Method and apparatus for implementing communication network optimization

Abstract

The present invention discloses a method and apparatus for implementing communication network optimization. The method includes: determining the number m of cells to be optimized and the number k of base station antenna weight parameters; taking the product of the number m of cells to be optimized and the number k of base station antenna weight parameters as the dimension of the position vector, and taking the optimization direction of the antenna weights as the velocity vector of a single particle; setting the number N of particles in the particle swarm; for each particle, initializing the position vector using random numbers and initializing the velocity vector using a zero vector; using the particle swarm algorithm to iteratively update the position vector and the velocity vector of each particle until the iteration end condition is reached, and outputting the globally historically optimal value; determining each antenna weight according to the globally historically optimal value. Using the present invention, the requirements for large-scale and multi-parameter optimization of the communication network can be met, and the efficiency and timeliness of network optimization can be improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and in particular, to a method and device for realizing communication network optimization. Background Art

[0002] At present, with the large-scale application of 5G (the fifth-generation mobile communication technology), a communication network with high speed, low latency, and wide coverage has become an urgent need and the goal pursued by the communication industry. However, with the advent of 5G, network optimization has become a thorny problem. In order to achieve goals such as large connections, low latency, and high capacity, it is necessary to reasonably optimize and adjust the base station parameters. Different from the past, due to its technical characteristics, 5G often requires a larger number of base stations to achieve the same coverage range. Especially after the commercial use of 5G, the number of base stations has increased sharply, and a larger number of base stations means a corresponding sharp increase in the base station parameters. The cooperation among a large number of base station clusters is required to achieve high-quality network coverage, and the order of magnitude of the parameters of a large number of base stations has far exceeded the scope that can be accommodated by the traditional expert experience adjustment method. Moreover, manual adjustment may result in overlooking some aspects while attending to others, being unable to coordinate among multiple base stations, and even worse, there may be interference among multiple base stations. It is often impossible to quickly find the optimal solution, let alone achieve daily or hourly optimization according to the service characteristics. At the same time, good optimization highly depends on the experience of experts and network optimization knowledge. People without a lot of industry experience and network optimization knowledge cannot perform accurate, fast, and good parameter adjustment. Summary of the Invention

[0003] The present invention provides a method and device for realizing communication network optimization, which can meet the requirements of large-scale and multi-parameter optimization of communication networks and improve the efficiency and timeliness of network optimization.

[0004] To this end, the present invention provides the following technical solutions:

[0005] On the one hand, the present invention provides a method for realizing communication network optimization, and the method includes:

[0006] Determine the number m of cells to be optimized and the number k of base station antenna weight parameters;

[0007] Take the product of the number m of cells to be optimized and the number k of base station antenna weight parameters as the dimension of the position vector, and take the optimization direction of the antenna weight as the velocity vector of a single particle;

[0008] Set the number N of particles in the particle swarm;

[0009] For each particle, initialize the position vector with a random number and initialize the velocity vector with a zero vector;

[0010] Use the particle swarm optimization algorithm to iteratively update the position vector and velocity vector of each particle until the iteration end condition is reached, and output the global historical optimal value;

[0011] Determine each antenna weight according to the global historical optimal value.

[0012] Optionally, the antenna weight parameters include: azimuth angle, downward tilt angle, horizontal beamwidth, vertical beamwidth.

[0013] Optionally, for each particle, the initialization of the position vector using random numbers further includes:

[0014] When the random number is greater than the set range, use the boundary value of the set range to initialize the position vector.

[0015] Optionally, the setting of the number N of particles in the particle swarm includes:

[0016] Set multiple candidate particle swarm sizes;

[0017] Based on each candidate particle swarm size respectively, use the particle swarm optimization algorithm to iteratively update the position vector and velocity vector of each particle until the set stop condition is reached;

[0018] Determine the best candidate particle swarm size among them as the number N of particles in the particle swarm according to the iteration results corresponding to each candidate particle swarm size.

[0019] Optionally, the method further includes:

[0020] Determine the objective function value based on the updated position vector of each particle;

[0021] Determine the score of each particle according to the objective function value;

[0022] Determine the historical optimal value of each current particle and the global historical optimal value according to the score of each particle.

