Wireless network optimization method, network device and computer readable storage medium

Abstract

The embodiment of the present invention provides a wireless network optimization method, a network device and a computer-readable storage medium, wherein a performance optimization model is obtained by training according to historical performance indicators of cells in a wireless network to be optimized and historical configuration parameters corresponding to the historical performance indicators, and the performance optimization model is used as a basis for subsequent optimization adjustments; since the performance optimization model is obtained based on the real historical performance indicators of the wireless network and the historical configuration parameters corresponding to the historical performance indicators, it is more in line with the real operating environment of the wireless network, has better adaptability and intelligence, and the optimization adjustment made based on this is more accurate, so that the performance of the adjusted wireless network is better.

Description

Technical Field

[0001] The embodiments of the present invention relate to, but are not limited to, the field of communications, and specifically, relate to, but are not limited to, a wireless network optimization method, a network device, and a computer-readable storage medium. Background Art

[0002] In the related art, the performance optimization of wireless networks is mostly based on artificially set expert rules, relying on expert rules to formulate adjustment rules to perform corresponding parameter adjustment actions to achieve the purpose of wireless network performance optimization. For example, for the network capacity optimization of wireless networks, the configuration parameters are modified based on pre-established adjustment rules to achieve the purpose of adjusting parameters to optimize network capacity. The disadvantage of this type of method is that it completely relies on established rules. If the rules are not set well, it will directly affect the effect of wireless network performance optimization; in addition, when the existing network environment changes, the set rules may not be able to adapt to the changes, which further leads to poor optimization effects. Summary of the invention

[0003] The embodiments of the present invention provide a wireless network optimization method, a network device, and a computer-readable storage medium, which mainly solve the technical problem of: solving the problem of poor optimization effect and poor adaptability of wireless network performance in related technologies.

[0004] In order to solve the above technical problems, an embodiment of the present invention provides a wireless network optimization method, including:

[0005] Obtaining performance indicators of the target cell pair to be adjusted;

[0006] Using the performance indicator as an input of a performance optimization model, and inputting the performance optimization model to obtain configuration parameters of the target cell pair;

[0007] Optimizing and adjusting at least one target cell in the target cell pair according to the configuration parameters;

[0008] The configuration parameters are parameters that affect the performance indicators, and the performance optimization model is obtained by training based on historical performance indicators of cells in the wireless network and historical configuration parameters corresponding to the historical performance indicators.

[0009] An embodiment of the present invention further provides a network device, including a processor, a memory and a communication bus;

[0010] The communication bus is used to realize the connection and communication between the processor and the memory;

[0011] The processor is used to execute one or more computer programs stored in the memory to implement the steps of the wireless network optimization method as described above.

[0012] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores one or more computer programs, and the one or more computer programs can be executed by one or more processors to implement the steps of the wireless network optimization method as described above.

[0013] According to the wireless network optimization method, network device and computer-readable storage medium provided by the embodiments of the present invention, a performance optimization model is obtained by training according to the historical performance indicators of the cells in the wireless network to be optimized and the historical configuration parameters corresponding to the historical performance indicators; for the target cell pair to be adjusted, the performance indicators of the target cell pair can be obtained, and the obtained performance indicators are used as inputs of the performance optimization model, so as to obtain the configuration parameters of the target cell pair using the performance optimization model, and optimize and adjust at least one target cell in the target cell pair according to the obtained configuration parameters; the performance optimization model in the embodiments of the present invention is obtained by training according to the historical performance indicators of the cells in the wireless network and the historical configuration parameters corresponding to the historical performance indicators, and is no longer based on simple expert rules (i.e., empirical rules) as the adjustment basis, so that the accuracy of the optimization adjustment is better, the performance of the wireless network after the optimization adjustment is better, and the adaptability and intelligence are better.

[0014] Other features and corresponding beneficial effects of the present invention are described in the latter part of the specification, and it should be understood that at least part of the beneficial effects become obvious from the description in the specification of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of a wireless network optimization method flow chart provided in Embodiment 1 of the present invention;

[0016] Figure 2 A schematic diagram of a network area to be optimized provided in the first embodiment of the present invention;

[0017] Figure 3 A schematic diagram of the incremental training process of the performance optimization model provided in the first embodiment of the present invention;

[0018] Figure 4 A schematic diagram of the trigger threshold value flow of the A1 event and the A2 event provided in the first embodiment of the present invention;

[0019] Figure 5 A schematic diagram of a safe rollback control process provided in the first embodiment of the present invention;

[0020] Figure 6 A schematic diagram of the structure of a wireless network optimization device provided in Embodiment 2 of the present invention;

[0021] Figure 7 A schematic diagram of the structure of a wireless network system provided in Embodiment 2 of the present invention;

[0022] Figure 8 A schematic diagram of a flow chart of a wireless network optimization method in an application example 1 provided in embodiment 2 of the present invention;

[0023] Fig. 9 A schematic diagram of a flow chart of a wireless network optimization method in an application example 2 provided in embodiment 2 of the present invention;

[0024] Fig.10 A schematic diagram of a flow chart of a wireless network optimization method in application example 3 provided in embodiment 2 of the present invention;

[0025] Fig.11 A schematic diagram of the network device structure provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the following is a further detailed description of the embodiments of the present invention through specific implementation methods combined with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0027] Embodiment 1:

[0028] In order to improve the accuracy and adaptability of wireless network performance optimization, the present embodiment provides a new wireless network optimization method, which obtains a performance optimization model by training based on historical performance indicators of cells in the wireless network to be optimized and historical configuration parameters corresponding to the historical performance indicators, and uses the performance optimization model as the basis for optimization adjustment; because it is obtained based on the real historical performance indicators of the wireless network and the historical configuration parameters corresponding to the historical performance indicators, it is more in line with the real operating environment of the wireless network, has better adaptability and intelligence, and the optimization adjustment made based on this is more accurate, so that the performance of the adjusted wireless network is better, and the comprehensive performance of the wireless network can be improved.

[0029] For ease of understanding, this embodiment is described below in conjunction with Figure 1 The wireless network optimization method shown is for example only. Figure 1 As shown, it includes but is not limited to:

[0030] S101: Obtain performance indicators of a target cell pair to be adjusted.

