Outer loop value determination method, device, equipment and storage medium

Abstract

The present application provides a method, apparatus, device and storage medium for determining an outer loop value. First, grid division is performed according to user feature data in the network to train a model; a pre-trained outer loop initialization model is determined based on current feature data of a user device; an initialization outer loop value of the user device is determined based on a current air interface measurement value of the user device and the outer loop initialization model, and the model is applied; the model is evaluated according to the application effect; and the model is corrected according to the evaluation effect.

Description

Technical Field

[0001] The present application relates to the field of wireless communication technology, and in particular to a method, apparatus, device and storage medium for determining an outer loop value. Background Art

[0002] Due to the rapid changes in wireless channels, it is difficult to fully utilize spectrum resources if fixed modulation and coding methods are used. Therefore, Adaptive Modulation and Coding (AMC) technology has been introduced in wireless communication technology to improve spectrum efficiency.

[0003] Traditional AMC uses Acknowledgement / Non-Acknowledgement (ACK / NACK) to adjust the outer loop value to adapt the modulation and coding strategy to user channel conditions, thereby achieving the network's target reliability requirements. However, this convergence process often requires dozens or even hundreds of transmissions from user devices. This can result in some user devices not converging before transmission ends, or even convergence but with the previous dozens or even hundreds of transmissions not utilizing the optimal modulation and coding strategy, limiting spectral efficiency.

[0004] With the introduction of Massive Mimo technology in fifth-generation mobile communications, information transmission fully utilizes airspace resources. In addition, due to the larger transmission bandwidth, these factors will result in 5G scheduling data packets less frequently under the condition of the same data packet size, further exacerbating the problem of resource waste and reduced user perception caused by the non-convergence of small packet users. Summary of the Invention

[0005] The present application provides a method, apparatus, device and storage medium for determining an outer loop value, so as to improve the convergence speed of adaptive modulation and coding technology.

[0006] In a first aspect, an embodiment of the present application provides a method for determining an outer loop value, comprising:

[0007] Determining a pre-trained outer loop initialization model based on current feature data of the user device;

[0008] An initialization outer loop value of the user equipment is determined based on a current air interface measurement value of the user equipment and the outer loop initialization model.

[0009] In one embodiment, the technical solution of the present application mainly includes:

[0010] Step 1: Train the outer loop initialization model based on historical feature data.

[0011] Among them, the training of the outer loop initialization model based on historical feature data includes: performing palace grid division and grid division based on the historical feature data; determining the number of ACK / NACK information and the outer loop value in each grid; calculating the block error rate of each grid based on the number of ACK / NACK information in each grid; and determining the outer loop initialization model based on the block error rate and outer loop value of each grid.

[0012] The grid division and raster division based on the historical feature data include: performing grid division on each cell in the network based on the historical feature data; and performing raster division on each grid according to the scheduled MCS and the corresponding air interface measurement value.

[0013] The determining of the number of ACK / NACK information and the outer loop value in each grid includes: obtaining confirmation response / negative response (ACK / NACK) information corresponding to the demodulation; delivering the ACK / NACK information and the corresponding outer loop value to the corresponding grid; and counting the number of ACK / NACK information and the outer loop value in each grid.

[0014] Among them, the outer loop initialization model is determined based on the block error rate and outer loop value of each grid, including: in each grid, determining the MCS corresponding to the air interface measurement value and the outer loop value corresponding to the air interface measurement value under the set target block error rate BLER interval based on the block error rate of each grid; determining the first outer loop model based on the correspondence between the air interface measurement value and the MCS; and determining the second outer loop model based on the correspondence between the air interface measurement value and the outer loop value.

[0015] Among them, the method of determining the MCS corresponding to the air interface measurement value under the set target BLER interval based on the block error rate of each grid and the air interface measurement value and the corresponding outer loop value includes: for an MCS, if the number of grids that meet the set BLER interval is greater than a preset value, calculating the weight factor of each grid; performing weighted averaging calculation on the air interface measurement value corresponding to each grid according to the weight factor of each grid to obtain the air interface measurement value corresponding to the MCS; performing weighted averaging calculation on the outer loop value corresponding to each grid according to the weight factor of each grid to obtain the outer loop value corresponding to the MCS; and determining the outer loop value corresponding to the air interface measurement value based on the air interface measurement value corresponding to the MCS and the outer loop value corresponding to the MCS.

[0016] Among them, the method of determining the MCS corresponding to the air interface measurement value of the set target BLER interval based on the block error rate of each grid and the air interface measurement value and the corresponding outer loop value includes: for an MCS, if the number of grids that meet the set BLER interval is less than a preset value, searching for the air interface measurement value interval corresponding to the target BLER interval corresponding to the MCS on the reference demodulation curve; using the difference between the median value in the searched air interface measurement value interval and the system default initialization outer loop value as the air interface measurement value corresponding to the MCS; and using the system default initialization outer loop value as the outer loop value corresponding to the air interface measurement value.

[0017] Step 2: Application of the outer loop initialization model.

[0018] The outer loop initialization model includes a first outer loop model and a second outer loop model; the pre-trained outer loop initialization model is determined based on the current feature data of the user device, including: determining the grid to which the user device belongs based on the current feature data of the user device and the grid division strategy; obtaining the first outer loop model and the second outer loop model corresponding to the grid to which the user device belongs.

[0019] Among them, the determining of the initialization outer loop value of the user equipment based on the current air interface measurement value of the user equipment and the outer loop initialization model includes: determining the first outer loop value based on the current air interface measurement value and the first outer loop model; determining the second outer loop value based on the current air interface measurement value and the second outer loop model; and determining the initialization outer loop value based on the first outer loop value and the second outer loop value.

[0020] The determining of the first outer loop value based on the current air interface measurement value and the first outer loop model includes:

[0021] Determine the modulation and coding strategy MCS corresponding to the current air interface measurement value based on the current air interface measurement value and the first outer loop model; wherein the first outer loop model is a correspondence model between the air interface measurement value and the MCS under the set target BLER; obtain the system mapping SINR corresponding to the target BLER based on the determined MCS and the reference demodulation curve; and determine the difference between the system mapping SINR and the current air interface measurement value as the first outer loop value.

[0022] Among them, determining the second outer loop value based on the current air interface measurement value and the second outer loop model includes: obtaining the second outer loop value through the second outer loop model based on the current air interface measurement value; wherein, the second outer loop model is a correspondence model between the air interface measurement value and the outer loop value under the set target BLER interval.

[0023] Step 3: Start online model learning.

[0024] Obtain a first network performance indicator before using the outer loop initialization model and a second network performance indicator after using the outer loop initialization model; when the first network performance indicator and the second network performance indicator meet preset conditions, start online learning to obtain an outer loop online model; use the outer loop online model to determine a new initialized outer loop value of the user equipment.

[0025] Step 4: Modify the outer ring model.

[0026] The outer loop model of the corresponding grid is modified based on the ACK / NACK information after the user in the grid uses the outer loop model for the first time; the outer loop model includes a pre-trained outer loop initialization model or an outer loop online model.

[0027] Among them, the outer loop model of the corresponding grid is corrected based on the ACK / NACK information after the user in the grid uses the outer loop model for the first time, including: for each grid, calculating the proportion of negative responses NACK in the ACK / NACK information; when the proportion of NACK is greater than the target value, reducing the outer loop value of the outer loop model according to the corresponding strategy; when the proportion of NACK is less than the target value, increasing the outer loop value of the outer loop model according to the corresponding strategy.

[0028] In a second aspect, an embodiment of the present application provides an outer loop value determination device, comprising:

[0029] a model determination module, configured to determine a pre-trained outer-loop initialization model based on current feature data of the user equipment;

[0030] The outer loop value determination module is configured to determine an initialized outer loop value of the user equipment based on a current air interface measurement value of the user equipment and the outer loop initialization model.

[0031] In a third aspect, an embodiment of the present application provides a device, including:

[0032] one or more processors;

[0033] a memory for storing one or more programs;

[0034] When the one or more programs are executed by the one or more processors, the one or more processors implement any one of the methods in the embodiments of the present application.

[0035] In a fourth aspect, an embodiment of the present application provides a storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, any one of the methods in the embodiments of the present application is implemented.

