A 5G base station deployment optimization method based on an improved quantum evolutionary algorithm

Abstract

This invention discloses a 5G base station deployment optimization method based on an improved quantum evolution algorithm, comprising: acquiring geographical data and defining initial conditions; constructing a multi-level candidate site pool; establishing an objective function that balances coverage performance and construction economy; performing population encoding based on qubit probability amplitude to achieve an equal-probability superposition state; obtaining discrete base station deployment schemes by observing the superposition state during quantum evolution, and ensuring that the schemes meet preset construction indicators through heuristic repair logic; guiding the population to evolve towards the optimal region by updating the probability amplitude of qubits based on an adaptive quantum rotation gate update mechanism; outputting the global optimal solution and performing spatial mapping to obtain the optimized deployment scheme. This invention, through the combination of differentiated modeling of heterogeneous networks and quantum search logic, can effectively reduce signal overlap redundancy between base stations and achieve optimal network coverage performance configuration with limited construction resources.

Description

Technical Field

[0001] This invention belongs to the field of communication network planning and relates to 5G base station deployment optimization technology, specifically to a 5G base station deployment optimization method based on an improved quantum evolution algorithm. Background Technology

[0002] With the full commercialization of fifth-generation mobile communication technology (5G), heterogeneous networks (HNs) have become the norm in modern communication networks. 5G networks are not merely a continuation of 4G networks, but a completely new architecture with higher speeds, lower latency, wider connectivity, and stronger reliability. In this process, the deployment of 5G base stations has become one of the key tasks in network planning.

[0003] In 5G network construction, operators face multi-layered base station deployment challenges. This includes not only building new sites in undeveloped areas but also upgrading existing 2G, 3G, and 4G base stations on-site to save construction costs and maximize the use of existing base stations and supporting facilities. 5G base station deployment relies not only on physical network expansion but also on the coordinated operation of existing multi-generational base stations (2G, 3G, and 4G). This multi-generational coexistence environment presents complex optimization challenges, including technical differences in coverage, spectrum allocation, and power control between different base stations, as well as how to meet coverage requirements while reducing construction costs and improving resource utilization efficiency.

[0004] Currently, base station site selection and deployment optimization methods mainly rely on traditional heuristic algorithms, such as greedy algorithms, genetic algorithms (GA), and simulated annealing (SA). Greedy algorithms typically select site locations based on the level of local demand, prioritizing areas with higher demand. However, this method easily leads to uneven base station distribution, potentially resulting in overly dense site distribution in some areas while other areas have signal coverage gaps, ultimately leading to decreased network quality and wasted resources. Furthermore, greedy algorithms only focus on local optima and struggle to escape the trap of finding local optima. While traditional heuristic algorithms like genetic algorithms and simulated annealing can explore the global solution space effectively, they exhibit "premature convergence" when dealing with large-scale discrete combinatorial optimization problems. This means the algorithm converges prematurely to certain suboptimal solutions, preventing further optimization and resulting in insufficient solution quality. Moreover, genetic algorithms and simulated annealing may not effectively balance the differences between different base station generations (such as long-wavelength 4G base stations versus short-wavelength 5G base stations) when handling complex problems with multiple objectives and constraints, especially when considering differences in physical coverage characteristics, easily overlooking the compatibility issues between 5G and existing base stations.

[0005] Quantum Evolutionary Algorithm (QEA), as an emerging intelligent optimization technique, possesses numerous advantages. Utilizing the superposition of qubits and the interference effect between quantum states, it provides a wider search space and higher parallelism, enabling the processing of large-scale optimization problems in a shorter time. Through quantum tunneling, QEA effectively avoids local optima, exhibiting stronger global search capabilities. Therefore, in solving complex base station deployment problems, QEA has the potential to surpass traditional algorithms in terms of global search efficiency and accuracy. However, despite its strong potential in multiple fields, particularly in complex combinatorial optimization problems, QEA still faces several challenges in its practical application to base station deployment optimization. First, due to the complex physical constraints and multi-objective optimization problems involved in base station deployment, the convergence accuracy and computational efficiency of QEA remain somewhat insufficient. For example, when considering the differentiated physical coverage characteristics of multi-generational base stations, QEA may suffer from insufficient optimization accuracy in balancing coverage effectiveness and construction costs. Therefore, improving QEA to enhance its convergence accuracy and computational efficiency in large-scale base station deployment optimization is a pressing technical challenge in the current communications field. Summary of the Invention

