WO2023197585A1 - 无线网络用户感知优化方法、装置、电子设备和存储介质 - Google Patents

无线网络用户感知优化方法、装置、电子设备和存储介质 Download PDF

Info

Publication number
WO2023197585A1
WO2023197585A1 PCT/CN2022/130435 CN2022130435W WO2023197585A1 WO 2023197585 A1 WO2023197585 A1 WO 2023197585A1 CN 2022130435 W CN2022130435 W CN 2022130435W WO 2023197585 A1 WO2023197585 A1 WO 2023197585A1
Authority
WO
WIPO (PCT)
Prior art keywords
cell
target cell
target
user perception
load
Prior art date
Application number
PCT/CN2022/130435
Other languages
English (en)
French (fr)
Inventor
胡江华
林韫
冯刚
谢勤政
彭能
Original Assignee
中兴通讯股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Publication of WO2023197585A1 publication Critical patent/WO2023197585A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/02—Arrangements for optimising operational condition
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/08—Load balancing or load distribution
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks

Definitions

  • the present application relates to the field of wireless communication technology, and in particular, to a wireless network user perception optimization method, device, electronic equipment and storage medium.
  • Wireless network optimization is to find out the reasons that affect the network quality through traffic data analysis, on-site test data collection, parameter analysis, hardware inspection and other means on the existing running network, and through parameter modification, network structure adjustment, The process of adjusting equipment configuration and adopting certain technical means to ensure high-quality operation of the system, obtain the best benefits from existing network resources, and obtain the greatest benefits with the most economical investment. Among them, the improvement of user perception has long been a difficult problem in network optimization.
  • the traditional way to improve user perception is generally for wireless network optimization engineers to solve local hotspot problems in the wireless network in a targeted manner based on experience.
  • this optimization method is not applicable to the overall wireless network, cannot guarantee the accuracy of optimization, and has problems that need to be solved such as high optimization cost and low optimization efficiency.
  • the purpose of the embodiments of the present application is to provide a wireless network user perception optimization method, device, electronic device and storage medium to improve the accuracy of optimizing user perception and improve the efficiency of optimizing user perception.
  • embodiments of the present application provide an optimization method for wireless network user perception, including: obtaining the relationship between the capacity indicator and the user perception indicator; according to the relationship between the capacity indicator and the user perception indicator, Determine the target cell to be optimized; obtain the load distribution strategy of the target cell based on the load conditions of the target cell and the main neighboring cells of the target cell; adjust the target cell and the target cell according to the load distribution strategy Network configuration parameters of the primary neighbor.
  • Embodiments of the present application also provide an optimization device for wireless network user perception, including: a capacity-aware relationship acquisition module for acquiring the relationship between capacity indicators and user-awareness indicators; and an optimization target determination module for determining according to the The relationship between the capacity indicator and the user perception indicator determines the target cell to be optimized; the allocation strategy acquisition module is used to obtain the load of the target cell based on the load conditions of the target cell and the main neighboring cells of the target cell. Distribution strategy; an adjustment and optimization module, configured to adjust network configuration parameters of the target cell and the main neighboring cell according to the load distribution strategy.
  • Embodiments of the present application also provide an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions Executed by at least one processor, so that at least one processor can execute the above wireless network user perception optimization method.
  • An embodiment of the present application also provides a computer-readable storage medium storing a computer program.
  • the computer program is executed by a processor, the above-mentioned wireless network user perception optimization method is implemented.
  • the relationship between the capacity index and the user perception index is first obtained, and the relationship between the user perception and the cell capacity can be learned, and then the user perception can be optimized by adjusting the cell capacity. Further, the target cell to be optimized is determined based on the relationship between the capacity index and the user perception index, and the target cell that needs to be optimized for user perception can be determined. Then, based on the load conditions of the target cell and the main neighboring cells of the target cell, the load distribution strategy of the target cell is obtained. For the target cell that needs to be optimized, a personalized load distribution strategy for the cell can be obtained based on the load conditions of the target cell and its main neighboring cells, effectively improving the accuracy of optimizing user perception and improving the effect of user perception optimization.
  • the network configuration parameters of the target cell and the main neighboring cells can be adjusted according to the load distribution strategy, and finally the user perception of the target cell can be improved.
  • the wireless network user perception optimization method provided by this application is applicable to each cell in the wireless network. Therefore, there is no need to continuously adjust the network optimization method for each cell, and can effectively improve the efficiency of optimizing user perception.
  • Figure 1 is a flow chart of a method for optimizing user perception of a wireless network according to an embodiment of the present application
  • Figure 2 is a correlation curve diagram between the capacity index and the user perception index of cell A in an embodiment of the present application
  • Figure 3 is a correlation curve diagram between the capacity index and the user perception index of cell B in an embodiment of the present application
  • Figure 4 is a correlation curve diagram between the capacity index and the user perception index of cell C in an embodiment of the present application
  • Figure 5 is a schematic diagram of load distribution of the target cell and the removable cell in an embodiment of the present application.
  • Figure 6 is a correlation curve diagram between the capacity index and the user perception index of cell A in an embodiment of the present application.
  • Figure 7 is a correlation curve diagram between the capacity index and the user perception index of cell B in an embodiment of the present application.
  • Figure 8 is a correlation curve diagram between the capacity index and the user perception index of Cell C in an embodiment of the present application.
  • Figure 9 is a schematic structural diagram of an optimization device for wireless network user perception according to an embodiment of the present application.
  • FIG. 10 is a schematic structural diagram of an electronic device according to another embodiment of the present application.
  • An implementation manner of the present application relates to a method for optimizing user perception of a wireless network. It is worth noting that the execution subject of this embodiment is a device such as a computer that can implement the above-mentioned optimization method for wireless network user perception, and there is no specific limitation here.
  • the relationship between the capacity index and the user perception index is obtained; the target cell to be optimized is determined based on the relationship between the capacity index and the user perception index; based on the relationship between the target cell and the target cell
  • the load distribution situation of the main neighboring cells is used to obtain the load distribution strategy of the target cell; and the network configuration parameters of the target cell and the main neighboring cells are adjusted according to the load distribution strategy.
  • Step 101 Obtain the relationship between the capacity index and the user perception index.
  • the capacity indicator involved in this step can generally be the number of users in the cell, and the user perception indicators can include user uplink rate, user downlink rate, user delay and other indicators that can reflect user perception.
  • the process of obtaining the relationship between the capacity index and the user perception index in this step may include the correlation between the capacity index and the user perception index, the correlation coefficient, the slope of the relationship curve, etc.
  • the capacity indicator and the user perception indicator may also include: obtaining the historical capacity indicator and the historical user perception indicator; in this example, the acquisition of the relationship between the capacity indicator and the user perception indicator The relationship between the two includes: determining the relationship between the capacity indicator and the user perception indicator based on the historical capacity indicator and the historical user perception indicator. Using historical data of the community to determine the relationship between capacity indicators and user perception indicators for the community can improve the credibility of the relationship between cell capacity and user perception, thereby improving the accuracy of network perception optimization.
  • Step 102 Determine the target cell to be optimized based on the relationship between the capacity index and the user perception index.
  • the relationship between the capacity index and the user perception index is a cell in which the user perception index becomes worse as the capacity index increases, and its current capacity index (i.e., the number of users) is greater than the target capacity index (i.e., the target number of users),
  • the cell whose current user perception index is worse than the target user perception index is the target cell. It is worth noting that the target number of users and target user perception indicators can be determined specifically based on the optimization scenario of the cell.
  • the user perception index becomes worse as the capacity index increases, indicating that the user perception index of the cell can be better when the capacity index decreases, and the current capacity index (i.e., the number of users) of the cell is greater than the target capacity index (i.e., the target number of users) It indicates that users need to move out of this community.
  • the current user perception indicators are worse than the target user perception indicators, indicating that the user perception indicators of the community are insufficient and need to be optimized and improved.
  • the target cell to be optimized may also include: determining a cell that can be moved into according to the relationship between the capacity indicator and the user perception indicator; and determining, among the cells that can be moved into, The cell whose historical handover times with the target cell is greater than a preset value is regarded as the main neighboring cell of the target cell.
  • the neighboring cell with a greater number of handovers to the target cell is determined as the main neighboring cell that shares the load for the target cell.
  • the moveable communities involved in this example can include the following two categories:
  • a type of remigable cell a cell where the correlation relationship decreases monotonically and the absolute value of the Pearson correlation coefficient ⁇ is greater than the preset value k (which means that the user perception index of the cell becomes worse as the number of users increases), and the cell’s
  • the current user perception index is better than the target perception index and the current user number is smaller than the target user number.
  • the second type of removable community the correlation relationship decreases monotonically (indicating that the user perception index of this community becomes better as the number of users increases), and the current user perception index y and the target perception index Y exist y>Y*(1+ ⁇ ) relationship (which means that the user perception index of the cell is higher than the target user perception index to a large extent). It is worth noting that the k and ⁇ mentioned above are all hyperparameters.
  • Step 103 Obtain the load distribution strategy of the target cell based on the load conditions of the target cell and the main neighboring cells of the target cell.
  • obtaining the load distribution strategy of the target cell based on the load conditions of the target cell and the main neighboring cells of the target cell may include: based on the current load number and the target load number of the target cell. , determine the number of loads that the target cell needs to move out of; determine the number of loads that the main neighboring cell can move in according to the current load number and target load number of the main neighboring cell; wherein, the target cell and the target load number The target load number of the main neighboring cell is determined according to the optimization requirement; based on the number of loads that need to be moved out and the number of loads that can be moved in, the load distribution strategy of the target cell is obtained.
  • the load distribution strategy of the target cell is obtained, which can make the load distribution strategy conform to the load conditions of the target cell and main neighboring cells. This makes subsequent optimization adjustments more targeted and more adaptable to the optimization needs of the target community, that is, more accurate network optimization.