[0031] It should be understood that the target cell pair in this step is a cell pair in the wireless network to be optimized, and the number of target cells included in a target cell pair may be two. Of course, according to the requirements of specific application scenarios, the number of target cells in a target cell pair may also be set to three or more.

[0032] The performance indicators of the target cell obtained in this embodiment can be flexibly set according to specific optimization requirements. For example, when it is necessary to optimize the network capacity of the wireless network, the performance indicators obtained include at least one of the various performance indicators used to characterize or determine the network capacity of the wireless network. Therefore, which performance indicators are specifically obtained in this embodiment can be flexibly set according to specific application requirements, and this embodiment does not limit it. For example, in an application example, the performance indicators of the target cell pair to be adjusted obtained may include but are not limited to the load difference between the target cells within the target cell pair. In this application example, at least one of the various performance indicators that can characterize the load of the target cell can be obtained, for example, the PRB (Physical Resource Block) utilization rate of the target cell, the number of activated users, and the number of RRC (Radio Resource Control) connections can be obtained, for example:

[0033] In one application scenario, the PRB usage rate of each target cell in the target cell pair may be obtained, and the difference of the PRB usage rates of each target cell may be used as the load difference;

[0034] In another application scenario, the number of activated users of each target cell in the target cell pair may be obtained, and the difference in the number of activated users of each target cell may be used as the load difference;

[0035] In another application scenario, the number of RRC connections of each target cell in the target cell pair may be obtained, and the difference in the number of RRC connections of each target cell may be used as the load difference;

[0036] In another application scenario, the difference between at least two of the PRB usage rate, the number of RRC connections, and the number of activated users of each target cell in the target cell pair can be obtained, and the two obtained differences can be normalized or processed by weight calculation as the load difference.

[0037] It should be understood that the load of the target cell is not limited to the several parameter characterization methods in the above examples, and other parameters that can characterize the cell load may also be used, which is not limited here.

[0038] Optionally, in order to more reasonably optimize the performance of the wireless network and avoid the ping-pong effect between cells during the optimization process, the wireless network can be divided into regions and optimized in units of regions. For example, in one example, the wireless network can be divided into at least one optimized network area, and the network area to be optimized is geographically continuous, and the boundary lines of the peripheral cells in the area form a convex figure area. The convex figure in this embodiment is a convex figure defined in a mathematical sense. In this embodiment, when dividing the wireless network into regions, a manual division method or an automatic division method can be used. For example, in one application scenario, the wireless network can be divided into regions by, but not limited to, calling the RCA (Root Cause Analysis) method. Please refer to Figure 2 As shown, it is assumed that the network area to be optimized obtained by dividing the wireless network into regions in this application scenario includes a network area A to be optimized and a network area B to be optimized, and the graph formed by the cells in the network area A to be optimized and the network area B to be optimized is a convex graph. Among them, the cell a and the cell f in the network area A to be optimized constitute a cell pair. In some application examples, the cells in the cell pair may be adjacent cells. In some examples of this embodiment, the target cell pair may be a cell pair in the network area to be optimized in the wireless network, and the performance indicators meet the preset optimization conditions.

[0039] For example, following the above example, when the performance indicator includes load difference, the target cell pair can be set to be a cell pair in the network area to be optimized in the wireless network, and the load difference satisfies the preset optimization condition. The preset optimization condition can be that the load difference is not within the preset load difference range. The preset load difference range can be flexibly set according to demand, for example, it can be set to 0, or it can be set to a specific range value. In this example, it is assumed that cell a and cell f in the network area A to be optimized constitute a cell pair, and cell a and cell b constitute a cell pair. It is assumed that the absolute value of the load difference between cell a and cell f is greater than 0, then cell a and cell f are determined to be the target cell pair; it is assumed that the absolute value of the load difference between cell a and cell b is equal to 0, then it is determined that cell a and cell b are not the target cell pair.

[0040] In this embodiment, the performance indicator of the target cell pair may be obtained manually or automatically through network management equipment or other communication equipment.

[0041] The historical performance indicators and historical configuration parameters in this step may be the historical performance indicators and historical configuration parameters of each cell in the entire wireless network, or the historical performance indicators and historical configuration parameters of cells in the network area to be optimized, and can be flexibly set according to specific needs.

[0042] S102: Using the acquired performance indicators as inputs to a performance optimization model, and inputting the performance optimization model to obtain configuration parameters of the target cell pair.

[0043] The performance optimization model in this embodiment is trained based on the historical performance indicators of the cells in the wireless network and the historical configuration parameters corresponding to the historical performance indicators; the historical performance indicators and historical configuration parameters in this embodiment can be obtained manually or automatically collected by network management equipment or other communication equipment.

[0044] The configuration parameters in this embodiment include various configuration parameters that can affect the performance indicators. For example, following the above example, when the performance indicator includes the load difference between the target cells in the target cell pair, the configuration parameters may include but not be less than the cell individual offset (Cell Ind Offset, CIO) parameter of the target cell in the target cell pair. The cell individual offset parameter can directly affect the switching threshold of the two cells in the cell pair and adjust the load of the two cells in the cell pair. Adjusting the cell individual offset parameter to different degrees can cause the load of the two cells in the cell pair to change to different degrees, so that the load of the two cells in the cell pair meets the requirements of performance optimization.

[0045] It should be understood that the performance optimization model in this embodiment can be at least one of various models that can achieve the above purpose. For ease of understanding, this embodiment is described below with several model examples. In this example, the performance optimization model can include but is not limited to at least one of the following models:

[0046] Fitted straight line model: A fitted straight line model is obtained by using a linear regression model, taking historical performance indicators as input and historical configuration parameters as output for regression fitting;

[0047] In this embodiment, the historical performance data used can be the performance data of each cell in the network area to be optimized where the target cell pair is located; specifically, it can be collected but not limited to the network management operation log, and the adjustment record of the relevant parameters can be obtained by parsing. Then, the various performance indicator data of the target cell corresponding to the adjustment record are obtained, including but not limited to the historical load difference data and other data, and the historical performance configuration data corresponding to each historical load, such as the cell individual offset parameter of the cell, and are sorted into (load difference data, cell individual offset parameter) binary data, and the fitted straight line model is obtained by regression fitting based on the integrated binary number. For example, a corresponding scatter plot can be drawn in a two-dimensional plane coordinate system, and the above binary data can be regressed and fitted using a linear regression algorithm, and the fitted straight line is the obtained fitted straight line model. For the target cell pair that needs to adjust the load difference, the load difference of the target cell pair is substituted into the fitted straight line model to obtain the cell individual offset parameters (such as the cell individual offset amplitude) of the target cell in the target cell pair that needs to be adjusted.