[0006] Purpose of the invention: To overcome the shortcomings of existing technologies, this invention provides a 5G base station deployment optimization method based on an improved quantum evolutionary algorithm. The aim is to maximize network coverage efficiency while minimizing construction costs by leveraging the advantages of quantum evolutionary algorithms combined with the local search capabilities of traditional algorithms. Especially in complex environments with multiple network generations coexisting, this method balances the physical characteristics and constraints of base stations from different generations. Through the combination of differentiated modeling of heterogeneous networks and quantum search logic, it can effectively reduce signal overlap and redundancy between base stations, achieving optimal network coverage efficiency configuration with limited construction resources.

[0007] Technical Solution: To achieve the above objectives, this invention provides a 5G base station deployment optimization method based on an improved quantum evolution algorithm, comprising the following steps:

[0008] S1: Acquire geographic data and define initial conditions;

[0009] S2: Based on geographical data and initial conditions, complete the spatial construction of a multi-level candidate site pool;

[0010] S3: Establish an objective function that balances coverage performance and construction economy;

[0011] S4: Based on the multi-level candidate site pool, perform population coding based on qubit probability amplitude to achieve an equal probability superposition state;

[0012] S5: During the quantum evolution process, discrete base station deployment schemes are obtained by observing superposition states, and heuristic repair logic is used to ensure that the schemes meet the preset construction indicators.

[0013] S6: Based on the adaptive quantum rotation gate update mechanism, the population is guided to evolve towards the optimal region by updating the probability amplitude of the qubit;

[0014] S7: Repeat steps S5 to S6 until the maximum number of iterations is reached, output the global optimal solution and perform space mapping to obtain the deployment optimization scheme.

[0015] Further, step S1 includes: obtaining a set of user distribution points within the target area. ,in Represent the spatial coordinates of the j-th user; simultaneously obtain the existing base station set. ,in , Indicate the communication generation type of the base station; establish a coverage radius mapping table. Different physical coverage radii are matched for base stations of different generations, including And those to be deployed .

[0016] Further, step S2 includes:

[0017] Upgrade candidate pool construction: targeting existing base stations Calculate its evaluation score ,in For Euclidean distance, To evaluate the radius; if but Otherwise f=0; press Sites are sorted in descending order, and the top P-percentage sites are selected to enter the upgrade candidate pool. ;

[0018] New candidate pool construction: for user distribution point set Using the K-Means clustering algorithm, users are divided into For each spatial cluster, calculate the centroid coordinates of that cluster. Use it as a new candidate pool ;

[0019] Combinatorial encoding dimension: Determine the total optimization dimension .

[0020] Furthermore, the objective function in step S3 is expressed as follows:

[0021]

[0022] in: Let V be the set of all base stations currently in the active state in the current scheme; V represents the logical OR operation; λ is the penalty coefficient; M is the total number of user points; Cost is the normalized construction cost, defined as:

[0023]

[0024] in, The site is in an open state. To upgrade the open status of candidate sites, This indicates that the creation of new candidate sites is enabled. and These are the upper limits for the construction scale of candidate sites for upgrading and candidate sites for creating new sites, respectively. and These are the first cost weight and the second cost weight, respectively.

[0025] Further, step S4 includes:

[0026] Initialize a quantum population containing n individuals Each quantum individual It consists of D qubits:

[0027]

[0028] Satisfying constraints In the initial state, let This ensures that all candidate sites are in a superposition state of equal probability between being open and closed.

[0029] Furthermore, the method for obtaining discrete base station deployment schemes by observing superposition states in step S5 includes: observing quantum populations. Take measurements for the first The first individual One qubit, generating a in Uniformly distributed random numbers in an interval The observation logic is: if Then the decision variable corresponding to the location A value of 0 indicates the base station is not activated; conversely, a value of 0 indicates the decision variable is active. A value of 1 indicates that the base station is enabled.