[0023] Optionally, the determination of the objective function based on the updated position vector of each particle includes:

[0024] Based on the updated position vector of each particle, simulate the reference signal received power Rsrp value and / or signal-to-interference-plus-noise ratio Sinr value of each grid in the target area;

[0025] Calculate the objective function value according to the Rsrp value and / or the Sinr value.

[0026] Optionally, the iteration end condition is: reaching the set maximum number of iterations, or the global historical optimal value reaching the set threshold.

[0027] On the other hand, the present invention further provides a device for optimizing a communication network. The device includes:

[0028] An unoptimized cell determination module, configured to determine the number m of unoptimized cells and the number k of base station antenna weight parameters; use the product of the number m of unoptimized cells and the number k of base station antenna weight parameters as the dimension of the position vector, and use the optimization direction of the antenna weights as the velocity vector of a single particle.

[0029] A particle scale determination module, configured to set the number N of particles in the particle swarm.

[0030] An initialization module, configured to initialize the position vector for each particle using a random number, and initialize the velocity vector using a zero vector.

[0031] An iterative processing module, configured to iteratively update the position vector and the velocity vector of each particle using the particle swarm algorithm until an iteration end condition is reached, and output the global historical optimal value.

[0032] An optimization parameter determination module, configured to determine each antenna weight according to the global historical optimal value.

[0033] Optionally, the initialization module is further configured to, when the random number is greater than a set range, initialize the position vector using the boundary value of the set range.

[0034] Optionally, the particle scale determination module includes:

[0035] A candidate particle swarm scale setting unit, configured to set multiple candidate particle swarm scales.

[0036] An iterative optimization processing unit, configured to respectively perform iterative update on the position vector and the velocity vector of each particle using the particle swarm algorithm based on each candidate particle swarm scale until a set stop condition is reached.

[0037] A selection unit, configured to determine the best candidate particle swarm scale among the iterative results corresponding to each candidate particle swarm scale as the number N of particles in the particle swarm.

[0038] Optionally, the iterative processing module is further configured to determine an objective function value based on the updated position vector of each particle, determine a score for each particle according to the objective function value, and determine the historical optimal value and the global historical optimal value of each current particle according to the score of each particle.

[0039] Optionally, the device further includes: a simulation module, configured to simulate the reference signal received power (Rsrp) value and / or the signal-to-interference-plus-noise ratio (Sinr) value of each grid in the target area based on the updated position vector of each particle, and calculate the objective function value according to the Rsrp value and / or the Sinr value;

[0040] The iterative processing module obtains the objective function value by invoking the simulation module.

[0041] The method and device for implementing communication network optimization provided by the present invention determine the number m of cells to be optimized and the number k of base station antenna weight parameters, use the product of the number m of cells to be optimized and the number k of base station antenna weight parameters as the dimension of the position vector, use the optimization direction of the antenna weights as the velocity vector of a single particle, and use the particle swarm algorithm to automatically calculate and optimize the parameters of the cells to be optimized. Compared with the traditional method of manually adjusting parameters on-site, the method and device for implementing communication network optimization provided by the present invention can realize automatic adjustment of network parameters, greatly save manpower, and improve the timeliness of adjustment at the same time.

[0042] Furthermore, when performing parameter adjustment, the solution of the present invention uses an artificial intelligence algorithm to realize automatic calculation and analysis of network parameters. During this process, no manual intervention or participation is required, and it does not rely on the network optimization-related knowledge and work experience of operators, enabling anyone to quickly start operating and achieve professional-level parameter optimization and adjustment, greatly reducing the personnel threshold and labor cost. Due to the high efficiency of the automatic algorithm, optimization solutions at the day level and hour level can be realized, which are more suitable for different service scenarios and achieve differentiated and personalized optimization solutions.