[0036] The outer loop value determination method, apparatus, device and storage medium provided in the embodiments of the present application solve the problems of resource waste and reduced user perception caused by non-convergence of small packet users and improve the convergence speed of adaptive modulation and coding technology by determining a pre-trained outer loop initialization model based on the current feature data of the user equipment and determining the initialization outer loop value of the user equipment based on the current air interface measurement value of the user equipment and the outer loop initialization model.

[0037] With respect to the above embodiments and other aspects of the present application and their implementation, further description is provided in the accompanying drawings, detailed description and claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a flow chart of a method for determining an outer loop value provided by an embodiment of the present application;

[0039] Figure 2 This is a flowchart of learning the outer loop initialization model in an offline state provided by an embodiment of the present application;

[0040] Figure 3 This is a schematic diagram of the grid division provided in an embodiment of the present application;

[0041] Figure 4 is a schematic diagram of grid division provided in an embodiment of the present application;

[0042] Figure 5 is a flow chart of the model application method provided in an embodiment of the present application;

[0043] Figure 6 This is a flowchart of a method for triggering offline learning to online learning provided by an embodiment of the present application;

[0044] Figure 7 This is a flowchart of the online learning method for the outer loop initialization model provided by an embodiment of the present application;

[0045] Figure 8 This is a flow chart of the outer loop initialization model correction method provided by an embodiment of the present application;

[0046] Figure 9 This is an overall flow chart for determining the outer loop value provided by an embodiment of the present application;

[0047] Figure 10 Schematic diagram of the structure of an outer loop value determination device provided in an embodiment of the present application;

[0048] Figure 11 It is a structural diagram of a device provided by this application. DETAILED DESCRIPTION

[0049] To make the purpose, technical solutions and advantages of this application more clear, the embodiments of this application will be described in detail below with reference to the accompanying drawings. It should be noted that, unless there is a conflict, the embodiments and features in the embodiments of this application can be combined with each other in any way.

[0050] The steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions. Also, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be performed in an order different from that shown here.

[0051] With the development of mobile communications technology, people's demand for communication has gradually increased, from voice-based communications to today's data-based communications. Fourth-generation mobile communications, in particular, have gradually transformed our lifestyles. Today, people's communication needs continue to grow. For example, they want to watch high-definition live broadcasts in real time at higher service rates; communicate face-to-face with loved ones thousands of miles away in a virtual environment; and feel as if they are actually there. While current networks offer excellent visual and auditory experiences, people also desire a sense of touch. There is also growing demand for the Internet of Things in vertical industries such as transportation, public services, and manufacturing, as well as for telemedicine. Fifth-generation mobile communications technology has emerged to meet these needs.

[0052] The future of 5G communications, with its diverse services and communication methods, will inevitably increase the complexity of wireless solution design. Using existing planning methods to solve current or future problems will likely exponentially increase the complexity, rendering them unsolvable. However, future communications will generate vast amounts of data. Machine learning can be used to develop relevant models based on this empirical data, leveraging these models to make predictions and develop strategies.

[0053] Compared to wired channels, wireless channels have narrower coherence bandwidths and shorter coherence times, which are notable characteristics of wireless channels. Due to the rapidly changing nature of wireless channels, using fixed modulation and coding schemes makes it difficult to fully utilize spectrum resources. Therefore, adaptive coding and modulation (ACM) has been introduced in wireless communications to improve spectrum efficiency. The basic concept of ACM is to adapt the modulation and coding strategy to channel conditions, thereby improving spectrum efficiency and user experience.

[0054] To adapt the modulation and coding strategy to user channel conditions, traditional adaptive modulation and coding technology requires adjusting the outer loop based on ACK / NACK information fed back by users to achieve the system's target reliability requirements. However, this convergence process often requires user devices to transmit dozens or even hundreds of times. This can result in some user devices completing transmissions before convergence, or even after convergence, but without using the optimal modulation and coding strategy for the first dozens or even hundreds of transmissions, thus limiting spectral efficiency improvements. Furthermore, the suboptimal modulation and coding scheme can also affect user experience.

[0055] Both 4G and 5G networks have a large number of small-packet users, meaning users with low traffic volumes. These characteristics are a key factor limiting the ability of adaptive coding and modulation schemes to improve spectral efficiency. Because of their small packets, there are fewer learning opportunities for the adaptive coding process to converge. Furthermore, because 5G utilizes a larger bandwidth and more antennas to fully utilize spatial dimensions, 5G will have fewer transmission opportunities for users under the same traffic volume.

[0056] In response to the above problems, this application provides the following technical solutions.

[0057] In one embodiment, the present application provides a method for determining an outer loop value. Figure 1 This is a flowchart of a method for determining an outer loop value provided in an embodiment of the present application. This embodiment is applicable to the case of adaptive coding modulation technology. The method can be executed by the outer loop value determination device provided in an embodiment of the present application, and the device can be implemented in software and / or hardware.

[0058] like Figure 1 As shown, the outer loop value determination method provided in the embodiment of the present application mainly includes steps S11 and S12.

[0059] S11. Determine a pre-trained outer-loop initialization model based on current feature data of the user equipment.

[0060] S12: Determine an initialization outer loop value of the user equipment based on a current air interface measurement value of the user equipment and the outer loop initialization model.

[0061] It should be noted that the pre-trained outer-loop initialization model is an outer-loop initialization model trained in an offline state, and the offline model in this application may refer to the pre-trained outer-loop initialization model.

[0062] In one embodiment, the outer loop initialization model includes a first outer loop model and a second outer loop model; the determination of the pre-trained outer loop initialization model based on the current feature data of the user device includes: determining the grid to which the user device belongs based on the current feature data of the user device and the grid division strategy; obtaining the first outer loop model and the second outer loop model corresponding to the grid to which the user device belongs.

[0063] In one embodiment, the method of determining the initialization outer loop value of the user equipment based on the current air interface measurement value of the user equipment and the outer loop initialization model includes: determining the first outer loop value based on the current air interface measurement value and the first outer loop model; determining the second outer loop value based on the current air interface measurement value and the second outer loop model; and determining the initialization outer loop value based on the first outer loop value and the second outer loop value.

[0064] In one embodiment, determining the first outer loop value based on the current air interface measurement value and the first outer loop model includes: determining the modulation and coding strategy MCS corresponding to the current air interface measurement value based on the current air interface measurement value and the first outer loop model; wherein the first outer loop model is a correspondence model between the air interface measurement value and the MCS under the set target BLER interval; obtaining the system mapping SINR corresponding to the target BLER based on the determined MCS and the reference demodulation curve; and determining the difference between the system mapping SINR and the current air interface measurement value as the first outer loop value.

[0065] In one embodiment, determining the second outer loop value based on the current air interface measurement value and the second outer loop model includes: obtaining the second outer loop value through the second outer loop model based on the current air interface measurement value; wherein, the second outer loop model is a correspondence model between the air interface measurement value and the outer loop value under the set target BLER interval.

[0066] In one embodiment, before determining the pre-trained outer-loop initialization model based on the current feature data of the user equipment, the method further includes: training the outer-loop initialization model based on the historical feature data of the user equipment.

[0067] In one embodiment, the training of the outer loop initialization model based on the historical feature data of the user device includes: performing grid division and grid division based on the historical feature data; determining the number of confirmation response / negative response ACK / NACK information and the outer loop value in each grid; calculating the block error rate of each grid based on the number of ACK / NACK information in each grid; and determining the outer loop initialization model based on the block error rate and outer loop value of each grid.

[0068] In one embodiment, performing grid division and raster division based on the historical feature data includes: performing grid division on each cell in the network based on the historical feature data; and performing raster division on each grid according to the scheduled MCS and the corresponding air interface measurement value.

[0069] In one embodiment, determining the number of ACK / NACK information and the outer loop value in each grid includes: obtaining confirmation response / negative response (ACK / NACK) information corresponding to the demodulation; delivering the ACK / NACK information and the corresponding outer loop information to the corresponding grid; and counting the number of ACK / NACK information and the outer loop value in each grid.

[0070] In one embodiment, the outer loop initialization model is determined based on the BLER and outer loop value of each grid, including: in each grid, determining the MCS corresponding to the air interface measurement value and the outer loop value corresponding to the air interface measurement value under the set target BLER interval based on the BLER of each grid; determining a first outer loop model based on the correspondence between the air interface measurement value and the MCS; and determining a second outer loop model based on the correspondence between the air interface measurement value and the outer loop value.