[0048] Fitting curve model: A fitting curve model is obtained by using a polynomial fitting model, taking historical performance indicators as input and historical configuration parameters as output for polynomial fitting.

[0049] The training process of the fitting curve model in this embodiment is similar to that of the fitting straight line model. The main difference is that a polynomial fitting model is used for fitting. After the corresponding scatter plot is drawn in the two-dimensional plane coordinate system, a polynomial fitting is performed on this basis to obtain a fitting curve as the fitting curve model.

[0050] Neural network model: A neural network model is obtained by using a neural network, calling a preset training algorithm, taking historical performance indicators as input, and historical configuration parameters as output for training.

[0051] In this embodiment, the historical performance data used can be the performance data of each cell in the network area to be optimized where the target cell pair is located; specifically, it can be collected but not limited to the network management operation log, and the adjustment record of the relevant parameters can be obtained by parsing. Then, the performance indicator data of the target cell corresponding to the adjustment record is obtained, including but not limited to the state of the cell pair (it can also include but not limited to the historical transmission power, bandwidth and other parameters of the cell), the historical load difference data of the cell pair, and the historical performance configuration data corresponding to each historical load, such as the individual cell offset parameter of the cell, and the above data are integrated into a training data set, and each data in the training data set includes the state of the cell pair, the historical load difference data of the cell pair, and the individual cell offset parameter. An artificial neural network can be initialized, for example, a network composed of but not limited to three linear layers can be used, and a preset training algorithm can be called (for example, an open source deep learning framework can be used but not limited to build and call BP (Back Propagation, back propagation algorithm) algorithm) to train the neural network using the above training data set to obtain a neural network model. The input of the neural network model is the state of the cell pair and the historical load difference data of the cell pair, and the output is the value of the individual cell offset parameter to be adjusted. In this embodiment, the data can be normalized when input into the neural network, and it needs to be denormalized when output. Normalization and denormalization remain consistent during model training and use.

[0052] It should be understood that the performance optimization model in this embodiment is not limited to the three models in the above examples, nor is it limited to using one of the three models in the above examples. Two or three of the three models in the above examples may also be used. For example:

[0053] In one application scenario, a fitted straight line model may be used to output a first configuration parameter, a neural network model may be used to output a third configuration parameter, and then the first configuration parameter and the third configuration parameter may be averaged or calculated using other weight algorithms to obtain the final configuration parameter.

[0054] In another application scenario, a first configuration parameter may be outputted using a fitted straight line model, and a second configuration parameter may be outputted using a fitted curve model. Then, the first configuration parameter and the second configuration parameter may be averaged or calculated using other weight algorithms to obtain the final configuration parameter.

[0055] In another application scenario, a fitting straight line model can be used to output the first configuration parameter, a fitting curve model can be used to output the second configuration parameter, and a neural network model can be used to output the third configuration parameter. The first configuration parameter, the second configuration parameter, and the third configuration parameter are then averaged or calculated using other weight algorithms to obtain the final configuration parameter.

[0056] S103: Optimize and adjust at least one target cell in the target cell pair according to the obtained configuration parameters.

[0057] In some examples of this embodiment, the obtained configuration parameters may be sent down by, but not limited to, pushing them to a network management device.

[0058] For example, when the performance indicator includes the load difference, the obtained cell individual offset parameter of the target cell in the target cell pair may be pushed to the network management device and sent to the base station corresponding to the target cell in the target cell pair for optimization and adjustment.

[0059] Optionally, in this embodiment, in order to further improve the adaptability and accuracy of the performance optimization model, after the performance optimization model is obtained, the obtained performance optimization model can also be dynamically updated. For an example of dynamic update, see Figure 3 As shown, it includes but is not limited to:

[0060] S301: Collecting performance indicators of cells in a wireless network and configuration parameters corresponding to the performance indicators.

[0061] In this embodiment, when the performance indicators of the cells in the wireless network and the configuration parameters corresponding to the performance indicators are collected, the collection can be performed regularly or randomly, or only when other set collection conditions are met (for example, a collection instruction sent by an administrator is received).

[0062] S302: Incrementally train the performance optimization model according to the collected performance indicators and configuration parameters, so as to update the performance optimization model.

[0063] In this example, the method of incrementally training the performance optimization model according to the collected performance indicators and configuration parameters can adopt but is not limited to the performance model training method in the above example, which will not be repeated here.

[0064] Optionally, in order to make the optimized and adjusted parameters more effective, in the above S102, the performance indicators are used as inputs of the performance optimization model. After the configuration parameters of the target cell pair are obtained by inputting the performance optimization model, the following may also be included but not limited to: Figure 4 The following process is shown:

[0065] S401: Acquire a first trigger threshold value of an A1 event and a second trigger threshold value of an A2 event of a target cell in a target cell pair, and an average RSRP (Reference Signal Receiving Power) value of the target cell.

[0066] S402: Adjust the first trigger threshold and the second trigger threshold according to the acquired average RSRP value.

[0067] In this embodiment, the target cell's A1 event, the serving cell is better than the absolute threshold (i.e., the first trigger threshold value), this event can be used to disable the measurement between certain cells. The A2 event, the serving cell is worse than the absolute threshold (i.e., the second trigger threshold value), this event can be used to enable the measurement between certain cells, because after this event occurs, operations such as handover may occur.