  • the load distribution strategy involved in this implementation mode can be determined according to any of the following algorithms: greedy algorithm, genetic algorithm or simulated annealing algorithm. It is worth mentioning here that the algorithm used to determine the load distribution strategy in this embodiment is not limited to the above three algorithms. The above load distribution strategy can also be determined using other applicable optimization algorithms, which is not specifically limited in this embodiment.
  • Step 104 Adjust network configuration parameters of the target cell and the main neighboring cell according to the load distribution policy.
  • adjusting the network configuration parameters of the target cell and the main neighboring cell according to the load distribution strategy may include: determining each of the target cell and the main neighboring cell according to the load distribution strategy.
  • the number of migrated loads using the number of migrated loads and pre-collected network configuration parameters as training data, train to obtain a mapping relationship model between the number of migrated loads and network configuration parameters; determine the target cell according to the mapping relationship model and the respective target network configuration parameters of the main neighboring cell, and adjust the network configuration parameters of the target cell and the main neighboring cell according to the target network configuration parameters.
  • the mapping relationship model between the number of migrated loads and network configuration parameters is used to convert the determined number of load migrations between the target cell and the main neighboring cells into easily adjustable network configuration parameters, which can reduce the difficulty of user perception optimization. .
  • the number of migrated loads and network configuration parameters used as training data are in one-to-one correspondence, and the number of migrated loads and their corresponding network configuration parameters form a data group, which is used as training data for the model. Conduct training.
  • the number of migrated loads and corresponding network configuration parameters can be data collected by relevant technical personnel in their daily work.
  • mapping relationship model between the migrated load number and network configuration parameters involved in this example can be trained based on any of the following models: linear regression model, tree model, or deep neural network model. It is worth noting that the training model here is not limited to the above three models, and other applicable models can also be used to determine the mapping relationship model, which is not specifically limited in this implementation.
  • This implementation method analyzes the fitting curve between the user perception index and the capacity index of the cell to efficiently filter and classify, and determine which cells can effectively improve the user perception index after the load is transferred, and which cells can receive the incoming users and User perception metrics do not deteriorate below the threshold.
  • the load between the main neighbor cell pairs can be adjusted to the most appropriate range, so that the cell with capacity problem can reach the target value of the perception optimization index to the greatest extent.
  • the trained model is used to calculate the network configuration parameters between each cell that need to be adjusted on the network management side.
  • the above network optimization process quickly determines regional-level capacity optimization strategies and outputs optimization parameters, which can reduce reliance on expert experience.
  • data analysis and optimization plan output for 6,000 cells can be completed within one hour.
  • efficiency is significantly improved.
  • the wireless network user perception optimization method provided by this embodiment can effectively reduce the proportion of cells with low user perception, and the specific reduction rate can range from 3% to 9%.
  • the relationship between the capacity index and the user perception index is first obtained, and the relationship between the user perception and the cell capacity can be learned, and then the user perception can be optimized by adjusting the cell capacity.
  • the target cell to be optimized is determined based on the relationship between the capacity index and the user perception index, and the target cell that needs to be optimized for user perception can be determined. Then, based on the load conditions of the target cell and the main neighboring cells of the target cell, the load distribution strategy of the target cell is obtained. For the target cell that needs to be optimized, a personalized load distribution strategy for the cell can be obtained based on the load conditions of the target cell and its main neighboring cells, effectively improving the accuracy of optimizing user perception and improving the effect of user perception optimization.
  • the network configuration parameters of the target cell and the main neighboring cells can be adjusted according to the load distribution strategy, and finally the user perception of the target cell can be improved.
  • the wireless network user perception optimization method provided by this application is applicable to each cell in the wireless network. Therefore, there is no need to continuously adjust the network optimization method for each cell, and can effectively improve the efficiency of optimizing user perception.
  • Another embodiment of the present application relates to a method for optimizing user perception of a wireless network. It is worth noting that the main difference between this embodiment and the previous embodiment is that in this embodiment, the relationship between the obtained capacity index and the user perception index is specifically the relationship between the number of users and the user downlink rate.
  • N days of historical data which may specifically include: the number of historical users and historical user downlink rate indicators.
  • Gaussian fitting is performed on the historical number of users and the historical user downlink rate index of each cell to determine the correlation between the number of users in each cell and the user downlink rate index (i.e., between the capacity index and the user perception index involved in the previous embodiment). ) and Pearson correlation coefficient ⁇ .
  • the target downlink rate index and its corresponding target user number ie, the target load number involved in the previous embodiment
  • the current downlink rate index and its corresponding user number are used.
  • the relationship between the numbers is used to select the target community and the community where users need to move out. It is worth noting that the target downlink rate indicator and target number of users involved here can be individually determined based on the current optimization needs.
  • the correlation diagram between the number of users and the user downlink rate index corresponding to cell A is shown in Figure 2.
  • its target downlink rate index and target number of users are determined as Y ms and X respectively, as shown in the coordinates in Figure 2 ( As shown in ) shown.
  • the correlation relationship in Cell A decreases monotonically and the absolute value of the Pearson correlation coefficient ⁇ is greater than the preset value k, indicating that the downlink rate index of users in Cell A decreases as the number of users increases and the two are strongly correlated. relation.
  • the current user downlink rate index is lower than the target downlink rate index, and the current user number is greater than the target user number, indicating that cell A needs to optimize user perception and move out users, so cell A is determined to be the target cell from which users need to be moved out.
  • the correlation diagram between the number of users and the user downlink rate index corresponding to cell B is shown in Figure 3.
  • cell B its target downlink rate index and target number of users are determined as Y ms and X respectively, as shown in the coordinates in Figure 3 ( X, Y) shown.
  • the current downlink rate index is the average downlink rate index y ms of the user in the previous N days, and the current number of users is determined as the number of users x corresponding to the current downlink rate index, as shown in the coordinates (x, y) in Figure 3.
  • the correlation relationship in cell B decreases monotonically and the absolute value of the Pearson correlation coefficient ⁇ is greater than the preset value k, indicating that the downlink rate indicator for cell B users decreases as the number of users increases.
  • the current user downlink rate index of cell B is higher than the target downlink rate index and the current number of users is less than the target number of users. This means that cell B can receive fewer migrating users. Therefore, cell B is determined to be a migrating cell and is specifically determined. It is a type of community that can be moved into.
  • the correlation diagram between the number of users and the user downlink rate index corresponding to cell C is shown in Figure 4.
  • its target downlink rate index and target number of users are determined as Y ms and X respectively, as shown in the coordinates in Figure 4 ( X, Y) shown.
  • the current downlink rate index is the average downlink rate index y ms of the user in the previous N days, and the current number of users is determined as the number of users x corresponding to the current downlink rate index, as shown in the coordinates (x, y) in Figure 4.
  • the correlation of cell C decreases monotonically, and the current user downlink rate index y of cell C and the target downlink rate index Y have a relationship of y>Y*(1+ ⁇ ), indicating that cell C can receive more
  • Community C is determined to be a community that can be moved into, and is specifically determined to be a type II community that can be moved into.
  • the load situation of the target cell and the main neighboring cells of the target cell is determined, specifically including: the number of loads that need to be moved out of the target cell and the number of loads that can be moved into the cell.
  • the main neighboring cells of the target cell are determined based on the number of historical handovers with the target cell. Specifically, the cell with a higher number of handovers (which may be higher than a preset value) with the target cell is the main neighboring cell.
  • the target cell determined at this time is the target cell that ultimately needs to be adjusted and optimized.
  • the load distribution strategy of the target cell is obtained based on the load conditions of the target cell and the main neighboring cells of the target cell.
  • the principle of load distribution strategy is to maximize the utilization of neighbor relationships.
  • a greedy-like algorithm is used to allocate in the following order (hereinafter, the load situation of each cell shown in Figure 5 is For illustration, it is worth noting that the cells on the left shown in Figure 5 are target cells, and the cells shown on the right are cells that can be moved into. Each value is the number of loads that need to be moved out of the target cell and the cells that can be moved into. Number of loads that can be moved in):
  • each time allocation starts from the target cell that has not been cycled to, how many users can be moved in to all the main neighboring cells corresponding to the current target cell. Then, according to the number of users who need to move out of the target cell, the allocation is based on the proportion of the number of users who can move in in each main neighboring cell. Then, the number of migratable users in each community after migration in and out is updated, and the cycle starts again. After completing a round of user allocation in the target cell, recheck whether there are any neighboring cells that can continue to receive incoming users.
  • network configuration parameters of the target cell and the main neighboring cell are adjusted according to the load distribution strategy. Specifically, you can choose empirical models or linear regression models, tree models, and algorithm models such as DNN (Deep Neural Network) to train to obtain the mapping relationship model between the number of migrated loads and network configuration parameters, and obtain the mapping relationship between each community.
  • DNN Deep Neural Network
  • the network configuration parameters that need to be adjusted are then connected to the network management between the target cell and its main neighboring cells to complete the modification of the network configuration parameters.
  • the network configuration parameters are specifically CIO (Cell Individual Offset, cell-level individual offset) parameters.
  • Another embodiment of the present application relates to a method for optimizing user perception of a wireless network. It is worth noting that the main difference between this embodiment and the previous embodiment is that in this embodiment, the relationship between the obtained capacity index and the user perception index is specifically the relationship between the number of users and the user delay.
  • N days of historical data which can specifically include: the number of historical users and the historical user delay. Then perform linear fitting on the number of users and user delay in each cell, and determine the correlation between the number of users in each cell and user delay (that is, the relationship between the capacity indicators and user perception indicators mentioned above) and the Pearson correlation Coefficient ⁇ .
  • the relationship between the target delay and its corresponding target user number (that is, the target load number involved in the foregoing content) and the current delay and its corresponding user number are used. Relationship, respectively select the target community and the community where users need to move out. It is worth noting that the target latency and target number of users involved here can be individually determined based on the current optimization needs.