[0030] Furthermore, the method in step S5 for ensuring that the solution meets the preset construction indicators through heuristic repair logic includes:

[0031] Establish upgrade scale constraints: ,in, To upgrade the total number of stations, This represents the maximum number of upgrade stations.

[0032] Establish new size constraints: ,in, This represents the total number of newly built stations. This represents the maximum number of newly built stations.

[0033] If the observed solution violates the above constraints, the repair function is called to execute the forced flip strategy, specifically: first, excess amount calculation is performed, calculating the difference between the current number of stations opened and the upper limit. Next, random downsampling is performed, randomly selecting from all site indices marked "1" in the current solution. One site; perform a forced state flip, forcibly changing the state of the extracted site from 1 to 0.

[0034] Further, step S6 includes:

[0035] Using quantum rotating doors Update probability amplitude:

[0036]

[0037] Among them, the rotation step size Adaptive decay strategy is adopted:

[0038]

[0039]

[0040] in, It is a function of rotation direction; This is the initial base step size; For the number of iterations The step size function for linear decay; T max This is the preset maximum number of iterations.

[0041] Furthermore, the rotation direction function in step S6 The rotation direction (forward or reverse) is determined based on the comparison between the qubit state and the optimal solution bit, ensuring that the probability amplitude evolves towards the optimal solution. The decision logic is: if the current global optimal solution... In the If the value of the bit is 1, and the current solution is 0, then the probability magnitude... Rotate the axis in the positive direction to increase the probability that the point will be opened in the next iteration; conversely, if the current global optimal solution is not found, rotate the axis in the positive direction. If the value is 0, and the current solution is 1, then the probability magnitude... The axis is rotated in the positive direction to increase the probability that the site is closed.

[0042] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0043] 1. Global optimization capability: The quantum evolution algorithm avoids local optima by utilizing the superposition property of qubits, thus improving global search capability.

[0044] 2. High efficiency and accuracy: The dynamic quantum rotation update mechanism accelerates the convergence speed and improves the accuracy of the optimization results.

[0045] 3. Adapt to the differences between different generations of base stations: By modeling differentiated physical envelopes, it adapts to the physical characteristics of base stations of different generations and optimizes the balance between coverage and cost. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of the overall process of the method of the present invention;

[0047] Figure 2 A comparison of the convergence curves of the quantum evolution algorithm and the comparative algorithm;

[0048] Figure 3 This is a schematic diagram of the optimal base station spatial deployment scheme for the output. Detailed Implementation

[0049] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0050] Example 1:

[0051] like Figure 1 As shown, this embodiment provides a 5G base station deployment optimization method based on an improved quantum evolution algorithm, including the following steps:

[0052] S1: Acquire geographic data and define initial conditions;

[0053] This embodiment takes Jiangning District of Nanjing City, Jiangsu Province as the study area and obtains multi-source geospatial data within this area.

[0054] First, a set of user distribution points within the target area is constructed based on the PopSE high-resolution population density dataset. ,in Represent the spatial coordinates of the j-th user; simultaneously obtain the existing base station set. ,in , Indicates the communication generation type of the base station;

[0055] To accurately simulate heterogeneous network environments, this embodiment establishes a coverage radius mapping table. Different physical coverage radii are matched for base stations of different generations, including And those to be deployed .

[0056] Existing low-frequency base stations (2G / 3G): have strong signal penetration and diffraction capabilities, and their physical coverage radius is set at 1200m;

[0057] Existing mid-band base stations (4G): Taking into account both coverage breadth and capacity, their physical coverage radius is set at 1000m;

[0058] High-frequency base stations (5G) to be deployed: Considering the spatial attenuation characteristics of 3.5GHz band signals and the need for high-density base station deployment, its physical coverage radius is set to 600m.