[0068] It should be understood that when the UE (User Equipment) does not start measurement, even if the configuration parameters of the target cell within the target cell pair (such as the cell individual offset parameter) are modified, it cannot be triggered. Therefore, in order to ensure that a reasonable number of UE measurement triggers are triggered, the first trigger threshold value and the second trigger threshold value can be adjusted according to the obtained average RSRP value to ensure that the optimized and adjusted parameters can take effect in a timely and normal manner.

[0069] For example, in one example, adjusting the first trigger threshold value and the second trigger threshold value according to the average reference signal received power value may include but is not limited to:

[0070] When the first trigger threshold value of the A1 event and the second trigger threshold value of the A2 event of the target cell in the target cell pair are both less than the average reference signal received power value (i.e., the trigger threshold values ​​of the A1 event and the A2 event are too low), and the target cell is a cell from which the load needs to be migrated, the second trigger threshold value and the first trigger threshold value are increased by the same magnitude based on the average reference signal received power value; for example, the second trigger threshold value may be adjusted to be the same as the average reference signal received power value, and the first trigger threshold value is correspondingly increased by the same magnitude;

[0071] When the first trigger threshold value of the A1 event and the second trigger threshold value of the A2 event of the target cell are both greater than the average reference signal received power value (i.e., the trigger threshold values ​​of the A1 event and the A2 event are too high), the second trigger threshold value and the first trigger threshold value are reduced by the same magnitude based on the average reference signal received power value; for example, the first trigger threshold value may be adjusted to be the same as the average reference signal received power value, and the second trigger threshold value may be reduced by the same magnitude accordingly;

[0072] When the first trigger threshold value of the A1 event of the target cell is greater than the average reference signal received power value, and the second trigger threshold value of the A2 event of the target cell is less than the average reference signal received power value, the second trigger threshold value and the first trigger threshold value are increased by the same amount based on the average reference signal received power value; for example, the second trigger threshold value can be adjusted to be the same as the average reference signal received power value, and the first trigger threshold value is increased by the same amount accordingly.

[0073] Optionally, in this embodiment, in order to further improve the reliability of the wireless network, during the process of optimizing and adjusting at least one target cell in the target cell pair according to the configuration parameters, or after optimizing and adjusting at least one target cell in the target cell pair according to the configuration parameters, it may also include: Figure 5 The protection process shown:

[0074] S501: Monitor performance indicators of cells in a wireless network.

[0075] In this step, the performance indicators of the cells in the wireless network are monitored. The performance indicators of all cells in the wireless network may be monitored, or the performance indicators of only the target cells after optimization adjustment may be monitored, or the performance indicators of at least one cell in the network area to be optimized may be monitored.

[0076] The performance indicators monitored in this embodiment may include optimized and adjusted performance indicators, such as cell load; other performance indicators may also be monitored as required, such as transmission power, bandwidth, interference conditions, etc.

[0077] S502: When it is monitored that the performance indicator reaches a preset deterioration condition, the configuration parameters of at least one target cell in the target cell pair are restored to the configuration parameters before the optimization adjustment.

[0078] It should be understood that the preset deterioration condition in this step can be flexibly set according to specific application scenarios and standards. When it is monitored that the performance indicator reaches the preset deterioration condition, the configuration parameters of the cell in the wireless network are restored to the configuration parameters before optimization, so as to ensure the normal operation of the wireless network and avoid additional continuous impact on the wireless network. Of course, in other examples of this embodiment, a set of default configuration parameters that can ensure its normal operation can also be pre-set for the wireless network, and when it is monitored that the performance indicator reaches the preset deterioration condition, the configuration parameters of the cell in the wireless network are set to the default configuration parameters.

[0079] The wireless network optimization method provided by the embodiment of the present invention obtains a performance optimization model by training according to the historical performance indicators of the cells in the wireless network to be optimized and the historical configuration parameters corresponding to the historical performance indicators, and uses the performance optimization model as the basis for subsequent optimization adjustment; because it is obtained based on the real historical performance indicators of the wireless network and the historical configuration parameters corresponding to the historical performance indicators, it is more in line with the real operating environment of the wireless network, improves the adaptability and intelligence of the wireless network optimization, and the optimization adjustment made based on the performance optimization model is more accurate, so that the performance of the adjusted wireless network is better.

[0080] Embodiment 2:

[0081] This embodiment provides a wireless network optimization device, which can be set on a network device (for example, but not limited to a server) to optimize the performance of the wireless network. Figure 6 As shown, it includes but is not limited to:

[0082] The data collection module 61 is used to collect the historical performance indicators of the cells in the wireless network and the historical configuration parameters corresponding to the historical performance indicators; the collection method can be to receive the historical performance indicators and the historical configuration parameters corresponding to the historical performance indicators manually input, or to automatically collect the historical performance indicators and the historical configuration parameters corresponding to the historical performance indicators through the network management device. The specific collection process is shown in the above example, which will not be repeated here. The data collection module 61 is also used to collect the performance indicators of the target cell pair to be adjusted. The specific collection process is shown in the above example, which will not be repeated here.

[0083] The model training module 62 is used to train a performance optimization model according to the historical performance indicators of the cells in the wireless network and the historical configuration parameters corresponding to the historical performance indicators. The specific training process is shown in the above embodiment and will not be repeated here.

[0084] The cell pair parameter generation module 63 is used to generate the configuration parameters of the target cell pair according to the performance indicators of the target cell pair to be adjusted and the performance optimization model. The specific generation process is shown in the above example and will not be repeated here.

[0085] The cell parameter generation module 64 is used to adjust the first trigger threshold value and the second trigger threshold value according to the first trigger threshold value of the A1 event and the second trigger threshold value of the A2 event of the target cell in the target cell pair, and the average reference signal received power value of the target cell. The specific adjustment process is shown in the above example and will not be repeated here.

[0086] The parameter sending module 65 is used to send the corresponding parameters generated by the cell parameter generating module 63 and the cell parameter generating module 64 to the network management device to optimize the configuration of the corresponding cell.

[0087] The security rollback module 66 is used to restore the configuration parameters of at least one target cell in the target cell pair to the configuration parameters before optimization and adjustment when the performance indicators of the cells in the wireless network collected by the data collection module 61 reach a preset deterioration condition; the specific security rollback process is shown in the above embodiment and will not be repeated here.