[0043] The artificial intelligence algorithm used in the solution of the present invention can autonomously learn the optimization solution according to historical data. Compared with the traditional expert experience algorithm, it has better universality and flexibility, is applicable to a variety of different types of service scenarios, and has certain self-adaptability and environmental perception. Description of the Drawings

[0044] Figure 1 is a schematic diagram of the velocity and position of particles during the iterative process of the existing particle swarm algorithm;

[0045] Figure 2 is a flowchart of a method for implementing communication network optimization according to the present invention;

[0046] Figure 3 is a schematic diagram of the initialization of the position vector in an embodiment of the present invention;

[0047] Figure 4 is a schematic diagram of the initialization of the velocity vector in an embodiment of the present invention;

[0048] Figure 5It is a schematic diagram showing the relationship between the particle swarm size and the communication network optimization effect in the embodiments of the present invention;

[0049] Figure 6 It is a schematic diagram showing the relationship between the particle swarm size and the computing resource consumption in the embodiments of the present invention;

[0050] Figure 7 It is a flowchart for iteratively updating the position vector and velocity vector of each particle using the particle swarm algorithm in the embodiments of the present invention;

[0051] Figure 8 It is an example of the historical optimal value of a single particle and the global historical optimal value in the embodiments of the present invention;

[0052] Figure 9 It is a schematic structural diagram of a device for implementing communication network optimization in the present invention. Detailed implementation manners

[0053] To enable those skilled in the art to better understand the solutions of the embodiments of the present invention, the embodiments of the present invention will be further described in detail below in conjunction with the accompanying drawings and implementation manners.

[0054] Aiming at the problem that in the current 5G network, the number of base stations has increased significantly, and the existing method for adjusting base station parameters no longer meets the requirements of such large-scale and multi-parameter optimization, the embodiments of the present invention provide a method and device for implementing communication network optimization, which realizes the optimization of network parameters based on historical network data and improves the efficiency and timeliness of network optimization.

[0055] The solution of the present invention uses the particle swarm algorithm to optimize and calculate network parameters. First, the particle swarm algorithm (Particle Swarm Optimization, PSO) will be briefly described below.

[0056] The particle swarm algorithm is a stochastic search algorithm based on group cooperation developed by simulating the foraging behavior of bird flocks, mainly involving the following key points:

[0057] 1) Initialize parameters:

[0058] Set the particle swarm size N, the dimension d of the position and velocity vectors. N represents how many particles there are in total. At each moment, each particle is in the solution space, that is, at each moment, each particle can calculate the value of the objective function. d represents the dimension of the solution space and is related to the objective function.

[0059] 2) Initialize particles:

[0060] For each particle, use random numbers to initialize the position vector Xi and the velocity vector Vi to make the particle positions as scattered as possible in the solution space.

[0061] 3) Iterative process:

[0062] For the t-th iteration, calculate the objective function value of each particle. For the i-th particle, record the position with the optimal objective function value of this particle in history as Pi(t), and record the position with the optimal objective function value of all particles in history as G(t).

[0063] Taking minimizing y = x1 2 + x2 2 as an example, for particle i, after calculating t values of y, take the position vector (x1, x2) corresponding to the minimum y as Pi(t). For all particles, after calculating N×t values of y, take the position vector (x1, x2) corresponding to the minimum y as G(t).

[0064] Refer to Figure 1 as shown, update the velocity and position for the next iteration. Vi(t) is the velocity of the particle at the start of the t-th iteration, and Xi(t) is the position of the particle at the start of the t-th iteration. Then:

[0065] Vi(t + 1) = wVi(t)+c1×r1×(Pi(t)-Xi(t)) + c2× r2× (G(t) - Xi(t));

[0066] Xi(t + 1) = Xi(t) + Vi(t + 1);

[0067] Among them, c1 and c2 are acceleration constants used to adjust the maximum learning step size; r1 and r2 are two random functions with a value range of [0,1] used to increase the search randomness; w is the inertia weight, a non-negative number, used to adjust the search range of the solution space.

[0068] 4) Determine convergence:

[0069] After reaching the maximum number of iterations or determining that the solution reaches a certain threshold, output the globally optimal historical optimal value.

[0070] As Figure 2 shown, it is the flowchart of the method for implementing communication network optimization in the present invention, including the following steps:

[0071] Step 201, determine the number m of cells to be optimized and the number k of base station antenna weight parameters.

[0072] In the embodiments of the present invention, the antenna weight parameters can be set as needed, for example, they can include but are not limited to the following: azimuth angle, downward tilt angle, horizontal wave width, vertical wave width.

[0073] Step 202, take the product of the number m of cells to be optimized and the number k of base station antenna weight parameters as the dimension of the position vector, and take the optimization direction of the antenna weight as the velocity vector of a single particle.