[0059] The basic information of the data used in this embodiment is shown in Table 1:

[0060] Table 1 Basic Data Information

[0061]

[0062] S2: Based on geographical data and initial conditions, complete the spatial construction of a multi-level candidate site pool;

[0063] Upgrade candidate pool construction: targeting existing base stations Calculate its evaluation score ,in For Euclidean distance, To evaluate the radius; if but Otherwise f=0; press Sites are sorted in descending order, and the top P-percentage sites are selected to enter the upgrade candidate pool. ;

[0064] New candidate pool construction: for user distribution point set Using the K-Means clustering algorithm, users are divided into For each spatial cluster, calculate the centroid coordinates of that cluster. Use it as a new candidate pool ;

[0065] Combinatorial encoding dimension: Determine the total optimization dimension .

[0066] Based on the above solution, in this embodiment:

[0067] For existing base stations, this embodiment calculates the "radar response potential" of each existing base station and assigns a quantitative score by assessing the user distribution within a 1500-meter radius around it. Specifically, if a certain existing base station... The number of user sampling points within a 1500-meter radius is Then the base station score Calculated using the following formula:

[0068]

[0069] in, This represents the coordinates of the existing base station, while The coordinates of the user sampling points are used. This scoring method can effectively assess the user distribution within the coverage area of ​​existing base stations, providing support for further site selection. To avoid overly concentrated site distribution, this embodiment introduces a ranking of demand scores and selects the top 30% of sites into the upgrade candidate pool based on the 70th percentile. These base stations will serve as targets for 5G network upgrades.

[0070] In the process of generating new candidate sites, user groups are clustered and the centroid of each cluster is calculated to determine these centroids as potential locations for new base stations. In this embodiment, the number of clusters K is set to 50 to effectively divide the area and avoid excessive redundant base stations. Simultaneously, a fixed random seed (random_state=42) is used to ensure the reproducibility of the results. This method can accurately generate 50 new candidate base station locations in areas with insufficient coverage, and these sites are used as a pool of new candidate sites. This is used for subsequent optimization and deployment.

[0071] In the initial stage of base station deployment, this embodiment sets a comprehensive site distribution map D based on the aforementioned ratio limit between upgraded and newly built sites. The total number of sites is 110, including 60 existing base stations and 50 new candidate sites. Each site's location is determined by two-dimensional coordinates, and site activation and deactivation are determined through binary decision-making. To control the scale and optimization effect of base station construction, this embodiment sets... It is 0.3, and it is specified that... The value is 0.005, which means that current sites account for 30% of the total number of sites, while newly built sites account for 0.5%. These constraints ensure that the calculation results of the optimization algorithm meet the coverage requirements while effectively controlling resource allocation and construction budget.

[0072] S3: Establish an objective function that balances coverage performance and construction economy;

[0073] First, calculate the coverage of the current site selection scheme within the study area. To reflect the true physical characteristics of heterogeneous networks, the algorithm dynamically matches the physical radius based on the base station type during evaluation: for selected upgrade sites and newly built sites, the 5G standard physical radius is uniformly applied. A determination is made; for existing sites that are not selected, their original intergenerational radius is retained (e.g., 1000m for 4G, 1200m for 2G / 3G). The mathematical expression for coverage determination is:

[0074]

[0075] in, The total number of points for all users. This refers to the set of all currently active base stations. This is a logical OR operation, meaning that a user point is considered covered as long as it falls within the radius of any active base station.

[0076] To constrain resource input, this embodiment introduces a normalized construction cost. It is composed of a weighted average of upgrade costs and new construction costs:

[0077]

[0078] in, Indicates the site's on / off status (1 for on, 0 for off). To upgrade the open status of candidate sites, This indicates that the creation of new candidate sites is enabled. and These represent the upper limits for the construction scale of candidate sites for upgrading and candidate sites for creating new sites, respectively. By dividing the actual number of sites by the maximum allowed number of sites, the cost is limited to the range of [0, 1] to facilitate multi-objective fusion with coverage.