[0071] In one embodiment, the method of determining the MCS corresponding to the air interface measurement value under the set target block error rate BLER interval based on the block error rate of each grid and the outer loop value corresponding to the air interface measurement value includes: for an MCS, if the number of grids that meet the set block error rate BLER interval is greater than a preset value, calculating the weight factor of each grid according to the number of ACK / NACK information in each grid; performing weighted averaging calculation on the air interface measurement value corresponding to each grid according to the weight factor of each grid to obtain the air interface measurement value corresponding to the MCS; performing weighted averaging calculation on the outer loop value corresponding to each grid according to the weight factor of each grid to obtain the outer loop value corresponding to the MCS; and determining the outer loop value corresponding to the air interface measurement value according to the air interface measurement value corresponding to the MCS and the outer loop value corresponding to the MCS.

[0072] In one embodiment, the BLER of each grid is used to determine the MCS corresponding to the air interface measurement value under the set target BLER interval and the outer loop value corresponding to the air interface measurement value, including: for an MCS, if the number of grids that meet the set block error rate interval is less than a preset value, searching for the air interface measurement value interval corresponding to the target BLER interval corresponding to the MCS on the reference demodulation curve; using the difference between the median value in the searched air interface measurement value interval and the system default initialization outer loop value as the air interface measurement value corresponding to the MCS; and using the system default initialization outer loop value as the outer loop value corresponding to the air interface measurement value.

[0073] In one embodiment, after determining the initialized outer loop value of the user equipment based on the current air interface measurement value of the user equipment and the outer loop initialization model, the method further includes: obtaining a first network performance indicator before applying the model and a second network performance indicator after using the outer loop initialization model; when the first network performance indicator and the second network performance indicator meet preset conditions, starting online learning to obtain the outer loop online model; and using the outer loop online model to determine a new initialized outer loop value for the user equipment.

[0074] It should be noted that the outer-loop online model is an outer-loop initialization model trained online, and the online models in this application may all refer to outer-loop online models.

[0075] In one embodiment, the method further includes: correcting the outer loop model of the corresponding grid based on the ACK / NACK information after the user in the grid uses the outer loop model for the first time; the outer loop model includes a pre-trained outer loop initialization model or an outer loop online model.

[0076] In one embodiment, the outer loop model of the corresponding grid is corrected based on the ACK / NACK information after the user in the grid uses the outer loop model for the first time, including: for each grid, calculating the proportion of negative responses NACK in the ACK / NACK information; when the proportion of NACK is greater than the target value, reducing the outer loop value of the outer loop model according to the corresponding strategy; when the proportion of NACK is less than the target value, increasing the outer loop value of the outer loop model according to the corresponding strategy.

[0077] In one embodiment, a method for learning an outer loop initialization model in an offline state is provided. Figure 2 As shown in Figure 2, the learning method of the outer loop initialization model in the offline state mainly includes the following steps:

[0078] S21. Divide the grid into sections based on historical feature data.

[0079] The grids are divided according to the historical feature data of users in the network collected.

[0080] Specifically, historical feature data includes: user transmission mode, scheduled air interface measurement SINR, air division label; number of air division streams; maximum supported modulation order (64QAM, 256QAM); corresponding demodulation ACK / NACK; outer loop information, etc. The user transmission mode includes: single port, closed loop multiplexing RI=1, closed loop multiplexing RI=2.

[0081] like Figure 3As shown, according to the set interference level (high interference, low interference) and path loss level division threshold, the cell is divided into four grids, corresponding to low interference and low path loss grid 1, high interference and low path loss grid 2, low interference and high path loss grid 3, and high interference and high path loss grid 4.

[0082] S22: For each grid, divide the grid into grids according to the scheduled MCS and the corresponding air interface measurement value SINR.

[0083] like Figure 4 As shown, MCS is divided into two dimensions with a granularity of 1 order as the horizontal axis, and the air interface measurement SINR value is divided into two dimensions with a granularity of 0.5 dB as the vertical axis.

[0084] S23, looping and offline capturing new feature data.

[0085] S24. Determine the grid to which the new feature data belongs and the grid under the corresponding grid based on the captured new feature data.

[0086] S25 : Delivering the ACK / NACK information and outer loop information corresponding to the actual demodulation to the corresponding grid of the palace.

[0087] S26: Determine whether the captured new feature data has been processed. If so, execute S27; if not, execute S23.

[0088] S27. Statistically calculate the BLER and outer ring value of each grid in each grid.

[0089] Specifically, the BLER of the corresponding grid and the corresponding outer ring value are calculated based on the number of ACK / NACK statistics of each grid in each grid. If the number of samples of the corresponding grid is lower than a certain threshold, the grid is marked as invalid and recorded as NULL.

[0090] S28. Determine a corresponding relationship model between the air interface measurement value SINR and the scheduling MCS and a corresponding relationship model between the air interface measurement value SINR and the outer loop value in the set target BLER interval according to the statistical BLER of each grid in the grid and the corresponding outer loop value.

[0091] The specific method is to search for a grid in which the corresponding BLER of each MCS satisfies the set BLER range, thereby determining the air interface measurement value SINR corresponding to the MCS and the outer loop value corresponding to the air interface measurement SINR.

[0092] S281: If the number of grids that meet the set BLER interval for an MCS is greater than one, calculate a weight factor for each grid.

[0093] The calculation method is: the ratio of the number of samples contained in the grid that meets the conditions (the total number of ACK / NACKs) to the sum of the number of samples of all grids that meet the conditions.

[0094] S282: Perform weighted averaging of the air interface measurement SINR values ​​corresponding to each grid according to the weight factor, and perform weighted averaging of the outer ring values ​​corresponding to each grid.

[0095] S283: Determine the scheduled MCS that satisfies the target BLER interval under the air interface measurement SINR, and the outer loop value corresponding to the air interface measurement SINR.

[0096] S284. If an MCS does not meet the set BLER interval grid, find the SINR value interval corresponding to the target BLER interval of this MCS on the system reference demodulation curve (usually a simulated curve), and then take the median of the SINR interval minus the system default initialization outer loop value to obtain the air interface measurement value SINR corresponding to the MCS. The default initialization outer loop value is its corresponding outer loop value.

[0097] S285. Through the above process, it is possible to determine the scheduling MCS that satisfies the target BLER interval corresponding to each air interface measurement value SINR, and the outer loop value corresponding to each air interface measurement value SINR, that is, the learned model is set to the target BLER interval.

[0098] In one embodiment, a model application method is provided. Figure 5 As shown in Figure 2, the model application method mainly includes the following steps:

[0099] S31. Determine the grid to which the user belongs based on the current feature data corresponding to the user, and schedule the model of the corresponding grid.

[0100] Specifically, the grid to which the user's device belongs is determined based on the user's current feature data and the grid division strategy, and the corresponding grid model is obtained. and

[0101] S32, according to the air interface measurement value SINR0 corresponding to the user equipment, the model The corresponding MCS is determined, and the median value of the corresponding SINR in the target BLER interval is obtained from the MCS and the system reference BLER curve and recorded as SINR1. The first outer loop value D_SINR0=SINR1-SINR0 is calculated from SINR0 and SINR1.

[0102] S33, the air interface measurement value SINR0 corresponding to the user equipment is obtained through the model The corresponding second outer loop value is obtained and recorded as: D_SINR1.

[0103] S34. Calculate an initialization outer loop value of the user equipment: D_SINR_Init according to the first outer loop value D_SINR0 and the second outer loop value D_SINR1.

[0104] In one embodiment, a triggering method for switching from offline learning to online learning is provided. Figure 6 As shown in Figure 1, the triggering method for offline learning to online learning mainly includes the following steps:

[0105] S41 , a set of corresponding relationships between SE and average air interface measurement value SINR in the corresponding network before the statistical model is applied, and a proportion of user-level BLER falling within the target BLER range.

[0106] S42: A corresponding relationship set between SE and average air interface measurement value SINR in the corresponding network after the statistical model is applied, and a proportion of user-level BLER falling within the target BLER range.

[0107] S43. Determine whether the corresponding relationship sets before and after the model application have an intersection. If so, execute S44; if not, execute S45.

[0108] S44. Determine whether to perform online learning based on the intersection and the convergence ratio. If yes, execute S46; if not, execute S47.