  • the correlation diagram between the number of users and user delay corresponding to cell A is shown in Figure 6.
  • its target delay and target number of users are determined as Y ms and X respectively, as shown in Figure 6.
  • the coordinates (X, Y); the current delay is the user's average delay y ms in the previous N days, and the current number of users is determined as the number of users x corresponding to the current delay, as shown in the coordinates (x, y) in Figure 6.
  • the correlation relationship in cell A increases monotonically and the absolute value of the Pearson correlation coefficient ⁇ is greater than the preset value k, indicating that the user delay in cell A increases as the number of users increases and the two are strongly correlated. .
  • the current user delay is higher than the target delay, and the current user number is greater than the target user number, indicating that Cell A needs to optimize user perception and migrate users out, so Cell A is determined to be the target cell from which users need to be migrated.
  • the correlation diagram between the number of users and user delay corresponding to cell B is shown in Figure 7.
  • cell B its target delay and target number of users are determined as Y ms and X respectively, as shown in the coordinates (X, Y) is shown.
  • the current delay is the user's average delay y ms in the previous N days, and the current number of users is determined as the number of users x corresponding to the current delay, as shown in the coordinates (x, y) in Figure 7.
  • the correlation relationship of cell B increases monotonically and the absolute value of the Pearson correlation coefficient ⁇ is greater than the preset value k, indicating that the user delay for cell B increases as the number of users increases.
  • the current user delay of Cell B is lower than the target delay and the current number of users is less than the target number of users, indicating that Cell B can receive fewer migrating users. Therefore, Cell B is determined to be a remigable cell, and is specifically determined to be a cell. Classes can move into the community.
  • the correlation diagram between the number of users and user delay corresponding to cell C is shown in Figure 8.
  • its target delay and target number of users are determined as Y ms and X respectively, as shown in the coordinates (X, Y) is shown.
  • the current delay is the user's average delay y ms in the previous N days, and the current number of users is determined as the number of users x corresponding to the current delay, as shown in the coordinates (x, y) in Figure 8.
  • the load situation of the target cell and the main neighboring cells of the target cell is determined, specifically including: the number of loads that need to be moved out of the target cell and the number of loads that can be moved into the cell.
  • the main neighboring cells of the target cell are determined based on the number of historical handovers with the target cell. Specifically, the cell with a higher number of handovers (which may be higher than a preset value) with the target cell is the main neighboring cell.
  • the target cell determined at this time is the target cell that ultimately needs to be adjusted and optimized.
  • the load distribution strategy of the target cell is obtained based on the load conditions of the target cell and the main neighboring cells of the target cell.
  • the principle of load distribution strategy is to maximize the utilization of neighbor relationships.
  • the population size can be set to 50 to 500.
  • genes the number of migrating users in each cell pair
  • individuals the collection of all genes. Since the number of migrated users is basically not an integer, floating point encoding is used here. In order to avoid that too many main neighbor cells can move in beyond the accommodating range during initialization, each main neighbor cell and its neighbors are divided into a group.
  • the dirichlet distribution uses the dirichlet distribution to randomly set a probability value of 0 to 1 and assign it to each
  • Crossover operation Traverse all individuals. If the random value (0 ⁇ 1) is less than the crossover threshold (generally set to 0.6 ⁇ 1), the crossover position cross_point will be randomly generated on the current individual, and only the genes before the crossover position will be intercepted, and then Randomly select another individual, obtain the genes after the cross_point, splice them together, and then add the new individual to the new population.
  • the crossover threshold generally set to 0.6 ⁇ 1
  • Mutation operation Traverse all individuals. If the random value (0 ⁇ 1) is less than the mutation threshold (generally set to 0.0001 ⁇ 0.1), the mutation position mutate_point is randomly generated. Since each cell has a unique value for the individual (assigned user value) Encoded by floating point numbers, binary inversion cannot be performed during mutation, so random Gaussian noise is added or subtracted to randomly selected genes. At the same time, it is necessary to ensure that the genes of new individuals are not negative (negative values are set to 0).
  • the individual corresponding to the maximum/minimum fitness is selected from the optimal population as the optimal load allocation strategy.
  • network configuration parameters of the target cell and the main neighboring cell are adjusted according to the load distribution strategy. Specifically, you can choose empirical models or linear regression models, tree models, and algorithm models such as DNN (Deep Neural Network) to train to obtain the mapping relationship model between the number of migrated loads and network configuration parameters, and obtain the mapping relationship between each community.
  • DNN Deep Neural Network
  • the network configuration parameters that need to be adjusted are then connected to the network management between the target cell and its main neighboring cells to complete the modification of the network configuration parameters.
  • the network configuration parameters are specifically CIO (Cell Individual Offset, cell-level individual offset) parameters.
  • One embodiment of the present application relates to an optimization device for wireless network user perception, as shown in Figure 9, including:
  • Capacity awareness relationship acquisition module 901 used to acquire the relationship between capacity indicators and user perception indicators
  • the optimization target determination module 902 is used to determine the target cell to be optimized based on the relationship between the capacity indicator and the user perception indicator;
  • the distribution strategy acquisition module 903 is configured to obtain the load distribution strategy of the target cell based on the load conditions of the target cell and the main neighboring cells of the target cell;
  • the adjustment and optimization module 904 is configured to adjust network configuration parameters of the target cell and the main neighboring cell according to the load distribution strategy.
  • the allocation policy acquisition module 903 can also be used to determine the load number that needs to be moved out of the target cell according to the current load number and target load number of the target cell; according to the current load number of the main neighboring cell; The load number and the target load number are determined to determine the load number that the main neighboring cell can move into; wherein the target load number of the target cell and the main neighboring cell is determined according to the optimization demand; based on the need.
  • the number of loads that are moving out and the number of loads that can be moved in are used to obtain the load distribution strategy of the target cell.
  • the adjustment and optimization module 904 can also be used to determine the number of migrating loads of the target cell and the main neighboring cell according to the load distribution strategy; combine the number of migrating loads and the pre-collected
  • the network configuration parameters are used as training data, and the mapping relationship model of the migrated load number and the network configuration parameters is obtained through training; the target network configuration parameters of the target cell and the main neighboring cell are determined according to the mapping relationship model, and the target network configuration parameters are determined according to the target
  • the network configuration parameters adjust the network configuration parameters of the target cell and the main neighbor cell.
  • the device for optimizing wireless network user perception may further include: a historical data acquisition module (not shown in the figure), configured to acquire the historical capacity index before acquiring the relationship between the capacity index and the user perception index. and historical user perception metrics.
  • the capacity-aware relationship acquisition module 901 is configured to determine the relationship between the capacity index and the user-perception index based on the historical capacity index and the historical user-perception index.
  • the wireless network user perception optimization device may further include: a main neighbor cell acquisition module (not shown in the figure), configured to determine the target cell to be optimized according to the capacity index and user perception after determining the target cell to be optimized.
  • the relationship between indicators is used to determine a cell that can be moved into; among the cells that can be moved into, a cell whose historical switching times with the target cell is greater than a preset value is determined as the main neighboring cell of the target cell.
  • the wireless network user perception optimization device provided in this embodiment first obtains the relationship between the capacity index and the user perception index, can learn the relationship between the user perception and the cell capacity, and can further optimize the user perception by adjusting the cell capacity.
  • the target cell to be optimized is determined based on the relationship between the capacity index and the user perception index, and the target cell that needs to be optimized for user perception can be determined. Then, based on the load conditions of the target cell and the main neighboring cells of the target cell, the load distribution strategy of the target cell is obtained. For the target cell that needs to be optimized, a personalized load distribution strategy for the cell can be obtained based on the load conditions of the target cell and its main neighboring cells, effectively improving the accuracy of optimizing user perception and improving the effect of user perception optimization.
  • the network configuration parameters of the target cell and the main neighboring cells can be adjusted according to the load distribution strategy, and finally the user perception of the target cell can be improved.
  • the wireless network user perception optimization method provided by this application is applicable to each cell in the wireless network. Therefore, there is no need to continuously adjust the network optimization method for each cell, and can effectively improve the efficiency of optimizing user perception.
  • each module involved in the above-mentioned embodiments of the present application is a logical module.
  • a logical unit can be a physical unit, or a part of a physical unit, or it can be in the form of multiple The combination of physical units is realized.
  • units that are not closely related to solving the technical problems raised in this application are not introduced in this embodiment, but this does not mean that other units do not exist in this embodiment.
  • An embodiment of the present application also provides an electronic device, as shown in Figure 10, including at least one processor 1001; and a memory 1002 communicatively connected to the at least one processor 1001; wherein the memory 1002 stores information that can be at least Instructions executed by one processor 1001 are executed by at least one processor 1001, so that at least one processor 1001 can execute the above-mentioned optimization method for wireless network user perception.
  • the memory 1002 and the processor 1001 are connected using a bus.
  • the bus may include any number of interconnected buses and bridges.
  • the bus connects various circuits of one or more processors 1001 and the memory 1002 together.
  • the bus may also connect various other circuits together such as peripherals, voltage regulators, and power management circuits, which are all well known in the art and therefore will not be described further herein.
  • the bus interface provides the interface between the bus and the transceiver.
  • a transceiver may be one element or may be multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium.
  • the data processed by the processor 1001 is transmitted on the wireless medium through the antenna. Furthermore, the antenna also receives the data and transmits the data to the processor 1001.
  • Processor 1001 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions.
  • the memory 1002 may be used to store data used by the processor 1001 when performing operations.
  • Embodiments of the present application also provide a computer-readable storage medium storing a computer program.
  • the computer program is executed by the processor, the above-mentioned optimization method for wireless network user perception is implemented.
  • the program is stored in a storage medium and includes several instructions to make a device (which may be A microcontroller, a chip, etc.) or a processor (processor) executes all or part of the steps of the methods described in various embodiments of this application.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program code. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本申请涉及无线通信技术领域，公开了一种无线网络用户感知优化方法、装置、电子设备和存储介质。本申请中，该无线网络用户感知优化方法，包括：获取容量指标和用户感知指标之间的关系；根据容量指标和用户感知指标之间的关系，确定待优化的目标小区；基于目标小区与目标小区的主要邻区的负荷情况，获取目标小区的负荷分配策略；根据负荷分配策略，调整目标小区与主要邻区的网络配置参数。