[0079] The final objective function is defined as:

[0080]

[0081] In this embodiment, a penalty coefficient is set. The design logic of this parameter is based on the premise that when the benefit of a 1% increase in coverage is greater than... When the cost increase is several times, the algorithm tends to increase the number of base stations; conversely, if increasing the number of base stations contributes little to improving coverage, a penalty term is applied. The algorithm will take the lead, guiding it to eliminate redundant sites while ensuring coverage, thereby optimizing the economic solution.

[0082] S4: Based on the multi-level candidate site pool, perform population coding based on qubit probability amplitude to achieve an equal probability superposition state;

[0083] Initialize a quantum population containing n individuals Each quantum individual It consists of D qubits:

[0084]

[0085] Satisfying constraints In the initial state, let This ensures that all candidate sites are in a superposition state of equal probability between being open and closed.

[0086] In this embodiment:

[0087] To leverage quantum parallelism to improve the search efficiency of site selection schemes, this embodiment maps base station deployment decisions to quantum space, using the probability amplitude of qubits to characterize the "on" or "off" state of a site. This embodiment initializes a... A quantum population of individuals Each quantum individual Depend on It consists of 10 qubits, each corresponding to a specific position in the candidate pool. The state of a qubit. Defined as:

[0088]

[0089] in, and For complex probability amplitudes, in the real number operations of this embodiment, they are expressed as satisfying The coordinate values.

[0090] In the initial stage of the algorithm startup, in order to ensure an unbiased global search of all candidate sites in Jiangning District, the probability amplitude of all qubits is initialized as follows:

[0091]

[0092] At this point, each candidate site is in a superposition of equal probabilities of being on (1) and off (0), that is, the probability of observing the base station being on is... The value is 0.5. This encoding method allows a single quantum entity to store data simultaneously. This provides several possible deployment options, greatly expanding the search coverage.

[0093] The 110 dimensions of the encoding vector correspond one-to-one with physical spatial locations: dimensions 1 to 60 are mapped to the upgrade candidate pool. Its probability amplitude evolution determines whether existing 4G / 3G sites are upgraded to 5G; dimensions 61 to 110 are mapped to a newly created candidate pool generated by K-Means clustering. The probability amplitude evolution determines whether to add a new 5G base station at the centroid of the user cluster area. During the evolution process, the algorithm strictly maintains the normalization constraint for each qubit, through... Applying trigonometric functions on a plane and The state is updated to ensure that the probability amplitude moves on the unit circle, thus mathematically guaranteeing the legality of the base station deployment decision and the stability of the algorithm.

[0094] S5: During the quantum evolution process, discrete base station deployment schemes are obtained by observing superposition states, and heuristic repair logic is used to ensure that the schemes meet the preset construction indicators.

[0095] Methods for obtaining discrete base station deployment schemes by observing superposition states include: quantum populations. Take measurements for the first The first individual One qubit, generating a in Uniformly distributed random numbers in an interval The observation logic is: if Then the decision variable corresponding to the location A value of 0 indicates the base station is not activated; conversely, a value of 0 indicates the decision variable is active. When the value is 1, the base station is activated. This process simulates wave function collapse in quantum mechanics, allowing the algorithm to maintain diversity while moving towards a high-fitness region through the evolution of probability amplitude.

[0096] Methods for ensuring that a solution meets preset construction targets through heuristic repair logic include:

[0097] Establish upgrade scale constraints: ,in, To upgrade the total number of stations, This represents the maximum number of upgrade stations.

[0098] Establish new size constraints: ,in, This represents the total number of newly built stations. This represents the maximum number of newly built stations.

[0099] If the observed solution violates the above constraints, the repair function is called to execute the forced flip strategy, specifically: first, excess amount calculation is performed, calculating the difference between the current number of stations opened and the upper limit. Next, random downsampling is performed, randomly selecting from all site indices marked "1" in the current solution. For each site, a forced state flip is performed, forcibly changing the state of the extracted site from 1 to 0, thereby ensuring the randomness of site selection and bringing the solution back into the feasible solution space.

[0100] The repaired binary vector and This will be used as the final candidate deployment scheme and substituted into the objective function for fitness evaluation. This "observe first, then fix" logic avoids the problem of frequent illegal solutions generated by traditional algorithms when dealing with complex constraints, and significantly improves the optimization efficiency.