[0109] S45. Determine whether to perform online learning based on the convergence ratio. If yes, execute S46; if not, execute S47.

[0110] S46. Trigger online learning.

[0111] S47. Continue to use the offline model.

[0112] In one embodiment, an online learning method for an outer loop initialization model is provided. Figure 7 As shown in Figure 2, the outer loop model online learning method mainly includes the following steps:

[0113] S51. Determine feature segmentation points based on the distribution of newly captured feature data, and then perform grid division.

[0114] S52: For each grid, divide it into grids according to SINR and scheduling MCS.

[0115] S53. Acquire feature data online.

[0116] S54: Determine the grid to which the data belongs and the grid under the corresponding grid according to the data characteristics.

[0117] S55: Deliver the ACK / NACK information and outer loop information corresponding to the data to the grid to which it belongs.

[0118] S56: Determine whether the sample size meets the requirements. If so, execute S57; if not, execute S53.

[0119] S57, statistically calculating the BLER under each grid and the outer ring value.

[0120] S58. Select the corresponding SINR and outer loop value within the target BLER interval for each MCS.

[0121] S59: Determine a relationship model between the online learned SINR and the scheduling MCS and a relationship model between the SINR and the outer loop value.

[0122] In one embodiment, a method for modifying the outer loop initialization model is provided. Figure 8 As shown in Figure 2, the outer loop initialization model correction method mainly includes the following steps:

[0123] S61 : Initialize ACK / NACK information corresponding to the outer loop scheduling according to the grid cache adoption model, and calculate the NACK ratio.

[0124] S62: Calculate a model adjustment amount according to the relationship between the calculated NACK ratio and the target BLER.

[0125] S63. Modify the model using the above adjustment amount.

[0126] S64. The modified model is passed to the model application module.

[0127] In one embodiment, an overall flow chart of outer loop value determination is provided, such as Figure 9 As shown, it mainly includes 7 steps: offline learning, model application, model evaluation, online learning, model application, effect evaluation and model modification.

[0128] In one embodiment, a method for determining an outer loop value is provided. The method for determining an outer loop value mainly includes the following steps:

[0129] S81. Divide the grid into sections based on historical feature data.

[0130] The grids are divided according to the historical feature data of users in the network collected.

[0131] Specifically, historical feature data includes: user transmission mode, scheduled air interface measurement SINR, air division label; number of air division streams; maximum supported modulation order (64QAM, 256QAM); corresponding demodulation ACK / NACK; outer loop information, etc. The user transmission mode includes: single port, closed loop multiplexing RI=1, closed loop multiplexing RI=2.

[0132] In the case of non-space division: divided into 24 grids according to three transmission modes (single port, closed-loop multiplexing RI=1, closed-loop multiplexing RI=2), two interference levels (high interference, low interference), the maximum supported modulation order (64QAM, 256QAM), and the number of RBs (large and small RBs). In the case of space division: divided into 40 grids according to the number of space division streams (2, 3, 4, 5, and greater than or equal to 6, divided into 5 types), two interference levels (high interference, low interference), the maximum supported modulation order (64QAM, 256QAM), and the number of RBs (large and small RBs). Furthermore, the grid-level division can also include the user's chip type information, demodulation algorithm information, user beam position information, etc. It should be noted that the above grid division method is only an example and not a limitation.

[0133] S82: For each grid, divide the grid into grids according to the scheduled MCS and the corresponding air interface measurement value SINR.

[0134] For example, MCS is divided into two dimensions with order 1 as the granularity and SINR value is divided into two dimensions with order 1 as the horizontal axis and SINR value is divided into two dimensions with 0.5 dB as the vertical axis.

[0135] S83. Determine the grid to which the data belongs and the grid under the corresponding grid based on the captured new feature data, and deliver the ACK / NACK information corresponding to the actual demodulation and the outer loop information to the corresponding grid of the grid.

[0136] Specifically, determine whether it is space-divided based on the captured new feature data. If it is not space-divided, find the grid to which the user data belongs according to the transmission mode, interference level, supported maximum modulation order configuration information, and the number of scheduled RBs, and deliver the corresponding ACK / NACK information corresponding to the actual demodulation and the outer loop information to the corresponding grid of the grid based on the scheduled MCS and the air interface measurement value SINR. Record the number of ACK / NACK information (two respectively) and the outer loop value in each grid.

[0137] If space division is used, find the grid to which the user data belongs based on the number of space division flows, interference level, supported maximum modulation order configuration information, and the number of scheduled RBs; and deliver the corresponding ACK / NACK information corresponding to the actual demodulation and the outer loop information to the corresponding grid of the grid based on the scheduled MCS and air interface measurement value SINR. Record the number of ACK / NACKs (two respectively) and the outer loop value in each grid.

[0138] S84. Statistically calculate the BLER under each grid and the outer ring value.

[0139] Specifically, the BLER of the corresponding grid and the outer ring value of the corresponding grid are calculated based on the number of ACK / NACK statistics of each grid in each grid. If the number of samples of the corresponding grid is lower than a certain threshold, the grid is marked as invalid and recorded as NULL.

[0140] S85. Determine a correspondence model between the air interface measurement value SINR and the scheduling MCS in the target BLER interval and a correspondence model between the air interface measurement value SINR and the outer loop value in the target BLER interval according to the BLER statistics of each grid in the grid and the corresponding outer loop value.

[0141] The specific method is to search for a grid in which the corresponding BLER of each MCS satisfies the set BLER range, thereby determining the air interface measurement value SINR corresponding to the MCS and the outer loop value corresponding to the air interface measurement value SINR.

[0142] S851: If the number of grids that meet the set BLER interval for an MCS is greater than one, calculate a weight factor for each grid.

[0143] The calculation method is: the ratio of the number of samples contained in the grid that meets the conditions (the total number of ACK / NACKs) to the sum of the number of samples of all grids that meet the conditions.

[0144] S852: Perform weighted averaging of the air interface measurement SINR values ​​corresponding to each grid according to the weight factor, and perform weighted averaging of the outer ring values ​​corresponding to each grid.

[0145] S853: Determine the air interface measurement value SINR and the outer loop value that meet the target BLER range under the MCS.

[0146] S854. If an MCS does not meet the set BLER interval grid, find the interval of the demodulation value SINR corresponding to the target BLER interval of this MCS on the system reference demodulation curve (usually a simulated curve), and then take the median of the interval minus the system default initialization outer loop to obtain the air interface measurement SINR value corresponding to the MCS. The default initialization outer loop value is its corresponding outer loop value.

[0147] S855. Through the above process, the scheduling MCS that satisfies the target BLER interval and the outer loop value corresponding to each air interface measurement SINR can be determined. That is, the learned model is:

[0148] The relationship model between the air interface measurement value SINR and the scheduling MCS in the target BLER interval is recorded as:

[0149]

[0150] The relationship model between the air interface measurement value SINR and the outer loop value ΔSINR in the target BLER interval is recorded as:

[0151]

[0152] Among them, [0.08, 0.12] is the target BLER interval, Indicates the combination of air interface measurement value and scheduling MCS that meets the target BLER interval; Indicates the combination of air interface measurement value and outer loop value that meets the target BLER interval; BLER(·) represents the BLER under the corresponding combination; Represents the relationship model between the air interface channel quality corresponding to the j-th grid and the scheduling MCS learned offline; represents the relationship model between the air interface channel quality and the outer loop value corresponding to the j-th grid learned offline; j = 1, ... 64; N represents the number of MCSs supported by the system.

[0153] S86. Before applying the outer loop initialization model, the relationship set between the SE in the corresponding network and the air interface measurement SINR is:

[0154]

[0155] The proportion of the statistical user-level BLER falling within the target BLER range is recorded as R0; where K0 is the number of sample points, and Set_p0 represents the network performance set before applying the model.

[0156] S87, outer loop initialization model application.

[0157] S871. Determine the grid to which the user belongs based on the user's corresponding feature data and the grid division strategy, and obtain the corresponding grid model. and model

[0158] S872, according to the current air interface measurement value SINR0 corresponding to the user, the model Determine the corresponding MCS, obtain the median of the corresponding SINR in the target BLER interval from the MCS and the system reference BLER curve, record it as SINR1, and calculate the outer loop value D_SINR0=SINR1-SINR0 from SINR0 and SINR1.