[0088] Optionally, the wireless network optimization device in this embodiment may further include a region division module 67 for performing region division on the wireless network to obtain a network region to be optimized. The specific division method is shown in the above examples and will not be described in detail here.

[0089] The adjacency relationship statistics module 68 is used to count the cell pairs in the network area to be optimized. The specific statistical method can be performed according to the configuration information and the corresponding adjacency relationship identification method, which will not be repeated here.

[0090] This embodiment provides a wireless network system. Figure 7 As shown, it includes a base station 701, a network management device 702 and a server 703 connected thereto. The server 703 may be provided with Figure 6 The wireless network optimization device shown in the figure can optimize the wireless network performance through the wireless network optimization method shown in the above examples. Figure 7 Based on the system architecture example shown, several wireless network optimization methods are exemplarily described.

[0091] Application example 1:

[0092] See also Figure 8 As shown, the wireless network optimization method in this application example includes but is not limited to:

[0093] S801: The data collection module obtains performance data, configuration parameters, switching data, parameter adjustment history logs and other data from the network management, and provides these data to modules in need in subsequent steps.

[0094] S802: The area division module uses the performance indicator data and the neighboring cell configuration data between cells to calculate the network area to be optimized.

[0095] Assume that one of the network areas to be optimized is recorded as area A. Area A is a geographically continuous area with a convex shape, including 200 cells. The load difference between neighboring cells in area A is large, there is a serious load imbalance, the network capacity is limited, and capacity optimization is needed.

[0096] S803: The adjacency statistics module obtains the handover data and configuration parameters from the data collection module, and obtains the target cell pairs existing in area A through statistics.

[0097] For example, it is calculated that there are 600 target cell pairs in area A. The two target cells in each target cell pair are configured with neighboring cell relations in the configuration parameters, and there are a certain number of handover records between them in the handover data.

[0098] S804: Before the optimization starts, the security rollback module obtains the network area to be optimized (ie, area A) from the area division module, and obtains the relevant configuration parameters of area A before the optimization starts from the data acquisition module and saves these parameters.

[0099] After the optimization starts, the safety rollback module obtains the performance index data of area A from the data acquisition module at regular intervals. When the performance index data of area A deteriorates beyond expectations, the safety rollback module can start the rollback process. The configuration parameters saved before the optimization are sent through the parameter sending module, and the cell stops working on the optimization parameter generation module and the cell parameter generation module, and the parameter sending module can be notified to stop sending the optimization parameters.

[0100] S805: Perform model training to obtain a performance optimization model.

[0101] Assume that this capacity optimization is used for the first time in the region, so there is no trained performance optimization model, and model training is required to obtain the performance optimization model. It should be understood that when it is not the first time to use, the previously trained performance optimization model can be directly called.

[0102] In this step, the model training module obtains the network management operation log from the data acquisition module, and parses it to obtain the adjustment record of the relevant parameters. Then the model training module obtains various performance indicators of the cell corresponding to the adjustment record from the data acquisition module, including data such as load change. In this embodiment, a linear regression model is used. The above data are organized into a binary group of (change value of the load difference of the cell pair, individual cell offset of the cell in the cell pair), and a scatter plot can be drawn in a two-dimensional plane coordinate system. The above data are regressed and fitted using a linear regression algorithm, and the straight line obtained by fitting is the obtained performance optimization model (i.e., the fitted straight line model). For the target cell pair that needs to adjust the load difference, the load difference of the target cell pair is substituted into the model to obtain the individual cell offset amplitude of the target cell that needs to be adjusted.

[0103] S806: The cell pair parameter generation module is used to generate the optimization parameters of the cell pair, that is, the cell individual offset of the target cell.

[0104] The cell pair parameter generation module can obtain the cell pair information from the adjacency statistics module and obtain the load difference between the two cells in the cell pair from the data acquisition module. The performance optimization model obtained by the above training is used to calculate the amplitude of the cell individual offset of the target cell of each target cell pair; and a cell pair parameter adjustment schedule containing the amplitude of the cell individual offset of each target cell is generated.

[0105] S807: The cell parameter generation module generates optimized parameters for the cell, which may include but is not limited to adjusting the trigger thresholds of the A1 and A2 events of the cell.

[0106] In this application example, the cell parameter generation module obtains the list of cell pairs to be adjusted from the adjacency statistics module, and obtains all relevant target cells; obtains the cell individual offset parameter adjustment plan of the target cell in each target cell pair from the cell pair parameter generation module, so as to obtain the direction and amplitude of the parameters of each target cell to be adjusted; obtains the performance indicator data such as the average RSRP of users of the corresponding target cell from the parameter acquisition module. In this embodiment, the cell parameter generation module adjusts the measurement event opening and closing threshold parameters of the corresponding target cell based on the above data, that is, the triggering threshold of the A1 and A2 events, and the adjustment method is as follows:

[0107] For the target cell where users need to be migrated, if the trigger thresholds of A1 and A2 of the target cell are too low compared to the average RSRP of users in the cell, the trigger thresholds of A1 and A2 events should be increased so that enough users start measurement to trigger the handover and make the optimized parameters effective for the target cell;

[0108] For all target cells, if the trigger threshold value of the target cell A1 and A2 events is set too high compared to the average RSRP of the users in the cell, too many users will start measuring, affecting network performance. In this case, the trigger threshold value of the A1 and A2 events can be lowered to achieve a reasonable ratio of users who start and stop measuring.

[0109] The above parameter configuration is calculated based on the cell. In one example, in the existing network area A, multiple cells share a parameter configuration under a configuration number. For each configuration number, the expected adjustment range of the parameters of all cells under it is counted, and the parameter adjustment range of the configuration number is determined using methods such as weighted average or median.

[0110] S808: The parameter delivery module obtains two parameter adjustment tables output by the cell pair parameter generation module and the cell parameter generation module, and pushes them to the network management device for delivery to perform optimization and adjustment.

