model represented by a mathematical formula according to the relevant information provided by the user;

[0046] The code programming agent is configured to: screen optimization algorithms based on mathematical models and generate solutions based on Python code templates provided by users;

[0047] The solution verification agent is configured to: run the code generated by the Programmer, determine whether there are syntax errors in the code, and further determine whether the coordinates of the newly built base station generated by the code meet the constraints;

[0048] The task management agent is configured to execute the steps of mathematical modeling, programming, and code verification until a feasible solution is found.

[0049] The beneficial effects of the present invention are:

[0050] In view of the shortcomings of traditional base station site selection methods, such as long task cycle, high labor cost and inability to reflect the dynamic changes of network performance over time in real time, the present invention proposes a multi-agent automated base station site selection method driven by a large language model. The present invention introduces a large model "team" consisting of Manager, Formulator, Programmer and Evaluator. First, Formulator mathematically models the base station site selection problem and generates a mathematical model represented by LaTeX. Then, Programmer analyzes the type of problem based on the mathematical model, selects a suitable optimization algorithm and generates Python code. Finally, Evaluator checks and runs the code to verify whether the generated base station type and coordinates meet the constraints. The whole process is carried out under the overall coordination of Manager, who is responsible for the information exchange between the major models, analyzes the progress and status of problem solving, and selects a suitable large model to perform the task. The present invention effectively reduces the labor cost and site selection cycle of the base station site selection process with a small amount of human participation. The method proposed in the present invention can significantly reduce the cost of base station construction on the premise of meeting the minimum business volume coverage. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic diagram of the base station site selection problem;

[0052] Figure 2 It is a flowchart of multiple large models collaborating to solve the base station location problem;

[0053] Figure 3 This is the base station site selection diagram of the present invention under the premise of meeting 90% business volume coverage. DETAILED DESCRIPTION

[0054] The present invention will be further described below with reference to the accompanying drawings and embodiments, but is not limited thereto.

[0055] Example 1

[0056] A multi-agent automated base station site selection method based on a large language model Figure 2 As shown, including:

[0057] (1) Mathematical modeling: Based on the relevant information provided by the user, the base station site selection problem is converted into a mathematical model represented by a mathematical formula;

[0058] (2) Programming: Screen optimization algorithms based on mathematical models and generate solutions based on Python code templates provided by users;

[0059] (3) Code verification: Run the code generated by the Programmer to determine whether there are syntax errors in the code, and further determine whether the coordinates of the newly built base station generated by the code meet the constraints;

[0060] (4) Under overall coordination, execute steps (2), (3) and (4) until a feasible solution is found.

[0061] Example 2

[0062] The difference between the multi-agent automatic base station site selection method based on large language model driving described in Example 1 is that:

[0063] In step (1), the base station site selection problem is converted into a mathematical model represented by a mathematical formula according to the relevant information provided by the user; specifically, the following steps are performed:

[0064] 1) Collect relevant information provided by users, including the coordinates of established base stations, the coordinates and traffic volume of users with weak coverage, and the coverage range, construction cost, distance constraints between base stations and optimization goals set by users (e.g., achieving 90% coverage) of different types of base stations. Different types of base stations include macro base stations and micro base stations; Figure 1 shown.

[0065] The present invention uses the dataset of Problem D of the 2022 Math Cup Undergraduate Mathematical Modeling Challenge [“Mathorcup undergraduate mathematical modeling challenge 2022,” http: / / www.mathorcup.org.], which includes a 2500×2500 grid area, and the coordinate range of each grid point is 0 to 2499. The dataset provides the coordinates of weak coverage points and the corresponding traffic volume as well as the coordinate data of the built base stations. The dataset also defines two types of base stations: a macro base station with a coverage range of 30 grids and a cost of 10, and a micro base station with a coverage range of 10 grids and a cost of 1.

[0066] 2) Analyze the data and clarify the objective function and constraints of the base station site selection problem; including identifying the objectives that need to be optimized (such as maximizing coverage and minimizing costs) and the constraints that need to be met (such as minimum coverage and base station interference);

[0067] The goal is to rationally plan the location and type of base stations, minimize costs, ensure reasonable layout and network optimization while covering at least 90% of the traffic; in order to avoid signal interference between base stations, the restriction is that the distance between new base stations and between new base stations and existing base stations must be greater than several (e.g. 10) grids; the present invention uses grids to simulate the actual geographical environment, and divides the geographical area where the base station needs to be deployed into small grid units, each of which represents a certain actual area. In actual application, the size of each grid unit needs to be determined according to the specific geographical environment and planning requirements, and the actual distance between base stations is calculated accordingly.