[0101] S6: Based on the adaptive quantum rotation gate update mechanism, the population is guided to evolve towards the optimal region by updating the probability amplitude of the qubit;

[0102] Using quantum rotating doors Update probability amplitude:

[0103]

[0104] Among them, the rotation step size Adaptive decay strategy is adopted:

[0105]

[0106]

[0107] in, It is a function of rotation direction; As the initial base step size, this embodiment sets it to . ; For the number of iterations The step size function for linear decay; T max This is the preset maximum number of iterations. In the early stages of the search, when... When the step size is small, the step size is close to The algorithm has a large probability amplitude, which is beneficial for escaping local optima; however, in the later stages of the search, as the number of iterations approaches the maximum... The step size gradually approaches 0, achieving high-precision, small-amplitude search of the neighborhood of the optimal solution.

[0108] Rotation direction function The rotation direction (forward or reverse) is determined based on the comparison between the qubit state and the optimal solution bit, ensuring that the probability amplitude evolves towards the optimal solution. The decision logic is: if the current global optimal solution... In the If the value of the bit is 1, and the current solution is 0, then the probability magnitude... Rotate the axis in the positive direction to increase the probability that the point will be opened in the next iteration; conversely, if the current global optimal solution is not found, rotate the axis in the positive direction. If the value is 0, and the current solution is 1, then the probability magnitude... The axis is rotated in the positive direction to increase the probability that the site is closed.

[0109] Through matrix transformation, the state of the qubit slides smoothly on the unit circle, thereby achieving precise control over the direction of population evolution without destroying the quantum superposition property.

[0110] The algorithm iteratively executes the "observation-evaluation-update" process described above. After each generation of evolution, the algorithm automatically updates the globally optimal fitness value. and the corresponding deployment vector When the preset algebraic limit of 100 is reached, the algorithm stops evolving and outputs the final list of coordinates for the upgrade and construction of 5G base stations in Jiangning District.

[0111] S7: Repeat steps S5 to S6 until the maximum number of iterations is reached, output the global optimal solution and perform space mapping to obtain the deployment optimization scheme.

[0112] Example 2:

[0113] To verify the effectiveness and efficacy of the method of the present invention, the following experiments and data analysis were conducted in this embodiment:

[0114] This embodiment introduces two algorithms for comparative analysis under the same geographical environment: the Genetic Algorithm (GA) uses classic binary encoding and crossover mutation mechanisms, with a population size of 30; the Simulated Annealing Algorithm (SA) uses a probabilistic acceptance criterion based on temperature decay, with an initial temperature of 10.0 and a cooling coefficient of 0.85. The comparison focuses on the following dimensions:

[0115] By comparing convergence characteristics, we analyze the search efficiency of QEA in the early and late stages of iteration, as well as its mechanism advantages in avoiding local optima traps.

[0116] Coverage effectiveness comparison: Under the same construction cost and scale constraints, the differences in the contribution of schemes generated by different algorithms to the overall regional coverage rate are analyzed.

[0117] This embodiment uses Jiangning District, Nanjing City, Jiangsu Province as a case study to conduct actual tests and perform multi-dimensional performance analysis. Figure 2 The convergence curves shown illustrate the evolution trajectory of fitness scores for different optimization algorithms over 100 iterations. Experimental results demonstrate that the QEA algorithm proposed in this invention achieves a final fitness score of 0.824, significantly outperforming the genetic algorithm (GA, final score 0.776) and simulated annealing algorithm (SA, final score 0.712) in the control group. This indicates that QEA possesses extremely strong global optimization capabilities when dealing with large-scale, high-dimensional heterogeneous network location problems.

[0118] From the perspective of convergence dynamics, QEA achieves a breakthrough improvement in convergence slope in the early stage of iteration by leveraging the parallel search advantage of quantum superposition states after a short period of exploration. In the later stage of iteration, it accurately locks the global optimal region through an adaptive rotating door mechanism, effectively overcoming the "premature convergence" defect of traditional GA and SA algorithms, which are prone to getting trapped in local optima.