[0111] S809: The safety rollback module and the model training module begin to monitor the performance indicators of the optimization area. The safety rollback module ensures that the optimization plan is stopped and rollback parameters are issued when the corresponding performance indicator deteriorates to a certain extent. The model training module regularly collects relevant performance data and performance configuration data, and uses them for incremental training of the model used in S806.

[0112] Application example 2:

[0113] In this application example, it is assumed that the server cannot directly connect to the network management system due to issues such as permissions or network security.

[0114] S901: The data collection module collects data such as performance indicator data, configuration parameters, switching data, and parameter adjustment history logs, and provides these data to modules in need in subsequent steps.

[0115] In this embodiment, since the server cannot be directly connected to the network management, the required data is input into the data acquisition module manually or by using methods such as FTP scheduled tasks. The above data includes performance data, configuration parameters, switching data, and parameter adjustment history logs. The data acquisition module organizes and saves these data, and provides corresponding data to the required modules in the subsequent process. The data acquisition module will notify the relevant staff to input new data when new data is needed.

[0116] S902: Relevant staff inputs division instructions of the network area to be optimized into the area division module.

[0117] The relevant staff inputs the network area to be optimized, and one of the network areas to be optimized is recorded as area B. Area B also needs to meet the requirements of the final draft of this embodiment for the optimization area: it is a geographically continuous area and a convex shape. It is assumed that there is also an unbalanced load in the cell pairs in area B, and the regional network capacity is limited.

[0118] S903: The adjacency statistics module obtains the adjacency of area B from the data collection module. The neighboring area configuration in the parameters of area B is obtained, and combined with the historical switching data statistics calculation in area B, the calculation method is the same as the above application example.

[0119] The adjacency statistics module outputs a list of cell pairs in area B. Each cell pair has a neighbor relationship configured in the configuration parameters, and there is a certain amount of handover between them.

[0120] S904: The security rollback module is responsible for monitoring the corresponding performance indicators of area B.

[0121] When the performance indicators deteriorate by more than a certain extent, the rollback parameters are output and an alarm is issued to notify the staff to review the rollback parameters and issue them, and the two modules that generate the optimization parameters stop working.

[0122] S905: Perform model training to obtain a performance optimization model.

[0123] In this embodiment, it is assumed that this optimization method is used for the first time in area B and there is no ready-made model. This embodiment uses an artificial neural network model to fit the parameter modification effect, and the model training module obtains the parameter adjustment log and the performance indicator data of the corresponding cell from the data acquisition module. The above data are integrated into a training data set, and each piece of data is the state of the cell pair during one adjustment, the change of the cell pair before and after the load adjustment, and the parameter adjustment record. Initialize an artificial neural network, and use a network consisting of 3 linear layers in this application example. Use an open source deep learning framework to build and call the BP algorithm to train the neural network using the above data set. The input of the neural network is the current state of the cell pair and the load difference, and the output is the expected adjustment parameter value. Data needs to be normalized when input into the neural network, and denormalized when output. Normalization and denormalization are consistent during model training and use.

[0124] S906: The cell pair parameter generation module is used to generate the optimization parameters of the cell pair, that is, the individual cell offset of the target cell. In this application example, the cell pair parameter generation module can obtain the cell pair information from the adjacency statistics module, and obtain the load difference between the two cells in the cell pair from the data acquisition module. The performance optimization model obtained by the above training is used to calculate the amplitude of the individual cell offset of the target cell of each target cell pair; and a cell pair parameter adjustment plan table containing the amplitude of the individual cell offset of each target cell is generated. In this application example, the neural network model trained in S905 above is used, and the state of each target cell pair (this parameter is an optional parameter and may not be carried) and the load difference are used as model inputs to obtain the parameter values ​​that need to be adjusted for each cell pair. The cell pair parameter adjustment table for area B is obtained.

[0125] S907: The cell parameter generation module generates optimized parameters for the cell, which may include but is not limited to adjusting the trigger thresholds of the A1 and A2 events of the cell.

[0126] The cell parameter generation module obtains the range of area B from the adjacency generation module, the cell pair parameter generation module obtains the individual offset parameter adjustment plan for each cell pair, and obtains the performance index data of the relevant cell from the data acquisition module (including the average RSRP of the cell users and other signal-related indicators). The A1 and A2 trigger threshold adjustment principles for each target cell are as follows: referring to the average signal strength (average RSRP, number of RSRP reports in each interval, etc.) received by the UE on the target cell, when there are too few users starting measurement on the cell (i.e., the trigger threshold of A1 and A2 events is significantly lower than the average RSRP and other indicators), the trigger threshold of A1 and A2 events is appropriately increased to promote more users to start measurement, so that the cell pair parameters are easier to take effect; when there are too many users starting measurement on the target cell (i.e., when the trigger threshold of A1 and A2 events is higher than the average RSRP and other indicators), the trigger threshold of A1 and A2 events is appropriately reduced to prevent excessive measurements from causing network performance degradation.

[0127] S908: The cell parameter generation module obtains the parameters of each cell through the calculation of S908 above, specifically to the configuration number level.

[0128] In this example, each base station usually has a configuration number corresponding to more than one cell, and these cells share the same set of configurations. In this embodiment, the control of network parameter configuration is relatively strict, and the addition of configuration numbers is not allowed. Therefore, the modification of each configuration number must meet the parameter modification requirements of as many cells as possible among the corresponding cells. For each cell under the configuration number, the average or median of the expected adjustment of each cell is used as the parameter adjustment value of this configuration number. In view of the different load differences of each cell in its respective cell pair, the weight of each cell can be increased when calculating the above average or median. The parameter modification weight of the cell with a large difference in compliance with the neighboring cell is larger, which can better affect the parameter modification range of the configuration number. Finally, the cell parameter generation module calculates the corresponding parameter modification schedule under each configuration number in area B.

[0129] S909: The parameter sending module is responsible for sending the parameter table generated by the cell pair parameter generating module and the cell parameter generating module.

[0130] In this embodiment, the parameter delivery module is not connected to the network management, so the parameter delivery module summarizes two parameter tables and sends them to the staff for review and delivery.