[0074] That is, the dimension of the position vector is set to m×k.

[0075] In the embodiment of the present invention, the position vector of a single particle is represented by Xi, and the position vector of each particle represents all parameters of all cells. For example, if 50 cells need to be optimized at one time, the dimension of the position vector is dimensions.

[0076] The dimension of the velocity vector is the same as that of the position vector.

[0077] Step 203, set the number N of particles in the particle swarm.

[0078] The number N of particles in the particle swarm represents the scale of the particle swarm. The position vector of each particle represents a solution, and the particle swarm scale N represents that there are N solutions. The particle swarm algorithm will select a global optimal value from these N solutions as the final result.

[0079] Step 204, for each particle, initialize the position vector with a random number and initialize the velocity vector with a zero vector.

[0080] For each particle, its position vector Xi(0) can be initialized with a random number, and its velocity vector Vi(0) can be initialized with a zero vector, where i = 1, 2, …, N.

[0081] In a specific application, the initial values of the position vectors Xi of all particles can be limited within a certain range. When the random number is greater than this range, the boundary value of the range can be selected.

[0082] As Figure 3 and Figure 4 shown, the schematic diagrams of the position vector initialization and the velocity vector initialization are respectively shown.

[0083] Step 205, use the particle swarm algorithm to iteratively update the position vector and the velocity vector of each particle until the iteration end condition is reached, and output the global historical optimal value.

[0084] Step 206, determine each antenna weight according to the global historical optimal value.

[0085] The iteration end condition is: reaching the set maximum number of iterations, or the global historical optimal value reaching the set threshold.

[0086] It should be noted that the setting of the particle swarm size N will affect the consumption degree of computer resources by the algorithm and the optimization effect. Specifically, when N is too large, it is easier to find the global optimal value, that is, the best parameter setting value, but at the same time, it will increase the consumption of computing resources; when N is too small, it may fall into a local optimal value and fail to find the actual global optimal value, which will affect the optimization effect to a certain extent. Figure 5 and Figure 6 respectively show the relationship between the particle swarm size and the communication network optimization effect, and the relationship between the particle swarm size and the computing resource consumption. Considering these situations, in practical applications, an appropriate N value can be set in combination with the actual situation.

[0087] In a non-limiting embodiment, when setting the particle swarm size N in the above step 203, it can be set according to empirical values.

[0088] Furthermore, in order to balance the relationship between the optimization effect and the computing resource consumption, in another non-limiting embodiment, multiple candidate particle swarm sizes can be set first; respectively based on each candidate particle swarm size, use the particle swarm algorithm to iteratively update the position vector and velocity vector of each particle until the set stop condition is reached; determine the best candidate particle swarm size among them as the number N of particles in the particle swarm according to the iteration results corresponding to each candidate particle swarm size. For example, first set a relatively conservative value, and then increment the value of N by a certain step size (such as 3 - 5) to obtain multiple candidate N values, and perform iterative optimization respectively based on multiple different N values but the same other parameters until the set stop condition is reached (the stop condition can specifically be iterating a certain number of times, or iterating for a certain period of time, or the optimization effect has no obvious change, etc.), and determine the best number N of particles in the particle swarm among them according to the iteration results corresponding to the number N of particles in the multiple different particle swarms. Specifically, it can be comprehensively considered according to the optimization effect and computing resource consumption in each case of N value, and the N value that best balances the optimization effect and computing resource consumption is selected.

[0089] It should be noted that in the solution of the present invention, the Rsrp (Reference Signal Receiving Power) and / or Sinr (Signal to Interference plus Noise Ratio) mentioned later can be used as the index parameters for evaluating the optimization effect, and the optimization effect represents the sum of the objective function values corresponding to the optimized cells. The optimization effect having no obvious change means that the change in the optimization effect does not exceed the set value, such as not exceeding 5%.

[0090] The following combines Figure 7The above iterative update process will be described in detail. The iterative process of this embodiment will be described by taking the achievement of the set maximum number of iterations as an example.

[0091] Referring to Figure 7 , Figure 7 is a flowchart for iteratively updating the position vector and velocity vector of each particle using the particle swarm algorithm in an embodiment of the present invention, including the following steps:

[0092] Step 701, input the initialization parameters: the particle size N, the position vector dimension d, the acceleration constants c1 and c2, the inertia weight w, the maximum number of iterations T, the random functions r1 and r2.