[0068] 3) Use LaTeX to define the mathematical expressions of the objective function and constraints, and define the necessary symbols to represent parameters and variables.

[0069] During the mathematical modeling process, the generated mathematical expressions will be checked to see if they conform to the LaTeX specifications and whether there are undefined or incorrectly defined parameters and variables. If the above errors exist, the mathematical expressions will be modified according to the error information until the mathematical model meets all the specifications predefined by the user.

[0070] The mathematical model expressed in mathematical formula is as follows:

[0071]

[0072] in,

[0073]

[0074] In formulas (I) and (II), i represents the i-th existing base station; the subscript h represents a macro base station, and Ch represents the cost of the macro base station, d h Indicates the coverage of the macro base station; x n represents the horizontal coordinate of the nth candidate base station; y n represents the vertical coordinate of the nth candidate base station; P i,h,t and P i,d,t They represent the probability that the device at position t∈Gi (coordinates (xt, yt)) is covered by the macro base station or micro base station located at (xi, yi); G i represents the entire area in grid i; w t Represents the position (x t ,y t ) related business volume; represents the coordinate set of all candidate new base station locations; N represents the total number of grid points; represents the coordinate set of all existing base stations; d h and d d Respectively represent the coverage radius of the macro base station and micro base station; C h and C d Denote the cost of macro base station and micro base station respectively; D min represents the minimum distance between any two base stations; p i and q i Boolean variables representing whether a macro base station or a micro base station is deployed at the center point of grid i; (x i ,y i ) are the coordinates of the center point of grid i; are the coordinates of the existing base station j.

[0075] In step (2), the optimization algorithm is selected according to the mathematical model, and a solution is generated according to the Python code template provided by the user; including:

[0076] 1) Determine the type of mathematical model according to the characteristics of the generated mathematical model; the characteristics of the mathematical model include the form of the objective function, the complexity of the constraints, the number and type of variables; the type of mathematical model includes linear problems or nonlinear problems;

[0077] Linear model: There is a linear relationship between the variables in the model, that is, the model can be expressed as a first-order polynomial of the variables. Nonlinear model: The model contains non-first-order polynomial variables, such as x 2 , xy or e x The Programmer agent can determine the type of model based on the generated mathematical model expression.

[0078] 2) Access Python's built-in optimization algorithm library, which includes a variety of algorithms suitable for different types of optimization problems; for example, genetic algorithms, particle swarm optimization, simulated annealing, and linear programming. Here are the optimization algorithms that the Programmer agent may use in the optimization algorithm library. The specific method to be used requires the Programmer agent to select it based on the specific mathematical model;

[0079] 3) Evaluate the applicability of candidate algorithms, consider the algorithm's convergence speed, computational complexity and solution quality, algorithm versatility, stability and scalability factors, and finally select the best optimization algorithm.

[0080] Different optimization algorithms have different characteristics and are suitable for different mathematical problems. The Programmer agent is a large ChatGPT model. The input information sets it to have the ability to analyze mathematical problems. Therefore, it will analyze the characteristics of mathematical problems to select appropriate optimization algorithms and perform programming.

[0081] In order to generate code that meets the expected results, in step (2), the user provides a Python code template, including: outputting the type and coordinates of the newly built base station, calling the verification function generated based on the large model to verify whether the constraints are met, and printing the results including printing the final coordinates and type of the newly built base station, and drawing the base station site selection map; in addition, the code and path for loading the dataset are added to the Python code template, so that the dataset can be loaded correctly.

[0082] In step (3), the code generated by the Programmer is run to determine whether there are syntax errors in the code, and further determine whether the coordinates of the newly created base station generated by the code meet the constraint conditions; including:

[0083] 4) First check whether there are compilation errors in the code; if there are errors, return the error code and the error information of the programmer's code, such as undefined variables and unimported library functions; modify according to the feedback information and check the code again; the above process will be repeated until the code is error-free;

[0084] 5) Run the code and check whether the type and coordinates of the newly built base station output by the code meet the constraints; if not, the unsatisfied constraint information will be fed back and the code will be modified in a targeted manner; the above process will be iterated until the maximum number of iterations is reached or the generated base station type and coordinates meet all constraints.