Description

无线网络用户感知优化方法、装置、电子设备和存储介质
相关申请
本申请要求于2022年4月13号申请的、申请号为202210388264.3的中国专利申请的优先权。
技术领域
本申请涉及无线通信技术领域，尤其是涉及一种无线网络用户感知优化方法、装置、电子设备和存储介质。
背景技术
随着科学技术的不断进步，无线通信网络技术也得到了不断的发展。与此同时，无线网络优化已成为无线通信网络技术领域越来越为人关注的课题。无线网络优化是通过对现有已运行的网络进行话务数据分析、现场测试数据采集、参数分析、硬件检查等手段，找出影响网络质量的原因，并且通过参数的修改、网络结构的调整、设备配置的调整和采取某些技术手段，确保系统高质量的运行，使现有网络资源获得最佳效益，以最经济的投入获得最大的收益的过程。其中，用户感知的提升长期以来一直是网络优化中的一个难题。
提升用户感知的传统方式，一般是由无线网优工程师根据经验，对无线网络中的局部热点问题有针对性的解决。然而这种优化方式并不适用于无线网络全局，无法保证优化的准确性，且存在优化成本较高、优化效率低下等亟待解决的问题。
发明内容
本申请实施方式的目的在于提供一种无线网络用户感知优化方法、装置、电子设备和存储介质，用以提高优化用户感知的准确度，且提高优化用户感知的效率。
为了解决上述问题，本申请的实施方式提供了一种无线网络用户感知的优化方法，包括：获取容量指标和用户感知指标之间的关系；根据所述容量指标和用户感知指标之间的关系，确定待优化的目标小区；基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目标小区的负荷分配策略；根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数。
本申请的实施方式还提供了一种无线网络用户感知的优化装置，包括：容量感知关系获取模块，用于获取容量指标和用户感知指标之间的关系；优化目标确定模块，用于根据所述容量指标和用户感知指标之间的关系，确定待优化的目标小区；分配策略获取模块，用于基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目标小区的负荷分配策略；调整优化模块，用于根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数。
本申请的实施方式还提供了一种电子设备，包括：至少一个处理器；以及，与所述至少一个处理器通信连接的存储器；其中，存储器存储有可被至少一个处理器执行的指令，指令 被至少一个处理器执行，以使至少一个处理器能够执行上述的无线网络用户感知优化方法。
本申请的实施方式还提供了一种存储有计算机程序的计算机可读存储介质，计算机程序被处理器执行时实现上述的无线网络用户感知优化方法。
在本申请的实施方式中，首先获取容量指标和用户感知指标之间的关系，能够获知用户感知与小区容量的关系，进而能够通过调整小区容量实现优化用户感知。进一步地，根据所述容量指标和用户感知指标之间的关系，确定待优化的目标小区，能够确定需要进行用户感知优化的目标小区。再基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目标小区的负荷分配策略。对于需进行优化的目标小区，能够基于目标小区及其主要邻区的负荷情况获取针对该小区的个性化的负荷分配策略，有效提高优化用户感知的准确度，提升用户感知优化的效果。进而能够根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数，能够最终实现提升目标小区的用户感知。且本申请提供的无线网络用户感知优化方法适用于无线网络中的各小区，因而无需针对各小区不断调整网络优化的方式，能够有效提升优化用户感知的效率。
附图说明
一个或多个实施方式通过与之对应的附图中的图片进行示例性说明，这些示例性说明并不构成对实施方式的限定，附图中具有相同参考数字标号的元件表示为类似的元件，除非有特别申明，附图中的图不构成比例限制。
图1是根据本申请一实施方式中的无线网络用户感知的优化方法流程图；
图2是根据本申请一实施方式中的小区A的容量指标和用户感知指标关联关系曲线图；
图3是根据本申请一实施方式中的小区B的容量指标和用户感知指标关联关系曲线图；
图4是根据本申请一实施方式中的小区C的容量指标和用户感知指标关联关系曲线图；
图5是根据本申请一实施方式中的目标小区和可迁入小区的负荷分配示意图；
图6是根据本申请一实施方式中的小区甲的容量指标和用户感知指标关联关系曲线图；
图7是根据本申请一实施方式中的小区乙的容量指标和用户感知指标关联关系曲线图；
图8是根据本申请一实施方式中的小区丙的容量指标和用户感知指标关联关系曲线图；
图9是根据本申请一实施方式中的无线网络用户感知的优化装置的结构示意图；
图10是根据本申请另一实施方式中的电子设备的结构示意图。
具体实施方式
为使本申请实施例的目的、技术方案和优点更加清楚，下面将结合附图对本申请的各实施方式进行详细的阐述。然而，本领域的普通技术人员可以理解，在本申请各实施方式中，为了使读者更好地理解本申请而提出了许多技术细节。但是，即使没有这些技术细节和基于以下各实施方式的种种变化和修改，也可以实现本申请所要求保护的技术方案。
本申请的一实施方式涉及一种无线网络用户感知的优化方法。值得说明的是，本实施方式的执行主体是计算机等能够实施上述无线网络用户感知的优化方法的设备，此处不进行具体限制。
在本实施方式中，获取容量指标和用户感知指标之间的关系；根据所述容量指标和用户感知指标之间的关系，确定待优化的目标小区；基于所述目标小区与所述目标小区的主要邻 区的负荷情况，获取所述目标小区的负荷分配策略；根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数。
下面对本实施例中的无线网络用户感知的优化方法的实现细节进行具体的说明，以下内容仅为方便理解本方案的实现细节，并非实施本方案的必须。具体流程如图1所示，可包括如下步骤：
步骤101，获取容量指标和用户感知指标之间的关系。
本步骤中涉及的容量指标一般可以是小区的用户数，用户感知指标可以包括用户上行速率、用户下行速率以及用户时延等能够反映用户感知的指标。
本步骤中获取的容量指标和用户感知指标之间的关系的过程可以包括容量指标和用户感知指标的关联关系、相关系数以及关系曲线的斜率等。
在一个例子中，在所述获取容量指标和用户感知指标之间的关系之前，还可以包括：获取历史容量指标和历史用户感知指标；在本例中，所述获取容量指标和用户感知指标之间的关系，包括：根据所述历史容量指标与所述历史用户感知指标，确定所述容量指标和用户感知指标之间的关系。利用小区的历史数据为小区确定容量指标和用户感知指标之间的关系，能够提高小区容量与用户感知的关系的可信度，进而能够提高网络感知优化的准确度。
步骤102，根据容量指标和用户感知指标之间的关系，确定待优化的目标小区。
可以确定容量指标和用户感知指标之间的关系为用户感知指标随着容量指标的增加而变差的小区，且其当前的容量指标(即用户数)大于目标容量指标(即目标用户数)、其当前的用户感知指标劣于目标用户感知指标的小区为目标小区。值得说明的是，目标用户数和目标用户感知指标可以根据该小区的优化场景针对性地确定。
用户感知指标随着容量指标的增加而变差，说明该小区在容量指标下降时用户感知指标能够更佳，且该小区当前的容量指标(即用户数)大于目标容量指标(即目标用户数)说明该小区需要进行用户迁出。此外，当前的用户感知指标劣于目标用户感知指标，说明该小区的用户感知指标不足，需要进行优化提升。
在一个例子中，在所述确定待优化的目标小区之后，还可以包括：根据所述容量指标和用户感知指标之间的关系，确定可迁入小区；在所述可迁入小区中，确定与所述目标小区的历史切换次数大于预设值的小区作为所述目标小区的所述主要邻区。在本例中，确定与目标小区之间切换次数较多的邻区作为为目标小区分担负荷的主要邻区。
本例中涉及的可迁入小区可以包括以下两类：
一类可迁入小区：关联关系单调递减且皮尔逊相关系数θ的绝对值大于预设值k的小区(说明对于该小区用户感知指标随着用户数的增加而变差)，且该小区的当前用户感知指标优于目标感知指标而当前用户数小于目标用户数。
二类可迁入小区：关联关系单调递减(说明对于该小区用户感知指标随着用户数的增加而变优)，且当前用户感知指标y与目标感知指标Y存在y>Y*(1+δ)的关系(说明该小区的用户感知指标较大程度地高于目标用户感知指标)的小区。值得说明的是，上述涉及的k，δ均为超参数。
步骤103，基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目标小区的负荷分配策略。