[0159] S873, the current air interface measurement value SINR corresponding to the user is calculated through the model The corresponding outer loop value is obtained and recorded as: D_SINR1.

[0160] S874. Calculate the outer loop value initialized to the user:

[0161] D_SINR_Init=min{max{(D_SINR0+D_SINR1) / 2,-V_limt},V_limt}

[0162] Among them, V_limt is the protection limit value, its value is greater than 0, and the default value is 10.

[0163] S88. After applying the model, the relationship set between the SE of the corresponding network and the average air interface measurement SINR is: The proportion of the statistical user-level BLER falling within the target BLER range is recorded as R1; K1 is the number of sample points, and Set_p1 represents the network performance set before applying the model.

[0164] S89 . Determine whether the air interface channel qualities SINR in Set_p0 and Set_p1 have an intersection.

[0165] S891. If there is an intersection, calculate the performance difference corresponding to the intersection and record it as: {D_SE i |i=1,2,…M}

[0166] where D_SE i It represents the difference between the SE after applying the model and the SE before applying the model corresponding to the i-th element in the above intersection.

[0167] S8911、Calculate {D_SE i The proportion R2 of elements with a value of |i=1,2,…M} less than 0, and the mean value D_SE_mean of the elements in the set.

[0168] S8912: If D_SE_mean≥Th0 and R2≥Th1 and R0≥R1 or D_SE_mean<Th0, execute S810. Otherwise, continue to use the current model to initialize the outer loop of the newly connected user.

[0169] S892: If there is no intersection and R0≥R1, execute S810. Otherwise, continue to use the current model to initialize the outer loop of the newly connected user.

[0170] S810: Stop using the current model and start online learning.

[0171] S8101. Collect data related to user level and scheduling information online, including user transmission mode: single port, closed-loop multiplexing RI=1, closed-loop multiplexing RI=2; interference level; scheduling MCS; number of allocated RBs; air interface measurement SINR; space division label; number of space division streams; maximum supported modulation order (64QAM, 256QAM); corresponding demodulation ACK / NACK; outer loop information; and initialize the model update flag Model_change_index=0; and initially set the model performance evaluation set AI_ΔSINR_int_set to empty.

[0172] S8102. Select appropriate segmentation points based on the distribution of characteristic quantities used for grid division in the current cell, such as selecting the 50% quantile of the distribution as a threshold to perform grid division.

[0173] S8103, using S82 to S86 to learn the relationship model between the air interface measurement value SINR and the scheduling MCS under the given target BLER interval [0.08, 0.12] at the grid level, is recorded as:

[0174]

[0175] The relationship model between the air interface measurement SINR and the outer loop ΔSINR is as follows:

[0176]

[0177] Where [0.08, 0.12] is the target BLER interval, Indicates the combination of the air interface measurement value SINR and the scheduling MCS that meets the target BLER interval; Indicates the combination of the air interface measurement value SINR and the outer loop value that meets the target BLER interval; BLER(·) represents the BLER under the corresponding combination; Represents the relationship model between the air interface measurement value SINR corresponding to the j-th grid and the scheduling MCS; It represents the relationship model between the air interface measurement value SINR and the outer loop value corresponding to the j-th grid; j = 1, ... 64; N represents the number of MCSs that the system can support.

[0178] S811, apply online model.

[0179] S8111. Determine the grid to which the user belongs based on the user's corresponding feature data and the grid division strategy, and obtain the corresponding grid model. and model

[0180] S8112, according to the air interface measurement SINR0 corresponding to the user, the model Determine the corresponding MCS, obtain the median of the corresponding SINR in the target BLER interval from the MCS and the system reference BLER curve, record it as SINR1, and calculate the outer loop value D_SINR0=SINR1-SINR0 from SINR0 and SINR1.

[0181] S8113, the air interface corresponding to the user measures SINR, and the model The corresponding outer loop value is obtained and recorded as: D_SINR1.

[0182] S8114. Calculate the outer loop initialized for the user as follows:

[0183] D_SINR_Init=min{max{(D_SINR0+D_SINR1) / 2,-V_limt},V_limt}

[0184] Among them, V_limt is the protection limit value, its value is greater than 0, and the default value is 10.

[0185] S8115. Determine whether the model update flag Model_change_index is true. If so, initialize AI_ΔSINR_int_set to empty, Model_change_index = 0, and record the corresponding demodulation information ACK / NACK (not recorded for conservative scheduling) when the user is scheduled using this D_SINR_Init outer loop. Otherwise, record the corresponding demodulation information ACK / NACK (not recorded for conservative scheduling) when the user is scheduled using this D_SINR_Init outer loop and record it in AI_ΔSINR_int_set.

[0186] S812. Determine whether AI_ΔSINR_int_set≥Th_A / N_NUM holds.

[0187] If so, the NACK ratio is calculated based on the elements in AI_ΔSINR_int_set as NACK_R. The model adjustment amount is calculated based on the following relationship:

[0188]

[0189] Among them, BLER_tart is the set target BLER, and λ is the model adjustment step size.

[0190] For the model The channel quality dimension in the model is updated using the following formula: SINR i =max{min(SINR i -δ,SINR i +β),SINRi -β}, and get the new model:

[0191] For the model The channel quality dimension in the model is updated using the following formula:

[0192] Get the new model: Set Model_change_index=1 and return to execute S811; otherwise, the model is not updated and return to execute S811.

[0193] In one embodiment, a method for determining an outer loop value by initializing an outer loop model offline is provided. The method for determining an outer loop value by initializing an outer loop model offline mainly comprises the following steps:

[0194] S91. Divide the grid into grids based on the collected characteristic data of users in the network.

[0195] Specifically, the user's characteristic data includes user transmission mode: single port, closed-loop multiplexing RI=1, closed-loop multiplexing RI=2; interference level; scheduling MCS; number of allocated RBs; air interface measurement SINR; space division label; number of space division streams; maximum supported modulation order (64QAM, 256QAM); corresponding demodulation ACK / NACK; and outer loop information.

[0196] Without loss of generality, in the non-space division case: the three transmission modes (single port, closed-loop multiplexing RI=1, closed-loop multiplexing RI=2), two interference levels (high interference, low interference), the maximum supported modulation order (64QAM, 256QAM), and the number of RBs (large and small RBs) are divided into 24 grids; in the space division case: the five types of space division streams (2, 3, 4, 5, and greater than or equal to 6), two interference levels (high interference, low interference), the maximum supported modulation order (64QAM, 256QAM), and the number of RBs (large and small RBs) are divided into 40 grids. The grid-level division can also include user chip type information, demodulation algorithm information, user beam position information, etc.

[0197] S92: Divide each grid into grids according to the scheduled MCS and the corresponding air interface measured channel quality value SINR.

[0198] For example, MCS is divided into two dimensions with order 1 as the granularity and SINR value ...

[0199] S93. Determine whether it is space-divided according to the user's new transmission characteristic data. If it is not space-divided, find the grid to which the user data belongs according to the transmission mode, interference level, supported maximum modulation order configuration information, and the number of scheduled RBs; and deliver the corresponding ACK / NACK corresponding to the actual demodulation and the outer loop information to the corresponding grid of the corresponding grid according to the scheduled MCS and air interface channel quality SINR, and record the number of ACK / NACKs (two are recorded separately) and the outer loop value in each grid; if it is space-divided, find the grid to which the user data belongs according to the number of space-divided flows, interference level, supported maximum modulation order configuration information, and the number of scheduled RBs; and deliver the corresponding ACK / NACK information corresponding to the actual demodulation and the outer loop information to the corresponding grid of the corresponding grid according to the scheduled MCS and air interface channel quality SINR, and record the number of ACK / NACKs (two are recorded separately) and the outer loop value in each grid.

[0200] S94. Calculate the BLER of each grid and the outer ring value corresponding to each grid based on the number of ACK / NACK counted for each grid in each grid. If the number of samples of the corresponding grid is lower than a certain threshold, the grid is marked as invalid and recorded as NULL.

[0201] S95 : learning a demodulation curve according to the BLER of each grid in the grid and the corresponding outer loop value.

[0202] The specific method is to find the grid whose corresponding BLER meets the set BLER range for each MCS in the grid, thereby determining the air interface measurement value SINR corresponding to the grid and the outer loop value corresponding to the grid; and initialize the model update flag Model_change_index = 0; and initialize the model performance evaluation set AI_ΔSINR_int_set to empty.