[0093] Among them, the position vector dimension d is the product of the number of cells m to be optimized mentioned above and the number of base station antenna weight parameters k.

[0094] Among them, the acceleration constant c1 is used to learn its own experience, and the acceleration constant c2 is used to learn the group experience. When the c value is too small, the convergence speed will be prolonged, and when the c value is too large, the speed will be too fast and the optimal value will be missed. In specific applications, the acceleration constants c1 and c2 can be determined according to experiments. For example, c1 = c2 = 2 is set.

[0095] Among them, the random function r1 is used to introduce random fluctuations for learning its own experience, and r2 is used to introduce random fluctuations for learning the group experience. The purpose of setting the random function is to introduce non-linear capabilities into the algorithm to prevent the algorithm from falling into local optimal values. The values of r1 and r2 should not be too large or too small. When they are too large, it will be difficult to converge due to excessive randomness, and when they are too small, it will fall into local optimal values due to insufficient non-linear capabilities, and thus the optimal effect cannot be achieved. For this reason, in the embodiment of the present invention, r1 and r2 can be set to random(0,1), and better results can be obtained.

[0096] Among them, the inertia weight w will affect the convergence speed of the algorithm. When the inertia weight w is too small, the convergence speed is too slow and it takes a long time or multiple iterations. When the inertia weight w is too large, the optimal value may be missed and the algorithm is also difficult to converge. For this reason, in the embodiment of the present invention, the value of the inertia weight w can be set in the following way: the value of w is set to linearly decrease as the number of iterations increases. Specifically, a larger value is set at the beginning to accelerate the convergence speed of the algorithm, and this value can be appropriately adjusted smaller as the number of iterations increases to achieve the optimal effect. For example, a linearly decreasing manner from 0.9 to 0.1 can be selected.

[0097] Among them, the maximum number of iterations T affects the final optimization effect. When the value of T is too small, the number of algorithm iterations is insufficient, and it ends before finding the optimal value, resulting in poor optimization effect. When the value of T is too large, it takes too much time, but the optimization effect does not increase significantly. Therefore, it is particularly important to select an appropriate maximum number of iterations, and a selection method similar to the particle swarm size N value can be used.

[0098] Step 702, input the initial particle positions X and the velocity vector V.

[0099] It should be noted that the particle position X refers to the position vectors of all particles in the particle swarm. Similarly, the velocity vector V refers to the velocity vectors of all particles.

[0100] Step 703, set the iteration round t = 0. Then start the iterative calculation, that is, the subsequent steps 704 to 707.

[0101] First, in step 704, determine the objective function value yi(t) based on the position vector Xi(t) of each particle.

[0102] In practical applications, the simulation engine can be used to perform simulation calculations on the input data (i.e., the position vectors of each particle in the current round), and calculate the objective function value yi(t) according to the simulation results. Specifically, input the weights of all cells (i.e., the position vectors of each particle in the current round) into the simulation software, and obtain the Rsrp value and / or Sinr value of each grid in the target area through simulation. It is recommended to select a target area that can contain all cells and is slightly larger as the target area, and the grid size is recommended to be set to a 20-meter or 50-meter grid. The larger the Rsrp value, the better the signal, and the better it is; the larger the Sinr value, the higher the ratio of the received signal strength to the noise, and the better it is. Then, calculate the objective function value yi(t) according to the simulation results.

[0103] For example, when considering both the Rsrp value and the Sinr value of each grid in the target area, the objective function value yi(t) can be calculated according to the following formula:

[0104] ;

[0105] Where:

[0106] avgRsrp ;

[0107] ;

[0108] Among them, ScoreMin is the minimum value of yi that is allowed to appear, ScoreMax is the maximum value of yi that is allowed to appear, area represents the number of grids: a1 and a2 are weight coefficients, and specific different values can be taken according to the business scenario. For example, in a business scenario that attaches great importance to Rsrp and does not pay attention to Sinr, a1 can be adjusted larger and a2 can be adjusted smaller.

[0109] Indicates the allowed minimum value of, generally set to -140; Indicates the allowed maximum value of, Indicates the allowed minimum value of, generally set to -126.