在一个例子中，基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目 标小区的负荷分配策略，可以包括：根据所述目标小区的当前的负荷数以及目标负荷数，确定所述目标小区需要迁出的负荷数；根据所述主要邻区的当前的负荷数以及目标负荷数，确定所述主要邻区可迁入的负荷数；其中，所述目标小区和所述主要邻区的所述目标负荷数根据所述优化需求确定；基于所述需要迁出的负荷数以及所述可迁入的负荷数，获取所述目标小区的负荷分配策略。
在本例中，基于所述需要迁出的负荷数以及所述可迁入的负荷数，获取所述目标小区的负荷分配策略，能够使得负荷分配策略符合目标小区与主要邻区的负荷情况，使得后续的优化调整更具有针对性，更加适应目标小区的优化需求，即进行更为准确的网络优化。
本例中确定目标小区需要迁出的负荷数的过程可以是：确定目标小区需要卸载的负荷(即需要迁出的用户数)为L out＝x-X+α。其中，目标小区的当前的负荷数为x，目标负荷数为X。值得说明的是，若L out>0，则此时确定的目标小区最终进行调整优化。
本例中确定主要邻区可迁入的负荷数的过程可以是：对于前述的一类可迁入小区，可迁入的负荷数为L in＝X-x-α。其中，主要邻区的当前的负荷数为x，目标负荷数为X。值得说明的是，若L in>0，则此时确定的可迁入小区为最终需要接收迁入的负荷的小区。此外，对于二类可迁入小区，可直接确定其为最终需要接收负荷的小区，确定其可迁入的负荷数为L in＝C。
值得说明的是，上述涉及的α，C均为超参数。
值得一提的是，本实施方式中涉及的负荷分配策略可以根据以下任一算法确定：贪心算法、遗传算法或者模拟退火算法。此处值得说明的是，本实施方式中用于确定负荷分配策略的算法并不局限于上述三种算法。利用其他适用的寻优算法也能够确定上述负荷分配策略，本实施方式中并未对此进行具体限制。
步骤104，根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数。
在一个例子中，根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数，可以包括：根据所述负荷分配策略，确定所述目标小区和所述主要邻区各自的迁移的负荷数；将所述迁移的负荷数以及预先采集的网络配置参数作为训练数据，训练得到迁移的负荷数和网络配置参数的映射关系模型；根据所述映射关系模型确定所述目标小区和所述主要邻区各自的目标网络配置参数，根据所述目标网络配置参数调整所述目标小区与所述主要邻区的网络配置参数。在本例中，利用迁移的负荷数和网络配置参数的映射关系模型，将确定的目标小区与主要邻区之间的负荷迁移数转化为方便调整的网络配置参数，能够降低用户感知优化的难度。
值得一提的是，此处作为训练数据的迁移的负荷数以及网络配置参数一一对应，且迁移的负荷数和与其对应的网络配置参数形成一组数据组，作为训练数据用于对模型进行训练。此外，迁移的负荷数和与其对应的网络配置参数可以是相关技术人员在日常工作中采集的数据。
本例中涉及的迁移的负荷数和网络配置参数的映射关系模型可以基于以下任一模型训练得到：线性回归模型、树模型或者深度神经网络模型。值得说明的是，此处训练模型并不局限于以上三种模型，其他适用的模型也可用于确定映射关系模型，本实施方式中并未对此进行具体限制。
无线网络技术领域还存在一种网络优化方式是使用SON(Self-Organized Network，自 组织网络)工具中的一些功能用例，自动迭代地调整无线网络中的参数以解决无线网络的问题。进而间接提升用户感知。通过传统的SON中的负荷均衡功能，通过设置的一定固定规则，识别出高负荷小区，小步迭代修改参数，能在一定程度上解决该问题。然而各小区的覆盖场景、用户分布、参数配置差异较大，统一的调整标准很难个性化地针对不同小区进行网络优化，进而难以针对性的提升用户感知。
传统网络优化方式中只能凭借小区当前负荷状况，判断该小区是否需要卸载负荷，且难以确定不同的主邻小区之间转移用户数量，很难保证当前调整幅度能达到目标或者使得某些小区的感知更差，而且中间步骤常需要人为判断是否合理。本实施方式通过分析用户感知指标和小区的容量指标之间的拟合曲线，高效地筛选分类，确定哪些小区转移出负荷后用户感知指标能得到有效改善，以及哪些小区能够接收迁入的用户且用户感知指标不会恶化到阈值以下。此外可以根据各小区的负荷情况，调整主邻小区对之间的负荷到一个最合适的范围，最大程度地使容量问题小区达到感知优化指标目标值。最终利用已训练好的模型，计算出网管侧需调整的每个小区之间网络配置参数。上述网络优化流程是端到端的，中间不需要任何人为判断。
上述网络优化过程迅速决策区域级的容量优化策略，输出优化参数，能够减少对专家经验的依赖。在一个例子中，1小时内能够完成6000小区的数据分析和优化方案的输出。相较于人工优化，显著提升效率。在另外的例子中，本实施方式提供的无线网络用户感知的优化方法能有效降低低用户感知小区的占比，具体降低比率可达3％～9％不等。
在本实施方式中，首先获取容量指标和用户感知指标之间的关系，能够获知用户感知与小区容量的关系，进而能够通过调整小区容量实现优化用户感知。在一实施方式中，根据所述容量指标和用户感知指标之间的关系，确定待优化的目标小区，能够确定需要进行用户感知优化的目标小区。再基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目标小区的负荷分配策略。对于需进行优化的目标小区，能够基于目标小区及其主要邻区的负荷情况获取针对该小区的个性化的负荷分配策略，有效提高优化用户感知的准确度，提升用户感知优化的效果。进而能够根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数，能够最终实现提升目标小区的用户感知。且本申请提供的无线网络用户感知优化方法适用于无线网络中的各小区，因而无需针对各小区不断调整网络优化的方式，能够有效提升优化用户感知的效率。
本申请的另一实施方式涉及一种无线网络用户感知的优化方法。值得说明的是，本实施方式与上一实施方式的主要区别在于：在本实施方式中，获取的容量指标和用户感知指标之间的关系具体为用户数和用户下行速率的关系。
下面对本实施例中的无线网络用户感知的优化方法的实现细节进行具体的说明，以下内容仅为方便理解本方案的实现细节，并非实施本方案的必须。
首先取N天历史数据，具体可以包括：历史用户数和历史用户下行速率指标。进而对各小区的历史用户数和历史用户下行速率指标进行高斯拟合，确定每个小区的用户数和用户下行速率指标的关联关系(即上一实施方式中涉及的容量指标和用户感知指标之间的关联关系)和皮尔逊相关系数θ。
在一实施方式中，基于该关联关系和相关系数θ，利用目标下行速率指标及其对应的目标用户数(即上一实施方式中涉及的目标负荷数)和当前下行速率指标及其对应的用户数之 间的关系，分别选出需迁出用户的目标小区和可迁入小区。值得说明的是，此处涉及的目标下行速率指标和目标用户数可以根据当前的优化需求个性化地确定。
小区A对应的用户数和用户下行速率指标的关联关系图如图2所示，对于小区A，其目标下行速率指标和目标用户数分别被确定为Y ms和X，如图2中的坐标(X,Y)所示；其当前下行速率指标取用户前N天平均下行速率指标y ms、当前用户数被确定为当前下行速率指标对应的用户数x，如图2中的坐标(x,y)所示。
由图2可见，小区A的关联关系单调递减且皮尔逊相关系数θ的绝对值大于预设值k，说明对于小区甲用户下行速率指标随着用户数的增加而下降且二者为强相关的关系。且当前用户下行速率指标低于目标下行速率指标，当前用户数大于目标用户数，说明小区A需要进行用户感知优化、迁出用户，因此确定小区A为需要迁出用户的目标小区。
小区B对应的用户数和用户下行速率指标的关联关系图如图3所示，对于小区B，其目标下行速率指标和目标用户数分别被确定为Y ms和X，如图3中的坐标(X,Y)所示。其当前下行速率指标取用户前N天平均下行速率指标y ms、当前用户数被确定为当前下行速率指标对应的用户数x，如图3中的坐标(x,y)所示。
由图3可见，小区B的关联关系单调递减且皮尔逊相关系数θ的绝对值大于预设值k，说明对于小区B用户下行速率指标随着用户数的增加而下降。且小区B的当前用户下行速率指标高于目标下行速率指标而当前用户数小于目标用户数，说明小区B能够接收较少迁入的用户，因此确定小区B为可迁入小区，且具体被确定为一类可迁入小区。
小区C对应的用户数和用户下行速率指标的关联关系图如图4所示，对于小区C，其目标下行速率指标和目标用户数分别被确定为Y ms和X，如图4中的坐标(X,Y)所示。其当前下行速率指标取用户前N天平均下行速率指标y ms、当前用户数被确定为当前下行速率指标对应的用户数x，如图4中的坐标(x,y)所示。