[0203] S951. If the number of grids that meet the set BLER interval for an MCS is greater than one, calculate a weight factor for each grid. The calculation method is: the ratio of the number of samples (total number of ACK / NACKs) contained in the grid that meets the condition to the sum of the number of samples of all grids that meet the condition.

[0204] S952: Perform weighted averaging of the air interface measurement SINR values ​​corresponding to each grid according to the weights, and perform weighted averaging of the outer ring values ​​corresponding to each grid;

[0205] S953: Determine the scheduling MCS and the corresponding outer loop value that meet the target BLER interval under the air interface measurement value SINR.

[0206] S954. If the grid that meets the set BLER interval is not met for an MCS, find the SINR interval corresponding to the target BLER interval of this MCS on the system reference demodulation curve (usually a simulated curve), and then take the median of the SINR interval minus the system default initialization outer loop to obtain the air interface measurement value SINR corresponding to the MCS. The default initialization outer loop is its corresponding outer loop.

[0207] S955. Through the above process, the scheduling MCS that satisfies the target BLER interval and the outer loop value corresponding to each measured air interface measurement value SINR can be determined, that is, the learned model is:

[0208] The relationship model between the air interface measurement value SINR and the scheduling MCS in the target BLER interval is recorded as:

[0209]

[0210] And, set the relationship model between the air interface measurement value SINR in the target BLER interval and the outer loop ΔSINR:

[0211]

[0212] Where [0.08, 0.12] is the target BLER interval, Indicates the combination of the air interface measurement value SINR and the scheduling MCS that meets the target BLER interval; Indicates the combination of the air interface measurement value SINR and the outer loop value that meets the target BLER interval; BLER(·) represents the BLER under the corresponding combination; Represents the relationship model between the air interface measurement value SINR corresponding to the j-th grid learned offline and the scheduling MCS; represents the relationship model between the air interface measurement SINR corresponding to the j-th grid and the outer loop value learned offline; j = 1, ... 64; N represents the number of MCSs supported by the system;

[0213] S96. Before the model is applied, the relationship between SE and air interface measurement value SINR in the corresponding network is:

[0214]

[0215] The proportion of the statistical user-level BLER falling within the target BLER range is recorded as R0; where K0 is the number of sample points, and Set_p0 represents the network performance set before applying the model.

[0216] S97, Model Application

[0217] S971, according to the user's corresponding feature data and the strategy of dividing the grid, determine the grid to which the user belongs, and obtain the corresponding MCS, the median of the SINR corresponding to the target BLER interval is obtained by MCS and the system reference BLER curve, recorded as SINR1, and the outer loop value D_SINR0 = SINR1-SINR0 is calculated by SINR0 and SINR1

[0218] S973, the current air interface measurement value SINR corresponding to the user is calculated through the model Get the corresponding outer loop value and record it as: D_SINR1;

[0219] S974. Calculate the outer loop initialized for the user:

[0220] D_SINR_Init=min{max{(D_SINR0+D_SINR1) / 2,-V_limt},V_limt}

[0221] Among them, V_limt is the protection limit value, its value is greater than 0, and the default value is 10;

[0222] S975. Determine whether the model update flag Model_change_index is true. If so, initialize AI_ΔSINR_int_set to empty, Model_change_index = 0 to record that when the user is scheduled using the D_SINR_Init outer loop, the corresponding demodulation information ACK / NACK (not recorded here for conservative scheduling) is counted in AI_ΔSINR_int_set.

[0223] Otherwise, record the demodulation information ACK / NACK (not recorded here for conservative scheduling) when the user is scheduled with this D_SINR_Init outer loop and include it in AI_ΔSINR_int_set

[0224] S98. After applying the model, the relationship set between the SE of the corresponding network and the air interface measurement value SINR is statistically calculated as follows:

[0225]

[0226] The proportion of the user-level BLER that falls within the target BLER range is recorded as R1; K1 is the number of sample points, and Set_p1 represents the network performance set before applying the model;

[0227] S99: Determine whether the air interface channel quality SINRs in Set_p0 and Set_p1 have an intersection.

[0228] S991. If there is an intersection, calculate the performance difference corresponding to the intersection and record it as: {D_SEi |i=1,2,…M}; where D_SE i represents the difference between the SE after applying the model and the SE before applying the model corresponding to the i-th element in the above intersection;

[0229] S992, calculate {D_SE i The proportion R2 of elements with a value of |i=1,2,…M} less than 0, and the mean value D_SE_mean of the elements in the set.

[0230] S993 . If D_SE_mean ≥ Th0 and R2 ≥ Th1 and R0 ≥ R1 or D_SE_mean < Th0 , go to S910 .

[0231] Otherwise, continue to use the current model to initialize the outer loop of the newly connected user.

[0232] S994. If there is no intersection, and if R0 ≥ R1, go to S910.

[0233] Otherwise, continue to use the current model to initialize the outer loop of the newly connected user.

[0234] S910: Start model correction.

[0235] Determine whether AI_ΔSINR_int_set ≥ Th_A / N_NUM. If so, calculate the NACK ratio based on the elements in AI_ΔSINR_int_set as NACK_R. Calculate the model adjustment amount based on the following relationship:

[0236]

[0237] Among them, BLER_tart is the set target BLER, and λ is the model adjustment step size.

[0238] For the model The channel quality dimension in the model is updated using the following formula:

[0239] SINR i =max{min(SINR i -δ,SINR i +β),SINR i -β}, and get the new model:

[0240] For the model The channel quality dimension in the model is updated using the following formula:

[0241]

[0242] Get the new model: Set Model_change_index = 1 and go to S97.

[0243] Otherwise, the model is not updated and the process goes to S97.

[0244] In one embodiment, a method of revising an online model is provided.

[0245] S101. Divide the grid into sections according to historical feature data.

[0246] The grids are divided according to the historical feature data of users in the network collected.

[0247] Specifically, historical feature data includes: user transmission mode, scheduled air interface measurement SINR, air division label; number of air division streams; maximum supported modulation order (64QAM, 256QAM); corresponding demodulation ACK / NACK; outer loop information, etc. The user transmission mode includes: single port, closed loop multiplexing RI=1, closed loop multiplexing RI=2.

[0248] In the case of non-space division: divided into 24 grids according to three transmission modes (single port, closed-loop multiplexing RI=1, closed-loop multiplexing RI=2), two interference levels (high interference, low interference), the maximum supported modulation order (64QAM, 256QAM), and the number of RBs (large and small RBs). In the case of space division: divided into 40 grids according to the number of space division streams (2, 3, 4, 5, and greater than or equal to 6, divided into 5 types), two interference levels (high interference, low interference), the maximum supported modulation order (64QAM, 256QAM), and the number of RBs (large and small RBs). Furthermore, the grid-level division can also include the user's chip type information, demodulation algorithm information, user beam position information, etc. It should be noted that the above grid division method is only an example and not a limitation.

[0249] S102: For each grid, divide the grid into grids according to the scheduled MCS and the corresponding demodulated air interface measurement value SINR.

[0250] For example, MCS is divided into two dimensions with order 1 as the granularity and SINR value is divided into two dimensions with order 1 as the horizontal axis and SINR value is divided into two dimensions with 0.5 dB as the vertical axis.

[0251] S103 : Determine the grid to which the data belongs and the grid under the corresponding grid based on the captured new feature data, and deliver the ACK / NACK information corresponding to the actual demodulation and the outer loop information to the corresponding grid of the grid.

[0252] Specifically, determine whether it is space-divided based on the captured new feature data. If it is not space-divided, find the grid to which the user data belongs according to the transmission mode, interference level, supported maximum modulation order configuration information, and the number of scheduled RBs, and deliver the corresponding ACK / NACK information corresponding to the actual demodulation and the outer loop information to the corresponding grid of the grid based on the scheduled MCS and the air interface measurement value SINR. Record the number of ACK / NACK information (two respectively) and the outer loop value in each grid.

[0253] If space division is used, find the grid to which the user data belongs based on the number of space division flows, interference level, supported maximum modulation order configuration information, and the number of scheduled RBs; and deliver the corresponding ACK / NACK information corresponding to the actual demodulation and the outer loop information to the corresponding grid of the grid based on the scheduled MCS and air interface measurement value SINR. Record the number of ACK / NACKs (two respectively) and the outer loop value in each grid.