[0110] Among them, The unit of is dBm, and the unit of Sinr is dB.

[0111] It should be noted that in practical applications, the objective function may also have other forms, and the embodiments of the present invention do not limit this.

[0112] In step 705, determine the score of each particle according to the objective function value, and determine the historical optimal value of the particle.

[0113] Specifically, in one embodiment, the objective function value corresponding to each particle can be used as the score of the particle. Of course, in practical applications, it is not limited to this score calculation method.

[0114] The score of the particle can be used to represent the optimization effect.

[0115] In step 706, determine the historical optimal values of all particles, that is, the global historical optimal value.

[0116] After obtaining the score of each particle, the historical optimal value Pi of each current particle and the global historical optimal value G(t) can be determined according to the score of each particle. Specifically, after obtaining the score of each particle, the score can be compared with the historical optimal solution of the particle, and the larger value is assigned to the historical optimal solution of the particle; similarly, the global historical optimal solution can be obtained in a similar manner. At this time, the score of the particle represents the quality degree of the current position, that is, the parameter. The higher the score, the more excellent the parameter.

[0117] In step 707, for each particle i, update its position vector Xi(t) and velocity vector Vi(t).

[0118] Specifically, the update can be performed according to the following formula:

[0119] ;

[0120] Xi(t + 1) = Xi(t) + Vi(t + 1);

[0121] In step 708, it is determined whether the iteration number t is less than or equal to the maximum iteration number T; if so, step 709 is executed; otherwise, step 710 is executed.

[0122] In step 709, the global historical optimal value G(t) is output.

[0123] In step 710, the iteration number t is incremented by 1, and then the process returns to step 704 to execute the next iteration optimization process.

[0124] As Figure 8 shown, taking the example of 20 particles iterating to the tenth round, the historical optimal value of particle 1 is the position where the fifth round of iteration is located, and the global optimal value is the position where particle 7 is located in the sixth round of iteration.

[0125] The method for realizing communication network optimization provided by the present invention determines the number m of cells to be optimized and the number k of base station antenna weight parameters, takes the product of the number m of cells to be optimized and the number k of base station antenna weight parameters as the dimension of the position vector, takes the optimization direction of the antenna weight as the velocity vector of a single particle, and uses the particle swarm algorithm to automatically calculate and optimize the parameters of the cells to be optimized. Compared with the traditional method of manually adjusting parameters on site, the method for realizing communication network optimization provided by the present invention can realize automatic adjustment of network parameters, greatly save manpower, and improve the timeliness of adjustment at the same time.

[0126] Correspondingly, the present invention also provides a device for realizing communication network optimization, as Figure 9 shown, which is a schematic structural diagram of the device for realizing communication network optimization of the present invention.

[0127] The device 900 for realizing communication network optimization provided by this embodiment includes the following modules:

[0128] A cell to be optimized determination module 901, configured to determine the number m of cells to be optimized and the number k of base station antenna weight parameters; take the product of the number m of cells to be optimized and the number k of base station antenna weight parameters as the dimension of the position vector, and take the optimization direction of the antenna weight as the velocity vector of a single particle;

[0129] A particle scale determination module 902, configured to set the number N of particles in the particle swarm;

[0130] An initialization module 903, configured to initialize the position vector for each particle using random numbers and initialize the velocity vector using a zero vector;

[0131] The iterative processing module 904 is used to iteratively update the position vector and velocity vector of each particle using the particle swarm algorithm until the iterative end condition is reached, and output the global historical optimal value;

[0132] The optimization parameter determination module 905 is used to determine each antenna weight according to the global historical optimal value.

[0133] Furthermore, when the random number is greater than the set range, the initialization module 903 can also initialize the position vector using the boundary value of the set range, so as to limit the initial values of the position vectors Xi of all particles within a certain range.