由图4可见，小区C的关联关系单调递减，且小区C的当前用户下行速率指标y与目标下行速率指标Y存在y>Y*(1+δ)的关系，说明小区C能够接收较多的迁入的用户，因此确定小区C为可迁入小区，且具体被确定为二类可迁入小区。
值得说明的是，上述涉及的k，δ均为超参数。
在一实施方式中，确定目标小区与所述目标小区的主要邻区的负荷情况，具体包括：目标小区需要迁出的负荷数以及可迁入小区可迁入的负荷数。
首先在通过前述过程中确定的可迁入小区中，对于与目标小区的历史切换次数确定目标小区的主要邻区。具体地，与目标小区之间切换次数较高(可以是高于一预设数值)的小区为主要邻区。
根据所述目标小区的当前的负荷数x以及目标负荷数X，确定所述目标小区需要迁出的负荷数，具体确定目标小区需要卸载的负荷(即需要迁出的用户数)为L out＝x-X+α。值得说明的是，若L out>0，则此时确定的目标小区为最终需要进行调整优化的目标小区。
根据所述主要邻区的当前的负荷数x以及目标负荷数X，确定所述主要邻区可迁入的负荷数。具体地，确定如小区B的一类可迁入小区的可迁入的负荷数为L in＝X-x-α。值得说明的是，若L in>0，则此时确定的可迁入小区为最终需要接收迁入的负荷的小区。对于如小区C的二类迁入小区，可直接确定其为最终需要接收负荷的小区。此外，确定其可迁入的负荷数为L in＝C。
值得说明的是，上述涉及的α，C均为超参数。
在一实施方式中，基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目标小区的负荷分配策略。负荷分配策略的原则是最大限度的利用邻区关系。在本实施方式中，为了使更多目标小区能尽量把用户全部迁移至符合条件的邻区，采用类贪心算法，按照以下顺序进行分配(以下以图5中示出的各小区的负荷情况为例进行说明，值得说明的是图5中示出左侧各小区为目标小区，右侧示出的各小区为可迁入小区，各数值为目标小区需要迁出的负荷数以及可迁入小区可迁入的负荷数)：
1)首先挑选某些邻区唯一出现的小区对，先把这些邻区的迁入用户分配完，如图5中示出的目标小区Y和其主要邻区z，邻区z可以接收4.3个用户，目标小区Y需要迁出3.6个用户，可以将目标小区Y用户全部分配给z小区；
2)其次挑选某些邻区不唯一出现的小区对，如果某目标小区的所有主要邻区的可迁入用户数之和大于需迁出用户数，则先对该目标小区的需迁出用户进行分配；如图5中目标小区A的需迁出用户数为10.7，其主要邻区a,b,c的可迁入用户之和为6.8+5.2+3.4＝15.4，大于10.7,那么优先分配目标小区A的需迁出用户。
3)最后分配邻区可迁入用户数之和小于需迁出用户数的小区，如图5中目标小区Z和其主要邻区x和y，主要邻区x,y接收用户之和为2.0+1.7<9.5，将如小区Z的这类小区的需迁出用户最后进行分配。
此外，值得一提的是，每轮确定主要邻区并分配迁出用户的过程，还可选择是优先选取与目标小区异频的邻区。
此外，值得说明的是，每次从未循环到的目标小区开始进行分配，当前目标小区对应的所有主要邻区一共可以迁入多少用户。然后根据目标小区的需迁出的用户数，按照各主要邻区的可迁入用户数的比例分配。进而更新迁入迁出后的各小区可迁移用户数量，重新循环。在完成一轮目标小区的用户分配之后，重新检查是否还有可以继续接收迁入用户的邻区。
在一实施方式中，根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数。具体可以选择经验模型或是线性回归模型、树模型以及DNN(Deep Neural Network，深度神经网络)等算法模型，训练得到迁移的负荷数和网络配置参数之间的映射关系模型，得到每个小区所需调整的网络配置参数，进而在目标小区和其主要邻区之间对接网管完成网络配置参数的修改。
其中，网络配置参数具体为CIO(Cell Individual Offset，小区级个体偏移)参数。
值得说明的是，上一实施方式中的技术细节在本实施方式中依然适用。为减少重复，本实施方式中不再赘述。
本申请的另一实施方式涉及一种无线网络用户感知的优化方法。值得说明的是，本实施方式与上一实施方式的主要区别在于：在本实施方式中，获取的容量指标和用户感知指标之间的关系具体为用户数和用户时延的关系。
下面对本实施例中的无线网络用户感知的优化方法的实现细节进行具体的说明，以下内容仅为方便理解本方案的实现细节，并非实施本方案的必须。
首先取N天历史数据，具体可以包括：历史用户数和历史用户时延。进而对各小区的用户数和用户时延进行线性拟合，确定每个小区的用户数和用户时延的关联关系(即前述涉及的容量指标和用户感知指标之间的关系)和皮尔逊相关系数θ。
在一实施方式中，基于该关联关系和相关系数θ，利用目标时延及其对应的目标用户数(即前述内容中涉及的目标负荷数)和当前时延及其对应的用户数之间的关系，分别选出需迁出用户的目标小区和可迁入小区。值得说明的是，此处涉及的目标时延和目标用户数可以根据当前的优化需求个性化地确定。
小区甲对应的用户数和用户时延的关联关系图如图6所示，对于小区甲，其目标时延和目标用户数分别被确定为Y ms和X，如图6中的坐标(X,Y)所示；其当前时延取用户前N天平均时延y ms、当前用户数被确定为当前时延对应的用户数x，如图6中的坐标(x,y)所示。
由图6可见，小区甲的关联关系单调递增且皮尔逊相关系数θ的绝对值大于预设值k，说明对于小区甲用户时延随着用户数的增加而增加且二者为强相关的关系。且当前用户时延高于目标时延，当前用户数大于目标用户数，说明小区甲需要进行用户感知优化、迁出用户，因此确定小区甲为需要迁出用户的目标小区。
小区乙对应的用户数和用户时延的关联关系图如图7所示，对于小区乙，其目标时延和目标用户数分别被确定为Y ms和X，如图7中的坐标(X,Y)所示。其当前时延取用户前N天平均时延y ms、当前用户数被确定为当前时延对应的用户数x，如图7中的坐标(x,y)所示。
由图7可见，小区乙的关联关系单调递增且皮尔逊相关系数θ的绝对值大于预设值k，说明对于小区乙用户时延随着用户数的增加而增加。且小区乙的当前用户时延低于目标时延且当前用户数小于目标用户数，说明小区乙能够接收较少迁入的用户，因此确定小区乙为可迁入小区，且具体被确定为一类可迁入小区。
小区丙对应的用户数和用户时延的关联关系图如图8所示，对于小区丙，其目标时延和目标用户数分别被确定为Y ms和X，如图8中的坐标(X,Y)所示。其当前时延取用户前N天平均时延y ms、当前用户数被确定为当前时延对应的用户数x，如图8中的坐标(x,y)所示。
由图8可见，小区丙的关联关系单调递减，且小区丙的当前用户时延y与目标时延Y存在y*(1+δ)<Y的关系，说明小区丙能够接收较多的迁入的用户，因此确定小区丙为可迁入小区，且具体被确定为二类可迁入小区。
值得说明的是，上述涉及的k，δ均为超参数。
在一实施方式中，确定目标小区与所述目标小区的主要邻区的负荷情况，具体包括：目标小区需要迁出的负荷数以及可迁入小区可迁入的负荷数。
首先在通过前述过程中确定的可迁入小区中，对于与目标小区的历史切换次数确定目标小区的主要邻区。具体地，与目标小区之间切换次数较高(可以是高于一预设数值)的小区为主要邻区。
根据所述目标小区的当前的负荷数x以及目标负荷数X，确定所述目标小区需要迁出的负荷数，具体确定目标小区需要卸载的负荷(即需要迁出的用户数)为L out＝x-X+α。值得说明的是，若L out>0，则此时确定的目标小区为最终需要进行调整优化的目标小区。
根据所述主要邻区的当前的负荷数x以及目标负荷数X，确定所述主要邻区可迁入的负荷数。具体地，确定如小区B的一类可迁入小区的可迁入的负荷数为L in＝X-x-α。值得说明的是，若L in>0，则此时确定的可迁入小区为最终需要接收迁入的负荷的小区。对于如小区C的二类迁入小区，可直接确定其为最终需要接收负荷的小区。此外，确定其可迁入的负荷数为L in＝C。
值得说明的是，上述涉及的α，C均为超参数。
在一实施方式中，基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目标小区的负荷分配策略。负荷分配策略的原则是最大限度的利用邻区关系。
此外，本实施方式与上一实施方式的另一主要区别在于：在本实施方式中，为了使更多目标小区能尽量把用户全部迁移至符合条件的邻区，采用遗传算法，按照以下步骤建模，寻找最优分配方案：
1)先对筛选的目标小区与其主要邻区组成的小区对，按照需迁出的负荷数与可迁入负荷数的比值从小到大排序，值越小越早分配。