[0131] S910: The safety rollback module and the model training module start monitoring the performance indicators of the optimization area. The safety rollback module ensures that the optimization plan is stopped and rollback parameters are issued when the corresponding performance indicator deteriorates to a certain extent. The model training module regularly collects relevant performance data and performance configuration data, and uses them for incremental training of the model used in S906.

[0132] Application Example 3:

[0133] S1001: The data collection module obtains performance data, configuration parameters, switching data, parameter adjustment history logs and other data from the network management, and provides these data to modules in need in subsequent steps.

[0134] S1002: The area division module uses the performance indicator data and the neighboring cell configuration data between cells to calculate the network area to be optimized.

[0135] Assume that one of the network areas to be optimized is denoted as area C. Area C is a geographically continuous area with a convex shape, including 500 cells. The load difference between neighboring cells in area C is large, there is a serious load imbalance, the network capacity is limited, and capacity optimization is needed.

[0136] S1003: The adjacency statistics module obtains the handover data and configuration parameters from the data collection module, and obtains the target cell pairs existing in area C through statistics.

[0137] For example, it is calculated that there are 1600 target cell pairs in area C. The two target cells in each target cell pair are configured with neighboring cell relations in the configuration parameters, and there are a certain number of handover records between them in the handover data.

[0138] S1004: Before the optimization starts, the safety rollback module obtains the waiting area C from the area division module, and obtains the relevant configuration parameters of the area C before the optimization starts from the data acquisition module and saves these parameters.

[0139] After the optimization starts, the safety rollback module obtains the performance indicator data of area C from the data acquisition module at regular intervals. When the performance indicator data of area C deteriorates beyond expectations, the safety rollback module can start the rollback process. The configuration parameters saved before the optimization are sent through the parameter sending module, and the cell stops working on the optimization parameter generation module and the cell parameter generation module, and the parameter sending module can be notified to stop sending the optimization parameters.

[0140] S1005: Perform model training to obtain a performance optimization model.

[0141] Assume that this capacity optimization is used for the first time in the region, so there is no trained performance optimization model, and model training is required to obtain the performance optimization model. It should be understood that when it is not the first time to use, the previously trained performance optimization model can be directly called.

[0142] In this step, the model training module obtains the network management operation log from the data acquisition module, and parses it to obtain the adjustment record of the relevant parameters. Then the model training module obtains various performance indicators of the cell corresponding to the adjustment record from the data acquisition module, including data such as load change. In this embodiment, a linear regression model is used. The above data are organized into a binary group of (cell pair load difference change value, cell individual offset of the cell in the cell pair), and a scatter plot can be drawn in a two-dimensional plane coordinate system. Using a polynomial fitting model, a linear regression algorithm is used to perform regression fitting on the above data, and the fitting curve obtained by fitting is the obtained performance optimization model (i.e., fitting curve model). For the target cell pair that needs to adjust the load difference, the load difference of the target cell pair is substituted into the model to obtain the cell individual offset amplitude of the target cell that needs to be adjusted.

[0143] S1006: The cell pair parameter generation module is used to generate optimization parameters of the cell pair, that is, the cell individual offset of the target cell.

[0144] The cell pair parameter generation module can obtain the cell pair information from the adjacency statistics module and obtain the load difference between the two cells in the cell pair from the data acquisition module. The performance optimization model (i.e., the fitting curve model) obtained by the above training is used to calculate the amplitude of the cell individual offset of the target cell of each target cell pair; and a cell pair parameter adjustment schedule containing the amplitude of the cell individual offset of each target cell is generated.

[0145] S1007: The cell parameter generation module generates optimized parameters for the cell, which may include but is not limited to adjusting the trigger thresholds of the A1 and A2 events of the cell.

[0146] In this application example, the cell parameter generation module obtains the list of cell pairs to be adjusted from the adjacency statistics module and obtains all relevant target cells; obtains the cell individual offset parameter adjustment plan of the target cells in each target cell pair from the cell pair parameter generation module, thereby obtaining the direction and magnitude of parameter adjustment required for each target cell; obtains performance indicator data such as the average RSRP of users of the corresponding target cell from the parameter acquisition module. In this embodiment, the cell parameter generation module adjusts the measurement event on and off threshold parameters of the corresponding target cell based on the above data, that is, the trigger threshold of A1 and A2 events. The adjustment method is as follows: for the target cell that needs to migrate out users, if the trigger threshold value settings of A1 and A2 of the target cell are too low compared to the average RSRP of the users in the cell, the trigger threshold value of A1 and A2 events should be increased so that enough users can start measurement to trigger switching and make the optimized parameters effective for the target cell; for all target cells, if the trigger threshold value settings of A1 and A2 events of the target cell are too high compared to the average RSRP of the users in the cell, it will cause too many users to start measurement, affecting network performance. At this time, the trigger threshold value of A1 and A2 events can be lowered to make the users who turn on and off measurement reach a reasonable ratio. The above parameter configuration is based on cell calculation. In an example, in the existing network area C, multiple cells share a parameter configuration under a configuration number. For each configuration number, the expected adjustment range of the parameters of all cells under it is counted, and the parameter adjustment range of the configuration number is determined using methods such as weighted average or median.

[0147] S1008: The parameter delivery module obtains two parameter adjustment tables output by the cell pair parameter generation module and the cell parameter generation module, and pushes them to the network management device for delivery to perform optimization and adjustment.

[0148] S1009: In this application example, after a period of time after the optimization, the safety rollback module and the model training module begin to monitor the various performance indicators of the optimization area. The safety rollback module ensures that the optimization plan is stopped and the rollback parameters are issued when the corresponding performance indicator deteriorates to a certain extent. The model training module regularly collects relevant performance data and performance configuration data, and uses them for incremental training of the model used in S1006.

[0149] It can be seen that in the wireless network optimization method provided in this embodiment, the interoperability parameters and measurement-related parameters in the network management device can be adjusted to achieve the purpose of load balancing. For each cell pair with load differences, a fitting model or a neural network model is used to calculate the parameter value that needs to be adjusted for load balancing, thereby achieving the effect of precise adjustment and avoiding the problem of too long fixed step iterative adjustment cycle. In addition, the measurement-related parameters of the corresponding cell are adjusted at the same time to avoid the problem that the switching-related parameters cannot take effect due to the UE not turning on the measurement; it also avoids the problem that too many UEs turn on the measurement and cause the network performance to deteriorate.