[0085] Preferably, in step (4), task decomposition and coordinated management, information exchange and cooperation are performed, specifically including:

[0086] 6) Divide the task into three stages: mathematical modeling, programming, and code verification based on the information provided by the user;

[0087] 7) Information interaction: Analyze the progress and status of solving the current mathematical problem based on the information returned by each large model, and select the appropriate large model to perform the task based on the characteristics of each large model provided by the user.

[0088] Large models usually refer to models with a large number of parameters in the field of artificial intelligence, especially in machine learning and deep learning, such as ChatGPT. Because of the large number of parameters, these models can capture and learn complex patterns and relationships in the data, and thus perform well in various tasks, such as image recognition, natural language processing, and mathematical problem solving. The Manager, Formulator, Programmer, and Evaluator given in this article are all ChatGPT large models; these large models will combine the user's input information and pre-trained data to give text answers.

[0089] Run the program of the entire solution, print out the type and coordinates of the newly built base station in the console, and draw an image.

[0090] In order to evaluate the performance of the proposed solution, the present invention uses two main indicators: traffic coverage and cost. The traffic coverage is calculated as the percentage of weak coverage points covered by the base station to the total traffic, and the cost is measured by the total cost of the base station used to achieve the expected traffic coverage. Figure 3 This is the base station site selection diagram of the present invention under the premise of meeting 90% business volume coverage. For the simplicity of the problem, the present invention uses a 100x100 grid area in the data set with a coordinate range of (300-399, 0-99) and sets the minimum business volume coverage ratio to 90% to minimize the construction cost.

[0091] Example 3

[0092] A multi-agent automatic base station site selection system driven by a large language model, comprising:

[0093] The mathematical modeling agent is configured to: convert the base station site selection problem into a mathematical model represented by a mathematical formula according to the relevant information provided by the user;

[0094] The code programming agent is configured to: screen optimization algorithms based on mathematical models and generate solutions based on Python code templates provided by users;

[0095] The solution verification agent is configured to: run the code generated by the Programmer, determine whether there are syntax errors in the code, and further determine whether the coordinates of the newly built base station generated by the code meet the constraints;

[0096] The task management agent is configured to execute the steps of mathematical modeling, programming, and code verification until a feasible solution is found.

Claims

1. A multi-agent automated base station site selection method based on a large language model, characterized in that: include: (1) Mathematical modeling: Based on the relevant information provided by the user, the base station site selection problem is converted into a mathematical model represented by a mathematical formula; (2) Programming: Screen optimization algorithms based on mathematical models and generate solutions based on Python code templates provided by users; (3) Code verification: Run the code generated by the Programmer to determine whether there are syntax errors in the code, and further determine whether the coordinates of the newly built base station generated by the code meet the constraints; (4) Under overall coordination, execute steps (2), (3) and (4) until a feasible solution is found.

2. According to claim 1, a multi-agent automatic base station site selection method based on large language model driving is characterized in that: In step (1), the base station site selection problem is converted into a mathematical model represented by a mathematical formula according to the relevant information provided by the user; specifically, the following steps are performed: 1) Collect relevant information provided by users, including the coordinates of established base stations, the coordinates and traffic volume of users with weak coverage, and the coverage range, construction cost, distance constraints between base stations and optimization goals set by users of different types of base stations. Different types of base stations include macro base stations and micro base stations. 2) Analyze the data and clarify the objective function and constraints of the base station site selection problem, including identifying the objectives to be optimized and the constraints to be met; The goal is to rationally plan the location and type of base stations, minimize costs, and ensure reasonable layout and network optimization while covering at least 90% of the traffic. The constraint is that the distance between new base stations and between new base stations and existing base stations must be greater than a number of grids. 3) Use LaTeX to define the mathematical expressions of the objective function and constraints, and define the necessary symbols to represent parameters and variables.