2)种群初始化：种群大小可设定为50～500，按照个体定义，基因＝每个小区对的迁移用户数，个体＝所有基因的集合。由于迁移用户数基本都不是整数，所以这里选用浮点数编码。为了避免初始化时太多主要邻区的可迁入用户超出可容纳范围，对每个主要邻区及其邻区分成一组，先利用dirichlet分布，随机设定0～1概率值，分配给每个分组后的小区对，概率总和为1，随机用户数＝随机概率*可迁入用户数，初始小区对迁移用户数＝min(小区对间最大可迁移用户数，主区需迁出用户数，随机用户数)。对所有个体都执行上述操作后，生成初始种群。
3)适应度计算：计算目标小区和其主要邻区的剩余需迁出用户数和可迁入用户数，适应度函数定义为：fitness＝out_cell_reward+in_cell_reward-in_cell_penalty-out_cell_penalty，reward和penalty分别代表奖励和惩罚，当剩余需迁出用户数和可迁入用户数在设定范围内，给予奖励或惩罚：
目标小区剩余用户数，out_cell_reward＝1-i，i<＝0；out_cell_penalty＝-np.sqrt(i)，i>0；
主要邻区剩余用户数，in_cell_penalty＝-i，i>＝0；in_cell_penalty＝i，i<0；
对目标小区来说，剩余用户数i>0时，i越大惩罚越多；x<＝0时，x越小奖励越多；对其主要邻区来说，不管剩余用户数i如何，都给惩罚(只有i＝0时无惩罚)；然后计算所有个体适应度总和以及累积概率。
4)选择操作：遍历当前种群中的所有个体，如果随机值(0～1)小于某个个体的cum_prob，就把此个体添加到新种群中。
5)交叉操作：遍历所有个体，如果随机值(0～1)小于交叉阈值(一般设定为0.6～1)，就随机在当前个体上产生交叉位置cross_point，只截取交叉位置前的基因，再随机选择另一个个体，得到cross_point后的基因，拼接起来，然后把新个体添加到新种群中。
6)变异操作：遍历所有个体，如果随机值(0～1)小于变异阈值(一般设定为0.0001～0.1)，就随机产生变异位置mutate_point，由于每个小区对个体(分配用户值)都是由浮点数编码，变异时无法针对二进制取反，所以选择对随机挑选到的基因加减随机的高斯噪声，同时需要保证新个体的基因不为负(负值设置为0)。
7)选择当前种群中的最大适应度及其对应的个体，添加到最优种群中。
8)100次迭代进化后，终止算法，否则返回步骤3)。
最终从最优种群中选择最大/最小适应度对应的个体，作为最优负荷分配策略。
在一实施方式中，根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数。具体可以选择经验模型或是线性回归模型、树模型以及DNN(Deep Neural Network，深度神经网络)等算法模型，训练得到迁移的负荷数和网络配置参数之间的映射 关系模型，得到每个小区所需调整的网络配置参数，进而在目标小区和其主要邻区之间对接网管完成网络配置参数的修改。
其中，网络配置参数具体为CIO(Cell Individual Offset，小区级个体偏移)参数。
值得说明的是，前述实施方式中的技术细节在本实施方式中依然适用。为减少重复，本实施方式中不再赘述。
本申请的一实施方式涉及一种无线网络用户感知的优化装置，如图9所示，包括：
容量感知关系获取模块901，用于获取容量指标和用户感知指标之间的关系；
优化目标确定模块902，用于根据所述容量指标和用户感知指标之间的关系，确定待优化的目标小区；
分配策略获取模块903，用于基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目标小区的负荷分配策略；
调整优化模块904，用于根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数。
在一个例子中，分配策略获取模块903，还可以用于根据所述目标小区的当前的负荷数以及目标负荷数，确定所述目标小区需要迁出的负荷数；根据所述主要邻区的当前的负荷数以及目标负荷数，确定所述主要邻区可迁入的负荷数；其中，所述目标小区和所述主要邻区的所述目标负荷数根据所述优化需求确定；基于所述需要迁出的负荷数以及所述可迁入的负荷数，获取所述目标小区的负荷分配策略。
在一个例子中，调整优化模块904，还可以用于根据所述负荷分配策略，确定所述目标小区和所述主要邻区各自的迁移的负荷数；将所述迁移的负荷数以及预先采集的网络配置参数作为训练数据，训练得到迁移的负荷数和网络配置参数的映射关系模型；根据所述映射关系模型确定所述目标小区和所述主要邻区各自的目标网络配置参数，根据所述目标网络配置参数调整所述目标小区与所述主要邻区的网络配置参数。
在一个例子中，无线网络用户感知的优化装置还可以包括：历史数据获取模块(图中未示出)，用于在所述获取容量指标和用户感知指标之间的关系之前，获取历史容量指标和历史用户感知指标。在本例中，容量感知关系获取模块901，用于根据所述历史容量指标与所述历史用户感知指标，确定所述容量指标和用户感知指标之间的关系。
在一个例子中，无线网络用户感知的优化装置还可以包括：主要邻区获取模块(图中未示出)，用于在所述确定待优化的目标小区之后，根据所述容量指标和用户感知指标之间的关系，确定可迁入小区；在所述可迁入小区中，确定与所述目标小区的历史切换次数大于预设值的小区作为所述目标小区的所述主要邻区。
本实施方式提供的无线网络用户感知的优化装置首先获取容量指标和用户感知指标之间的关系，能够获知用户感知与小区容量的关系，进而能够通过调整小区容量实现优化用户感知。在一实施方式中，根据所述容量指标和用户感知指标之间的关系，确定待优化的目标小区，能够确定需要进行用户感知优化的目标小区。再基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目标小区的负荷分配策略。对于需进行优化的目标小区，能够基于目标小区及其主要邻区的负荷情况获取针对该小区的个性化的负荷分配策略，有效提高优化用户感知的准确度，提升用户感知优化的效果。进而能够根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数，能够最终实现提升目标小区的用户感知。且 本申请提供的无线网络用户感知优化方法适用于无线网络中的各小区，因而无需针对各小区不断调整网络优化的方式，能够有效提升优化用户感知的效率。
值得一提的是，本申请上述实施方式中所涉及到的各模块均为逻辑模块，在实际应用中，一个逻辑单元可以是一个物理单元，也可以是一个物理单元的一部分，还可以以多个物理单元的组合实现。此外，为了突出本申请的创新部分，本实施方式中并没有将与解决本申请所提出的技术问题关系不太密切的单元引入，但这并不表明本实施方式中不存在其它的单元。
本申请的实施例还提供一种电子设备，如图10所示，包括至少一个处理器1001；以及，与所述至少一个处理器1001通信连接的存储器1002；其中，存储器1002存储有可被至少一个处理器1001执行的指令，指令被至少一个处理器1001执行，以使至少一个处理器1001能够执行上述无线网络用户感知的优化方法。
其中，存储器1002和处理器1001采用总线方式连接，总线可以包括任意数量的互联的总线和桥，总线将一个或多个处理器1001和存储器1002的各种电路连接在一起。总线还可以将诸如外围设备、稳压器和功率管理电路等之类的各种其他电路连接在一起，这些都是本领域所公知的，因此，本文不再对其进行进一步描述。总线接口在总线和收发机之间提供接口。收发机可以是一个元件，也可以是多个元件，比如多个接收器和发送器，提供用于在传输介质上与各种其他装置通信的单元。经处理器1001处理的数据通过天线在无线介质上进行传输，进一步，天线还接收数据并将数据传送给处理器1001。
处理器1001负责管理总线和通常的处理，还可以提供各种功能，包括定时，外围接口，电压调节、电源管理以及其他控制功能。而存储器1002可以被用于存储处理器1001在执行操作时所使用的数据。
上述产品可执行本申请实施例所提供的方法，具备执行方法相应的功能模块和功能，未在本实施例中详尽描述的技术细节，可参见本申请实施例所提供的方法。
本申请的实施例还提供一种计算机可读存储介质，存储有计算机程序。计算机程序被处理器执行时实现上述无线网络用户感知的优化方法。
本领域技术人员可以理解，实现上述实施例方法中的全部或部分步骤是可以通过程序来指令相关的硬件来完成，该程序存储在一个存储介质中，包括若干指令用以使得一个设备(可以是单片机，芯片等)或处理器(processor)执行本申请各个实施例所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。
上述实施例是提供给本领域普通技术人员来实现和使用本申请的，本领域普通技术人员可以在不脱离本申请的发明思想的情况下，对上述实施例做出种种修改或变化，因而本申请的保护范围并不被上述实施例所限，而应该符合权利要求书所提到的创新性特征的最大范围。