[0254] S104: Statistically calculate the BLER of each grid and the outer ring value.

[0255] Specifically, the BLER corresponding to each grid and the outer ring value corresponding to each grid are calculated based on the number of ACK / NACK statistics of each grid in each grid. If the corresponding grid sample volume is lower than a certain threshold, the grid is marked as invalid and recorded as NULL.

[0256] S105 : Determine a correspondence relationship model between the air interface measurement value SINR and the scheduling MCS and a correspondence relationship model between the air interface measurement value SINR and the outer loop value according to the BLER and the outer loop value in each grid in the grid.

[0257] The specific method is to search for a grid whose corresponding BLER satisfies a set BLER range for each MCS in the grid, thereby determining the air interface measurement value SINR corresponding to the grid and the outer ring value corresponding to the grid.

[0258] S1051: If the number of grids that meet the set BLER interval for an MCS is greater than one, calculate a weight factor for each grid.

[0259] The calculation method is: the ratio of the number of samples contained in the grid that meets the conditions (the total number of ACK / NACKs) to the sum of the number of samples of all grids that meet the conditions.

[0260] S1052: Perform weighted averaging of the air interface measurement SINR values ​​corresponding to each grid according to the weight factor, and perform weighted averaging of the outer ring values ​​corresponding to each grid.

[0261] S1053: Determine the scheduling MCS and the corresponding outer loop value that meet the target BLER interval under the air interface measurement SINR.

[0262] S1054. If an MCS does not meet the set BLER interval grid, find the interval of demodulation SINR values ​​corresponding to the target BLER interval of this MCS on the system reference demodulation curve (usually a simulated curve), and then take the median of the interval minus the system default initialization outer loop to obtain the air interface measured SINR value corresponding to the MCS, and the default initialization outer loop is its corresponding outer loop.

[0263] S1055. Through the above process, the scheduling MCS that satisfies the target BLER interval and the outer loop value corresponding to each air interface measurement SINR can be determined. That is, the learned model is:

[0264] The relationship model between the air interface measurement value SINR and the scheduling MCS in the target BLER interval is recorded as:

[0265]

[0266] The relationship model between the air interface measurement value SINR and the outer loop value ΔSINR in the target BLER interval is recorded as:

[0267]

[0268] Among them, [0.08, 0.12] is the target BLER interval, Indicates the combination of air interface measurement value and scheduling MCS that meets the target BLER interval; Indicates the combination of air interface measurement value and outer loop value that meets the target BLER interval; BLER(·) represents the BLER under the corresponding combination; Represents the relationship model between the air interface measurement value SINR corresponding to the j-th grid learned offline and the scheduling MCS; It represents the relationship model between the air interface measurement SINR and the outer loop value corresponding to the j-th grid learned offline; j = 1, ... 64; N represents the number of MCSs supported by the system.

[0269] S106. Before applying the outer loop initialization model, the relationship set between the SE and the average air interface measurement SINR in the corresponding network is:

[0270]

[0271] The proportion of the statistical user-level BLER falling within the target BLER range is recorded as R0; where K0 is the number of sample points, and Set_p0 represents the network performance set before applying the model.

[0272] S107: Initialize the outer loop model application.

[0273] S1071. Determine the grid to which the user belongs based on the user's corresponding feature data and the grid division strategy, and obtain the corresponding grid model. and model

[0274] S1072, according to the current air interface measurement value SINR0 corresponding to the user, the model Determine the corresponding MCS, obtain the median of the corresponding SINR in the target BLER interval from the MCS and the system reference BLER curve, record it as SINR1, and calculate the outer loop value D_SINR0=SINR1-SINR0 from SINR0 and SINR1.

[0275] S1073, the current air interface measurement value SINR corresponding to the user is calculated through the model The corresponding outer loop value is obtained and recorded as: D_SINR1.

[0276] S1074. Calculate the outer loop value initialized to the user:

[0277] D_SINR_Init=min{max{(D_SINR0+D_SINR1) / 2,-V_limt},V_limt}

[0278] Among them, V_limt is the protection limit value, its value is greater than 0, and the default value is 10.

[0279] S1010. After applying the model, the relationship set between the SE of the corresponding network and the air interface measurement value SINR is statistically calculated as follows:

[0280]

[0281] The proportion of the statistical user-level BLER falling within the target BLER range is recorded as R1; K1 is the number of sample points, and Set_p1 represents the network performance set before applying the model.

[0282] S109: Determine whether the air interface channel quality SINRs in Set_p0 and Set_p1 have an intersection.

[0283] S1091. If there is an intersection, calculate the performance difference corresponding to the intersection and record it as:

[0284] {D_SE i |i=1,2,…M}

[0285] where D_SE i It represents the difference between the SE after applying the model and the SE before applying the model corresponding to the i-th element in the above intersection.

[0286] S10911、Calculate {D_SE i The proportion R2 of elements with a value of |i=1,2,…M} less than 0, and the mean value D_SE_mean of the elements in the set.

[0287] S10912: If D_SE_mean≥Th0 and R2≥Th1 and R0≥R1 or D_SE_mean<Th0, then execute S1010. Otherwise, continue to use the current model to initialize the outer loop of the newly connected user.

[0288] S1092: If there is no intersection and R0≥R1, execute S1010. Otherwise, continue to use the current model to initialize the outer loop of the newly connected user.

[0289] S1010, stop using the current model and start online learning

[0290] S10101. Collect data related to user level and scheduling information online, including user transmission mode: single port, closed-loop multiplexing RI=1, closed-loop multiplexing RI=2; interference level; scheduling MCS; number of allocated RBs; air interface measurement SINR; space division label; number of space division streams; maximum supported modulation order (64QAM, 256QAM); corresponding demodulation ACK / NACK; outer loop information; and initialize the model update flag Model_change_index=0; and initially set the model performance evaluation set AI_ΔSINR_int_set to empty.

[0291] S10102. Select appropriate segmentation points based on the distribution of characteristic quantities used for grid division in the current cell, such as selecting the 50% quantile of the distribution as a threshold to perform grid division.

[0292] S10103. Using S102 to S106, learn the relationship model between the air interface measurement value SINR and the scheduling MCS under the given target BLER interval [0.08, 0.12] at the grid level:

[0293]

[0294] The relationship model between the air interface measurement SINR and the outer loop ΔSINR is as follows:

[0295]

[0296] Where [0.08, 0.12] is the target BLER interval, Indicates the combination of air interface measurement value and scheduling MCS that meets the target BLER interval; Indicates the combination of air interface measurement value and outer loop value that meets the target BLER interval; BLER(·) represents the corresponding combination Represents the relationship model between the air interface measurement value corresponding to the j-th grid and the scheduling MCS; Represents the relationship model between the air interface measurement value and the outer loop value corresponding to the j-th grid; j = 1, ... 64; N represents the number of MCSs that the system can support.

[0297] S1011. Apply online models.

[0298] S10111. Determine the grid to which the user belongs based on the user's corresponding feature data and the grid division strategy, and obtain the corresponding grid model. and model

[0299] S10112, according to the current air interface measurement value SINR0 corresponding to the user, the model Determine the corresponding MCS, obtain the median of the corresponding SINR in the target BLER interval from the MCS and the system reference BLER curve, record it as SINR1, and calculate the outer loop value D_SINR0=SINR1-SINR0 from SINR0 and SINR1.

[0300] S10113, the current air interface measurement value SINR corresponding to the user is calculated through the model The corresponding outer loop value is obtained and recorded as: D_SINR1.

[0301] S10114. Calculate the outer loop initialized for the user as follows:

[0302] D_SINR_Init=min{max{(D_SINR0+D_SINR1) / 2,-V_limt},V_limt}

[0303] Among them, V_limt is the protection limit value, its value is greater than 0, and the default value is 10;

[0304]

[0305] The proportion of the user-level BLER that falls within the target BLER range is recorded as R1; K1 is the number of sample points, and Set_p1 represents the network performance set before applying the model;

[0306] S1013: Determine whether the air interface channel quality SINRs in Set_p0 and Set_p1 have an intersection.

[0307] S10131. If there is an intersection.