[0119] In this embodiment, after each generation of evolution, the algorithm automatically updates the global optimal fitness value and the corresponding deployment vector. When the preset upper limit of 100 generations is reached, the algorithm stops evolving and outputs the final list of coordinates for the upgrade and construction of 5G base stations in Jiangning District.

[0120] This embodiment extracts the globally optimal vector. And reconstruct the physical coordinates based on the index location. For upgraded site positioning, identify... The index with a value of 1 in the first 60 bits is selected from the upgrade candidate pool. Extracting the corresponding existing base station coordinates, in this embodiment, the determined existing site is used for 5G hardware upgrade. For the location of a new site, identification... The index with a value of 1 in the last 50 bits is used to select candidates from the newly created pool. The corresponding K-Means cluster centroid coordinates are extracted, and in this embodiment, 5G base station construction is carried out at the identified high-potential locations. During the full data verification stage and the scheme output stage, this embodiment switches to the full user point set in Jiangning District for final evaluation to verify the comprehensive coverage level under the constraints of the 5% upgrade limit and the 0.5% new construction limit.

[0121] Using geographic information visualization technology, the optimized base station distribution was projected onto the administrative vector map of Jiangning District. Different geometric symbols were used to distinguish base station types on the map to ensure high legibility even in black and white printing: existing base stations were represented by circles, indicating unchanged existing communication sites; 5G upgrade sites were marked with triangles, indicating 5G sites upgraded from existing sites; and newly built 5G sites were marked with stars, indicating new 5G sites added at the centroid of population distribution. The visualization results intuitively demonstrate how 5G sites are filling gaps in densely populated areas and original network coverage in Jiangning District.

[0122] Regarding specific spatial site selection efficiency and engineering indicators, this embodiment uses the QEA algorithm to design a refined supplementary scheme for 3892 original base stations in the study area. Through intelligent evolution of the algorithm, and strictly adhering to the constraints of a 0.5% new construction cap and a 5% upgrade cap, a 5G base station deployment scheme consisting of 156 upgraded sites and 5 new sites was finally determined. The specific optimized deployment results are as follows: Figure 3As shown in Table 2, QEA achieves efficient reuse of existing infrastructure, specifically including hardware upgrades to the physical space resources of 71 4G sites, 8 3G sites, and 68 2G sites. This strategy of deep reuse of existing sites significantly reduces the initial investment cost of 5G network construction. On the other hand, QEA precisely identifies five new 5G base stations at the centroid of user distribution density, which, combined with large-scale generational upgrade sites, effectively fills the signal coverage gaps in the original network. Experimental data demonstrates that this invention can obtain the optimal deployment list with lower construction investment through collaborative optimization of multiple generational sites, providing a solution with significant technological advancement and engineering application value for refined 5G networking at the city level.

[0123] Table 2 Results of Optimized Deployment of 5G Base Stations

[0124] .

Claims

1. A method for optimizing 5G base station deployment based on an improved quantum evolutionary algorithm, characterized in that, Includes the following steps: S1: Acquire geographic data and define initial conditions; S2: Based on geographical data and initial conditions, complete the spatial construction of a multi-level candidate site pool; S3: Establish an objective function that balances coverage performance and construction economy; S4: Based on the multi-level candidate site pool, perform population coding based on qubit probability amplitude to achieve an equal probability superposition state; S5: During the quantum evolution process, discrete base station deployment schemes are obtained by observing superposition states, and heuristic repair logic is used to ensure that the schemes meet the preset construction indicators. S6: Based on the adaptive quantum rotation gate update mechanism, the population is guided to evolve towards the optimal region by updating the probability amplitude of the qubit; S7: Repeat steps S5 to S6 until the maximum number of iterations is reached, output the global optimal solution and perform space mapping to obtain the deployment optimization scheme.