[0134] It should be noted that in a specific embodiment, the particle swarm size N can be set according to empirical values. Considering that the setting of the particle swarm size N will affect the consumption degree of computer resources and the optimization effect of the algorithm. Specifically, when N is too large, it is easier to find the global optimal value, that is, the best parameter setting value, but at the same time, it will increase the consumption of computing resources; when N is too small, it may fall into a local optimal value and fail to find the actual global optimal value, which will affect the optimization effect to a certain extent. For this reason, in another specific embodiment, the particle size determination module 902 can first set multiple different N values, perform optimization iterations based on the same parameters and different N values, and select the best N value according to the iteration results. Correspondingly, in this embodiment, the particle size determination module 902 can include the following units:

[0135] The candidate particle swarm size setting unit is used to set multiple candidate particle swarm sizes;

[0136] The iterative optimization processing unit is used to respectively iteratively update the position vector and velocity vector of each particle using the particle swarm algorithm based on each candidate particle swarm size until the set stop condition is reached;

[0137] The selection unit is used to determine the best candidate particle swarm size among them as the number N of particles in the particle swarm according to the iteration results corresponding to each candidate particle swarm size.

[0138] In the embodiment of the present invention, the determination of the global historical optimal value can comprehensively consider the Rsrp value and / or Sinr value in the target area. For this reason, a target function related to the Rsrp value and / or Sinr value is set, and the score of each particle is determined according to the target function value, and then the global historical optimal value is obtained. Correspondingly, the iterative processing module 904 is also used to determine the target function value based on the updated position vector of each particle, determine the score of each particle according to the target function value, and determine the historical optimal value and global historical optimal value of each current particle according to the score of each particle.

[0139] For the convenience of calculation, a simulation module (not shown) may be set up to simulate the reference signal received power Rsrp value and / or the signal-to-interference-plus-noise ratio Sinr value of each grid in the target area based on the updated position vector of each particle, and calculate the objective function according to the Rsrp value and / or the Sinr value.

[0140] Correspondingly, the iterative processing module 904 may obtain the value of the objective function by calling the simulation module. Specifically, the iterative processing module 904 inputs the updated position vector of each particle into the simulation module, and after the simulation module calculates the objective function value, it outputs it to the iterative processing module 904.

[0141] The setting and calculation method of the objective function may refer to the description in the method embodiment of the present invention above, and will not be elaborated here.

[0142] The device for realizing communication network optimization provided by the present invention determines the number m of cells to be optimized and the number k of base station antenna weight parameters, takes the product of the number m of cells to be optimized and the number k of base station antenna weight parameters as the dimension of the position vector, takes the optimization direction of the antenna weight as the velocity vector of a single particle, and uses the particle swarm algorithm to automatically calculate and optimize the parameters of the cells to be optimized. Using the solution of the present invention, the optimization adjustment of the base station antenna parameters can be realized efficiently and accurately, without the need for manual on-site adjustment, saving manpower, improving timeliness, and reducing the limitation of weather on the adjustment; moreover, there is no need for the operator to have complete network optimization-related knowledge and work experience, reducing the threshold for users, and a continuous and stable optimization effect can be achieved.

[0143] The solution of the present invention uses an artificial intelligence algorithm, has characteristics such as environmental perception ability, wide application range, and rich adaptable service types, and is more universal than the expert experience algorithm, and can be adapted to different service scenarios.

[0144] It should be noted that the terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0145] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other, and the key point of each embodiment is to illustrate the differences from other embodiments. Moreover, the system embodiments described above are only illustrative. The modules and units described as separate components may or may not be physically separated, that is, they may be located on one network unit or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative work.

[0146] The embodiments of the present invention have been introduced in detail above. In this article, specific implementation manners are used to elaborate on the present invention. The description of the above embodiments is only used to help understand the method and device of the present invention. They are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by a person of ordinary skill in the art without creative work shall fall within the protection scope of the present invention. The content of this specification should not be construed as a limitation of the present invention. Therefore, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for realizing communication network optimization, characterized in that, The method includes: Determining the number m of cells to be optimized and the number k of base station antenna weight parameters; Taking the product of the number m of cells to be optimized and the number k of base station antenna weight parameters as the dimension of the position vector, and taking the optimization direction of the antenna weights as the velocity vector of a single particle; Setting the number N of particles in the particle swarm; For each particle, initializing the position vector using random numbers and initializing the velocity vector using a zero vector; Using the particle swarm algorithm to iteratively update the position vector and velocity vector of each particle until the iteration end condition is reached, and outputting the global historical optimal value; Determining each antenna weight according to the global historical optimal value; Wherein, the method further includes: Determining the objective function value based on the updated position vector of each particle; Determining the score of each particle according to the objective function value; Determining the historical optimal value and the global historical optimal value of each current particle according to the score of each particle; The determining the objective function based on the updated position vector of each particle includes: Based on the updated position vector of each particle, simulating the reference signal received power Rsrp value and the signal-to-interference-plus-noise ratio Sinr value of each grid in the target area; Calculating the objective function value yi(t) according to the Rsrp value and the Sinr value; The objective function value yi(t) is calculated according to the following formula: ； Where: avgRsrp ; ； Where ScoreMin is the minimum value of yi allowed to appear, ScoreMax is the maximum value of yi allowed to appear, area represents the number of grids: a1 and a2 are weight coefficients; Indicates allowed The minimum value of, set to -140; Indicates allowed The maximum value of, Indicates allowed The minimum value of, set to -126; Among them, The unit of is dBm, and the unit of Sinr is dB.