3. The multi-agent automatic base station site selection method based on large language model driving according to claim 1 is characterized in that: The mathematical model expressed in mathematical formula is as follows: in, In formulas (I) and (II), i represents the i-th existing base station; the subscript h represents a macro base station, and C h represents the cost of the macro base station, d h Indicates the coverage of the macro base station; x n represents the horizontal coordinate of the nth candidate base station; y n represents the vertical coordinate of the nth candidate base station; P i,h,t and P i,d,t They represent the probability that the device at position t∈Gi (coordinates (xt, yt)) is covered by the macro base station or micro base station located at (xi, yi); G i represents the entire area in grid i; w t Represents the position (x t ,y t ) related traffic; N represents the coordinate set of all candidate new base station locations; N represents the total number of grid points; T represents the coordinate set of all existing base stations; d h and d d Respectively represent the coverage radius of the macro base station and micro base station; C h and C d Denote the cost of macro base station and micro base station respectively; D min represents the minimum distance between any two base stations; p i and q i Boolean variables representing whether a macro base station or a micro base station is deployed at the center point of grid i; (x i ,y i ) are the coordinates of the center point of grid i; are the coordinates of the existing base station j.

4. The multi-agent automatic base station site selection method based on large language model driving according to claim 1 is characterized in that: In step (2), the optimization algorithm is selected according to the mathematical model, and a solution is generated according to the Python code template provided by the user; include: 1) Determine the type of mathematical model according to the characteristics of the generated mathematical model; the characteristics of the mathematical model include the form of the objective function, the complexity of the constraints, the number and type of variables; the type of mathematical model includes linear problems or nonlinear problems; 2) Access Python's built-in optimization algorithm library, which includes a variety of algorithms suitable for different types of optimization problems; 3) Evaluate the applicability of candidate algorithms, consider the algorithm's convergence speed, computational complexity and solution quality, algorithm versatility, stability and scalability factors, and finally select the best optimization algorithm.

5. The multi-agent automatic base station site selection method based on large language model driving according to claim 1 is characterized in that: In step (2), the user provides a Python code template, including: outputting the type and coordinates of the newly built base station, calling the verification function generated based on the large model to verify whether the constraints are met, and printing the results including printing the final coordinates and type of the newly built base station, and drawing a base station site selection map; in addition, the code and path for loading the dataset are added to the Python code template.

6. The multi-agent automatic base station site selection method based on large language model driving according to claim 1 is characterized in that: In step (3), the code generated by the Programmer is run to determine whether there are syntax errors in the code, and further determine whether the coordinates of the newly created base station generated by the code meet the constraint conditions; including: 4) First check whether there are compilation errors in the code; if there are errors, return the error code and the error information of the programmer's code, modify it according to the feedback information, and check the code again until there are no errors in the code; 5) Run the code and check whether the type and coordinates of the newly created base station output by the code meet the constraints; if not, feedback the unsatisfied constraint information and make targeted modifications to the code; until the maximum number of iterations is reached or the generated base station type and coordinates meet all constraints.

7. The multi-agent automatic base station site selection method based on large language model driving according to claim 1 is characterized in that: In step (4), task decomposition and coordination management, information exchange and cooperation are carried out, including: 6) Divide the task into three stages: mathematical modeling, programming, and code verification based on the information provided by the user; 7) Information interaction: Analyze the progress and status of solving the current mathematical problem based on the information returned by each large model, and select the appropriate large model to perform the task based on the characteristics of each large model provided by the user.

8. A multi-agent automated base station site selection system driven by a large language model, characterized in that: include: The mathematical modeling agent is configured to: convert the base station site selection problem into a mathematical model represented by a mathematical formula according to the relevant information provided by the user; The code programming agent is configured to: screen optimization algorithms based on mathematical models and generate solutions based on Python code templates provided by users; The solution verification agent is configured to: run the code generated by the Programmer, determine whether there are syntax errors in the code, and further determine whether the coordinates of the newly built base station generated by the code meet the constraints; The task management agent is configured to execute the steps of mathematical modeling, programming, and code verification until a feasible solution is found.

Citation Information

Patent Citations

  • Combined base station location optimization method under heterogeneous network convergence scene

    CN103619027A

  • Optimization problem solving framework based on large language model

    CN118468931A

  • Base station site planning method and system based on multi-model optimization

    CN118590899A

  • Base station deployment method and device

    WO2019037683A1