[0150] Embodiment three:

[0151] This embodiment also provides a network device, which can be but not limited to a server, see Fig.11 As shown, it includes a processor 1101, a memory 1102 and a communication bus 1103, wherein:

[0152] The communication bus 1103 is used to realize the connection and communication between the processor 1101 and the memory 1102;

[0153] The processor 1101 is configured to execute one or more computer programs stored in the memory 1102 to implement at least one step in the wireless network optimization method in each of the above embodiments.

[0154] The present embodiment also provides a computer-readable storage medium, which includes volatile or non-volatile, removable or non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, computer program modules or other data). Computer-readable storage media include, but are not limited to, RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable read only memory), flash memory or other memory technology, CD-ROM (Compact Disc Read-Only Memory), digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer.

[0155] The computer-readable storage medium in this embodiment can be used to store one or more computer programs, and the one or more computer programs stored therein can be executed by a processor to implement at least one step of the wireless network optimization method exemplified in the above-mentioned embodiments.

[0156] This embodiment also provides a computer program (or computer software), which can be distributed on a computer-readable medium and executed by a computing device to implement at least one step of the method in the above-mentioned embodiment X and embodiment XX; and in some cases, at least one step of the wireless network optimization method different from the examples in the above-mentioned embodiments can be adopted.

[0157] It should be understood that, in some cases, at least one step shown or described may be performed in an order different from that described in the above embodiments.

[0158] This embodiment further provides a computer program product, including a computer readable device, on which the computer program as shown above is stored. In this embodiment, the computer readable device may include the computer readable storage medium as shown above.

[0159] It can be seen that those skilled in the art should understand that all or some of the steps, systems, and functional modules / units in the above disclosed methods can be implemented as software (which can be implemented with computer program code executable by a computing device), firmware, hardware, and appropriate combinations thereof. In hardware implementations, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component can have multiple functions, or a function or step can be performed by several physical components in cooperation. Some or all physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit.

[0160] In addition, it is well known to those skilled in the art that communication media generally contain computer readable instructions, data structures, computer program modules or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery media. Therefore, the present invention is not limited to any specific hardware and software combination.

[0161] The above contents are further detailed descriptions of the embodiments of the present invention in combination with specific implementation methods, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. A wireless network optimization method, comprising: Obtaining performance indicators of the target cell pair to be adjusted; The performance indicator is used as an input of a performance optimization model, and input into the performance optimization model to obtain configuration parameters of the target cell pair, wherein the performance indicator includes a load difference between target cells in the target cell pair; and the configuration parameters include individual cell offset parameters of the target cells in the target cell pair; Acquire a first trigger threshold value of an A1 event and a second trigger threshold value of an A2 event of a target cell in the target cell pair, and an average reference signal received power value of the target cell; When the first trigger threshold value of the A1 event and the second trigger threshold value of the A2 event of the target cell are both less than the average reference signal received power value, and the target cell is a cell from which the load needs to be migrated, the second trigger threshold value and the first trigger threshold value are increased by the same magnitude based on the average reference signal received power value; When the first trigger threshold value of the A1 event and the second trigger threshold value of the A2 event of the target cell are both greater than the average reference signal received power value, reducing the second trigger threshold value and the first trigger threshold value by the same magnitude based on the average reference signal received power value; When the first trigger threshold value of the A1 event of the target cell is greater than the average reference signal received power value, and the second trigger threshold value of the A2 event of the target cell is less than the average reference signal received power value, increasing the second trigger threshold value and the first trigger threshold value by the same magnitude based on the average reference signal received power value; Optimizing and adjusting at least one target cell in the target cell pair according to the configuration parameters; The configuration parameters are parameters that affect the performance indicators, and the performance optimization model is obtained by training based on historical performance indicators of cells in the wireless network and historical configuration parameters corresponding to the historical performance indicators.

2. The wireless network optimization method according to claim 1, characterized in that: The target cell pair is a cell pair in the network area to be optimized in the wireless network, and the performance indicator meets the preset optimization condition; The performance optimization model is obtained by training according to historical performance indicators of cells in the network area to be optimized and historical configuration parameters corresponding to the historical performance indicators.

3. The wireless network optimization method according to claim 2, characterized in that: The network area to be optimized is geographically continuous, and the figure formed by the boundary lines of the peripheral cells in the area is a convex figure.

4. The wireless network optimization method according to any one of claims 1 to 3, characterized in that: The wireless network optimization method further includes: Collecting performance indicators of cells in the wireless network and configuration parameters corresponding to the performance indicators; The performance optimization model is incrementally trained according to the collected performance indicators and configuration parameters.

5. The wireless network optimization method according to any one of claims 1 to 3, characterized in that: The process of optimizing and adjusting at least one target cell in the target cell pair according to the configuration parameter, or after optimizing and adjusting at least one target cell in the target cell pair according to the configuration parameter, further comprising: Monitoring the performance indicators of cells in the wireless network; When it is monitored that the performance indicator reaches a preset deterioration condition, the configuration parameters of at least one target cell in the target cell pair are restored to the configuration parameters before the optimization adjustment.

6. The wireless network optimization method according to any one of claims 1 to 3, characterized in that: The performance optimization model includes at least one of the following models: A linear regression model is used to obtain a fitted straight line model by performing regression fitting with the historical performance index as input and the historical configuration parameter as output; A fitting curve model is obtained by performing polynomial fitting using a polynomial fitting model with the historical performance index as input and the historical configuration parameter as output; A neural network model is obtained by using a neural network and calling a preset training algorithm, taking the historical performance indicators as input and the historical configuration parameters as output for training.

7. A network device, characterized in that: including a processor, a memory and a communication bus; The communication bus is used to realize the connection and communication between the processor and the memory; The processor is used to execute one or more computer programs stored in the memory to implement the steps of the wireless network optimization method according to any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores one or more computer programs, and the one or more computer programs can be executed by one or more processors to implement the steps of the wireless network optimization method according to any one of claims 1 to 6.

Images