2. The method according to claim 1, wherein The antenna weight parameters include: azimuth angle, downward tilt angle, horizontal beamwidth, vertical beamwidth.

3. The method according to claim 1, wherein The initializing the position vector using random numbers for each particle further includes: When the random number is greater than the set range, initializing the position vector using the boundary value of the set range.

4. The method according to claim 1, characterized in that The setting the number N of particles in the particle swarm includes: Setting multiple candidate particle swarm sizes; Respectively based on each candidate particle swarm size, using the particle swarm algorithm to iteratively update the position vector and velocity vector of each particle until the set stop condition is reached; Determining the best candidate particle swarm size among them as the number N of particles in the particle swarm according to the iteration results corresponding to each candidate particle swarm size.

5. The method according to claim 1, wherein The iteration end condition is: reaching the set maximum number of iterations, or the global historical optimal value reaching the set threshold.

6. A device for implementing communication network optimization, characterized in that, The device includes: A cell to be optimized determination module, configured to determine the number m of cells to be optimized and the number k of base station antenna weight parameters; taking the product of the number m of cells to be optimized and the number k of base station antenna weight parameters as the dimension of the position vector, and taking the optimization direction of the antenna weights as the velocity vector of a single particle; A particle scale determination module, configured to set the number N of particles in the particle swarm; An initialization module, configured to, for each particle, initialize the position vector using random numbers and initialize the velocity vector using a zero vector; An iterative processing module, configured to iteratively update the position vector and velocity vector of each particle by using a particle swarm algorithm until an iterative end condition is reached, and output a global historical optimal value; An optimization parameter determination module, configured to determine each antenna weight according to the global historical optimal value; The iterative processing module is further configured to determine an objective function value based on the updated position vector of each particle, determine a score of each particle according to the objective function value, and determine the historical optimal value and global historical optimal value of each current particle according to the score of each particle; The apparatus further includes: a simulation module, configured to simulate the reference signal received power Rsrp value and / or signal-to-interference-plus-noise ratio Sinr value of each grid in a target area based on the updated position vector of each particle, and calculate an objective function value yi(t) according to the Rsrp value and the Sinr value; The objective function value yi(t) is calculated according to the following formula: ； Where: avgRsrp ; ； Among them, ScoreMin is the minimum value of yi that is allowed to appear, ScoreMax is the maximum value of yi that is allowed to appear, area represents the number of grids: a1 and a2 are weight coefficients; Indicates allowed The minimum value of, set to -140; Indicates allowed The maximum value of Indicates allowed The minimum value of, set to -126; Among them, The unit of is dBm, and the unit of Sinr is dB; The iterative processing module obtains the objective function value by calling the simulation module.

7. The apparatus according to claim 6, wherein The initialization module is further configured to, when the random number is greater than a set range, initialize the position vector by using the boundary value of the set range.

8. The device according to claim 6, wherein The particle scale determination module includes: A candidate particle swarm scale setting unit, configured to set a plurality of candidate particle swarm scales; An iterative optimization processing unit, configured to respectively iteratively update the position vector and velocity vector of each particle by using a particle swarm algorithm based on each candidate particle swarm scale until a set stop condition is reached; A selection unit, configured to determine the best candidate particle swarm scale among the iterative results corresponding to each candidate particle swarm scale as the number N of particles in the particle swarm.

Images