2. The 5G base station deployment optimization method based on an improved quantum evolutionary algorithm according to claim 1, characterized in that, Step S1 includes: obtaining a set of user distribution points within the target area. ,in Represent the spatial coordinates of the j-th user; simultaneously obtain the existing base station set. ,in , Indicate the communication generation type of the base station; establish a coverage radius mapping table. Different physical coverage radii are matched for base stations of different generations, including And those to be deployed .

3. The 5G base station deployment optimization method based on an improved quantum evolutionary algorithm according to claim 2, characterized in that, Step S2 includes: Upgrade candidate pool construction: targeting existing base stations Calculate its evaluation score ,in For Euclidean distance, To evaluate the radius; if but Otherwise f=0; press Sites are sorted in descending order, and the top P-percentage sites are selected to enter the upgrade candidate pool. ; New candidate pool construction: for user distribution point set Using the K-Means clustering algorithm, users are divided into For each spatial cluster, calculate the centroid coordinates of that cluster. Use it as a new candidate pool ; Combinatorial encoding dimension: Determine the total optimization dimension .

4. The 5G base station deployment optimization method based on an improved quantum evolutionary algorithm according to claim 3, characterized in that, The objective function in step S3 is expressed as follows: ； in: Let V be the set of all base stations currently in the active state in the current scheme; V represents the logical OR operation; λ is the penalty coefficient; M is the total number of user points; Cost is the normalized construction cost, defined as: ； in, The site is in an open state. To upgrade the open status of candidate sites, This indicates that the creation of new candidate sites is enabled. and These are the upper limits for the construction scale of candidate sites for upgrading and candidate sites for creating new sites, respectively. and These are the first cost weight and the second cost weight, respectively.

5. The 5G base station deployment optimization method based on an improved quantum evolutionary algorithm according to claim 4, characterized in that, Step S4 includes: Initialize a quantum population containing n individuals Each quantum individual It consists of D qubits: ； Satisfying constraints In the initial state, let This ensures that all candidate sites are in a superposition state of equal probability between being open and closed.

6. The 5G base station deployment optimization method based on an improved quantum evolutionary algorithm according to claim 5, characterized in that, The method for obtaining discrete base station deployment schemes by observing superposition states in step S5 includes: quantum population... Take measurements for the first The first individual One qubit, generating a in Uniformly distributed random numbers in an interval The observation logic is: if Then the decision variable corresponding to the location A value of 0 indicates the base station is not activated; conversely, a value of 0 indicates the decision variable is active. A value of 1 indicates that the base station is enabled.

7. The 5G base station deployment optimization method based on an improved quantum evolutionary algorithm according to claim 6, characterized in that, The method for ensuring that the solution meets the preset construction indicators through heuristic repair logic in step S5 includes: Establish upgrade scale constraints: ,in, To upgrade the total number of stations, This represents the maximum number of upgrade stations. Establish new size constraints: ,in, This represents the total number of newly built stations. This represents the maximum number of newly built stations. If the observed solution violates the above constraints, the repair function is called to execute the forced flip strategy, specifically: first, excess amount calculation is performed, calculating the difference between the current number of stations opened and the upper limit. Next, random downsampling is performed, randomly selecting from all site indices marked "1" in the current solution. One site; perform a forced state flip, forcibly changing the state of the extracted site from 1 to 0.

8. The 5G base station deployment optimization method based on an improved quantum evolutionary algorithm according to claim 7, characterized in that, Step S6 includes: Using quantum rotating doors Update probability amplitude: ； Among them, the rotation step size Adaptive decay strategy is adopted: ； ； in, It is a function of rotation direction; This is the initial base step size; For the number of iterations The step size function for linear decay; T max This is the preset maximum number of iterations.

9. A 5G base station deployment optimization method based on an improved quantum evolutionary algorithm according to claim 8, characterized in that, The rotation direction function in step S6 The rotation direction is determined based on the comparison between the qubit state and the optimal solution bit. The decision logic is as follows: if the current global optimal solution... In the If the value of the bit is 1, and the current solution is 0, then the probability magnitude... Rotate the axis in the positive direction; conversely, if the current global optimal solution is... If the value is 0, and the current solution is 1, then the probability magnitude... Rotation of the axis in the positive direction.

Images