Claims (10)

  1. 一种无线网络用户感知的优化方法，包括：
    获取容量指标和用户感知指标之间的关系；
    根据所述容量指标和用户感知指标之间的关系，确定待优化的目标小区；
    基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目标小区的负荷分配策略；
    根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数。
  2. 根据权利要求1所述的无线网络用户感知的优化方法，其中，所述基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目标小区的负荷分配策略，包括：
    根据所述目标小区的当前的负荷数以及目标负荷数，确定所述目标小区需要迁出的负荷数；
    根据所述主要邻区的当前的负荷数以及目标负荷数，确定所述主要邻区可迁入的负荷数；其中，所述目标小区和所述主要邻区的所述目标负荷数根据所述优化需求确定；
    基于所述需要迁出的负荷数以及所述可迁入的负荷数，获取所述目标小区的负荷分配策略。
  3. 根据权利要求1所述的无线网络用户感知的优化方法，其中，所述根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数，包括：
    根据所述负荷分配策略，确定所述目标小区和所述主要邻区各自的迁移的负荷数；
    将所述迁移的负荷数以及预先采集的网络配置参数作为训练数据，训练得到迁移的负荷数和网络配置参数的映射关系模型；
    根据所述映射关系模型确定所述目标小区和所述主要邻区各自的目标网络配置参数，根据所述目标网络配置参数调整所述目标小区与所述主要邻区的网络配置参数。
  4. 根据权利要求3所述的无线网络用户感知的优化方法，其中，所述迁移的负荷数和网络配置参数的映射关系模型基于以下任一模型训练得到：线性回归模型、树模型或者深度神经网络模型。
  5. 根据权利要求1所述的无线网络用户感知的优化方法，其中，在所述获取容量指标和用户感知指标之间的关系之前，还包括：
    获取历史容量指标和历史用户感知指标；
    所述获取容量指标和用户感知指标之间的关系，包括：
    根据所述历史容量指标与所述历史用户感知指标，确定所述容量指标和用户感知指标之间的关系。
  6. 根据权利要求1至5中任一项所述的无线网络用户感知的优化方法，其中，在所述确定待优化的目标小区之后，还包括：
    根据所述容量指标和用户感知指标之间的关系，确定可迁入小区；
    在所述可迁入小区中，确定与所述目标小区的历史切换次数大于预设值的小区作为所述目标小区的所述主要邻区。
  7. 根据权利要求1至5中任一项所述的无线网络用户感知的优化方法，其中，所述负荷分配策略根据以下任一算法确定：贪心算法、遗传算法或者模拟退火算法。
  8. 一种无线网络用户感知的优化装置，包括：
    容量感知关系获取模块，用于获取容量指标和用户感知指标之间的关系；
    优化目标确定模块，用于根据所述容量指标和用户感知指标之间的关系，确定待优化的目标小区；
    分配策略获取模块，用于基于所述目标小区与所述目标小区的主要邻区的负荷情况，获取所述目标小区的负荷分配策略；
    调整优化模块，用于根据所述负荷分配策略，调整所述目标小区与所述主要邻区的网络配置参数。
  9. 一种电子设备，包括：
    至少一个处理器；以及，
    与所述至少一个处理器通信连接的存储器；其中，
    所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行如权利要求1至7中任一项所述的无线网络用户感知的优化方法。
  10. 一种计算机可读存储介质，存储有计算机程序，其中，所述计算机程序被处理器执行时实现权利要求1至7中任一项所述的无线网络用户感知的优化方法。
PCT/CN2022/130435 2022-04-13 2022-11-07 无线网络用户感知优化方法、装置、电子设备和存储介质 WO2023197585A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210388264.3A CN116980917A (zh) 2022-04-13 2022-04-13 无线网络用户感知优化方法、装置、电子设备和存储介质
CN202210388264.3 2022-04-13

Publications (1)

Publication Number Publication Date
WO2023197585A1 true WO2023197585A1 (zh) 2023-10-19

Family

ID=88328879

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/130435 WO2023197585A1 (zh) 2022-04-13 2022-11-07 无线网络用户感知优化方法、装置、电子设备和存储介质

Country Status (2)

Country Link
CN (1) CN116980917A (zh)
WO (1) WO2023197585A1 (zh)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102065478A (zh) * 2009-11-13 2011-05-18 大唐移动通信设备有限公司 网络负荷调整目标自优化方法和装置
US20120005841A1 (en) * 2005-08-16 2012-01-12 Whirlpool Corporation Method of detecting an off-balance condition of a clothes load in a washing machine
CN109391486A (zh) * 2017-08-07 2019-02-26 中国移动通信集团浙江有限公司 基于用户体验的对互操作策略调整的预警方法及服务器
CN110505650A (zh) * 2018-05-16 2019-11-26 中国移动通信集团广东有限公司 随机异构分层网容量智能评估方法及装置
CN113259954A (zh) * 2020-02-13 2021-08-13 中国移动通信集团广东有限公司 一种质差处理策略的确定方法、装置及电子设备

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120005841A1 (en) * 2005-08-16 2012-01-12 Whirlpool Corporation Method of detecting an off-balance condition of a clothes load in a washing machine
CN102065478A (zh) * 2009-11-13 2011-05-18 大唐移动通信设备有限公司 网络负荷调整目标自优化方法和装置
CN109391486A (zh) * 2017-08-07 2019-02-26 中国移动通信集团浙江有限公司 基于用户体验的对互操作策略调整的预警方法及服务器
CN110505650A (zh) * 2018-05-16 2019-11-26 中国移动通信集团广东有限公司 随机异构分层网容量智能评估方法及装置
CN113259954A (zh) * 2020-02-13 2021-08-13 中国移动通信集团广东有限公司 一种质差处理策略的确定方法、装置及电子设备

Also Published As

Publication number Publication date
CN116980917A (zh) 2023-10-31

Similar Documents

Publication Publication Date Title
WO2021169577A1 (zh) 一种基于加权联邦学习的无线业务流量预测方法
CN110786046B (zh) 使用深度学习优化蜂窝网络
CN114595049A (zh) 一种云边协同任务调度方法及装置
CN111510319B (zh) 一种基于状态感知的网络切片资源管理方法
CN110996365B (zh) 一种基于多目标优化模型的异构网络垂直切换算法及系统
Lei et al. Learning-based resource allocation: Efficient content delivery enabled by convolutional neural network
CN112329997A (zh) 电力需求负荷预测方法及系统、电子设备和存储介质
CN113676357B (zh) 面向电力物联网中边缘数据处理的决策方法及其应用
WO2023197585A1 (zh) 无线网络用户感知优化方法、装置、电子设备和存储介质
WO2023011371A1 (en) Method and system for configuring a threshold value for a handover parameter of a wireless communication system
Wang et al. A dynamic channel-borrowing approach with fuzzy logic control in distributed cellular networks
CN111010695A (zh) 基于信道空闲时长预测的信道分配方法
Hu et al. Device scheduling and channel allocation for energy-efficient Federated Edge Learning
CN110245019A (zh) 一种自适应系统资源的线程并发方法及装置
WO2021208557A1 (zh) 参数调整的方法、服务器及存储介质
CN110224861A (zh) 基于学习的自适应动态异构网络选择策略的实现方法
CN115361453A (zh) 一种面向边缘服务网络的负载公平卸载与迁移方法
CN115119332A (zh) 意图驱动的无线网络资源冲突解决方法及其装置
CN103906197A (zh) 一种认知无线网络多接入选择的决策方法
CN115250156A (zh) 一种基于联邦学习的无线网络多信道频谱接入方法
CN113890653A (zh) 面向多用户利益的多智能体强化学习功率分配方法
Mollaei et al. New method for congestion control in wireless sensor network using neural network
CN105472641B (zh) 基于3D-Markov链的IEEE802.11ad网络用户QoE优化方法
Zhang et al. A novel heuristic for handover priority in mobile heterogeneous networks based on a multimodule Takagi–Sugeno–Kang fuzzy system
CN113473419B (zh) 基于强化学习的机器类通信设备接入蜂窝数据网络的方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22937217

Country of ref document: EP

Kind code of ref document: A1