[0308] The performance difference corresponding to the intersection calculation is recorded as: {D_SEi |i=1,2,…M}

[0309] Among them, D_SE i It represents the difference between the SE after applying the model and the SE before applying the model corresponding to the i-th element in the above intersection.

[0310] S101311、Calculate {D_SE i The proportion R2 of elements with a value of |i=1,2,…M} less than 0, and the mean value D_SE_mean of the elements in the set.

[0311] S101312. If D_SE_mean ≥ Th0 and R2 ≥ Th1 and R0 ≥ R1; or D_SE_mean < Th0, use the default outer loop of the cell and stop applying the model; otherwise, continue to use the current model to initialize the outer loop of the newly accessed user.

[0312] S10132. If there is no intersection, and if R0≥R1, use the default outer loop of the cell and stop applying the model; otherwise, continue to use the current model to initialize the outer loop of the newly accessed user.

[0313] In one embodiment, the present application provides a method for determining an outer loop value. Figure 10 This is a structural diagram of an outer loop value determination device provided in an embodiment of the present application. This embodiment is applicable to the case of adaptive coding modulation technology, and the device can be implemented in software and / or hardware.

[0314] like Figure 1 As shown, the outer loop value determination method provided in the embodiment of the present application mainly includes a step model determination module 101 and an outer loop value determination module 102.

[0315] A model determination module 101 is configured to determine a pre-trained outer loop initialization model based on current feature data of a user device;

[0316] The outer loop value determination module 102 is configured to determine an initialization outer loop value of the user equipment based on a current air interface measurement value of the user equipment and the outer loop initialization model, where the outer loop initialization model includes a first outer loop model and a second outer loop model.

[0317] In one embodiment, the model determination module 101 is configured to determine the grid to which the user equipment belongs based on current feature data of the user equipment and a grid division strategy; and obtain a first outer ring model and a second outer ring model corresponding to the grid to which the user equipment belongs.

[0318] In one embodiment, the outer loop value determination module 102 is configured to determine a first outer loop value based on the current air interface measurement value and the first outer loop model; determine a second outer loop value based on the current air interface measurement value and the second outer loop model; and determine an initialization outer loop value based on the first outer loop value and the second outer loop value.

[0319] In one embodiment, determining the first outer loop value based on the current air interface measurement value and the first outer loop model includes: determining the modulation and coding strategy MCS corresponding to the current air interface measurement value based on the current air interface measurement value and the first outer loop model; wherein the first outer loop model is a correspondence model between the air interface measurement value and the MCS under the set target block error rate BLER interval; obtaining the system mapping SINR corresponding to the target BLER based on the determined MCS and the reference demodulation curve; and determining the difference between the system mapping SINR and the current air interface measurement value as the first outer loop value.

[0320] In one embodiment, determining the second outer loop value based on the current air interface measurement value and the second outer loop model includes: obtaining the second outer loop value through the second outer loop model based on the current air interface measurement value; wherein, the second outer loop model is a correspondence model between the air interface measurement value and the outer loop value under the set target BLER interval.

[0321] In one embodiment, the apparatus further includes: a model training module configured to train an outer-loop initialization model based on historical feature data of the user device before determining a pre-trained outer-loop initialization model based on current feature data of the user device.

[0322] In one embodiment, the model training module is configured to perform grid division based on the historical feature data; determine the number of confirmation response / negative response ACK / NACK information and the outer loop value in each grid; calculate the block error rate of each grid based on the number of ACK / NACK information in each grid; and determine the outer loop initialization model based on the block error rate and outer loop value of each grid.

[0323] In one embodiment, performing grid division based on the historical feature data includes: performing grid division on each cell in the network based on the historical feature data; and performing grid division on each grid according to the scheduled MCS and the corresponding air interface measurement value.

[0324] In one embodiment, determining the number of ACK / NACK information in each grid includes: obtaining confirmation response / negative response ACK / NACK information corresponding to demodulation; delivering the ACK / NACK information and corresponding outer loop information to the corresponding grid; and counting the number of ACK / NACK information and the outer loop value in each grid.

[0325] In one embodiment, the outer loop initialization model is determined based on the block error rate and outer loop value of each grid, including: in each grid, determining the MCS corresponding to the air interface measurement value and the outer loop value corresponding to the air interface measurement value under the set target block error rate BLER interval based on the block error rate of each grid; determining the first outer loop model based on the correspondence between the air interface measurement value and the MCS; and determining the second outer loop model based on the correspondence between the air interface measurement value and the outer loop value.

[0326] In one embodiment, the method of determining the MCS corresponding to the air interface measurement value under the set target block error rate BLER interval based on the block error rate of each grid and the air interface measurement value and the corresponding outer loop value includes: for an MCS, if the number of grids that meet the set block error rate interval is greater than a preset value, calculating the weight factor of each grid; performing weighted averaging calculation on the air interface measurement value corresponding to each grid according to the weight factor of each grid to obtain the air interface measurement value corresponding to the MCS; performing weighted averaging calculation on the outer loop value corresponding to each grid according to the weight factor of each grid to obtain the outer loop value corresponding to the MCS; and determining the outer loop value corresponding to the air interface measurement value based on the air interface measurement value corresponding to the MCS and the outer loop value corresponding to the MCS.

[0327] In one embodiment, the method of determining the MCS corresponding to the air interface measurement value under the set target block error rate BLER interval based on the block error rate of each grid and the air interface measurement value and the corresponding outer loop value includes: for an MCS, if the number of grids that meet the set block error rate interval is less than a preset value, looking up the air interface measurement value interval corresponding to the target block error rate interval corresponding to the MCS on the reference demodulation curve; taking the difference between the median value in the air interface measurement value interval and the system default initialization outer loop value as the air interface measurement value corresponding to the MCS; and taking the system default initialization outer loop value as the outer loop value corresponding to the air interface measurement value.

[0328] In one embodiment, the device also includes: an online model learning module, which is configured to obtain a first network performance indicator before applying the model and a second network performance indicator after using the outer loop initialization model after determining the initialization outer loop value of the user equipment based on the current air interface measurement value of the user equipment and the outer loop initialization model; when the first network performance indicator and the second network performance indicator meet preset conditions, start online learning to obtain the outer loop online model; and use the outer loop online model to determine a new initialization outer loop value for the user equipment.

[0329] In one embodiment, the device also includes: an online model adjustment module, which is configured to correct the outer loop model of the corresponding grid based on the ACK / NACK information after the user in the grid uses the outer loop model for the first time; the outer loop model includes an outer loop initialization model or an outer loop online model.

[0330] In one embodiment, adjusting the online model based on ACK / NACK information after using the online model includes: calculating, for each grid, a proportion of negative acknowledgements (NACKs) in the ACK / NACK information; and reducing an outer loop value of the outer loop model according to a corresponding strategy when the proportion of NACKs is greater than a target value;

[0331] When the proportion of NACKs is less than the target value, the outer loop value of the outer loop model is increased according to a corresponding strategy.

[0332] The outer loop value determination device provided in this embodiment can execute the outer loop value determination method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the method. For technical details not fully described in this embodiment, please refer to the outer loop value determination method provided in any embodiment of the present invention.

[0333] It is worth noting that in the embodiment of the above-mentioned outer loop value determination device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application.

[0334] The embodiment of the present application also provides a device, Figure 11 This is a schematic diagram of the structure of a device provided by this application, such as Figure 11 As shown, the device includes a processor 111, a memory 112, an input device 113, and an output device 114; the number of processors 111 in the device can be one or more. Figure 11 In the embodiment, a processor 111 is used as an example; the processor 111, the memory 112, the input device 113 and the output device 114 in the device can be connected by a bus or other means. Figure 11 The bus connection is taken as an example.

[0335] The memory 112, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the outer loop value determination method in the embodiments of the present application (for example, the model determination module 101 and the outer loop value determination module 102 in the outer loop value determination device). The processor 111 executes the various functional applications and data processing of the device by running the software programs, instructions, and modules stored in the memory 112, that is, implementing any method provided in the embodiments of the present application.

[0336] The memory 112 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system and at least one application required for a function; the data storage area may store data created based on the use of the device, etc. In addition, the memory 112 may include a high-speed random access memory and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 112 may further include a memory remotely located relative to the processor 111, and these remote memories may be connected to the device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0337] The input device 113 may be used to receive input digital or character information and generate key signal input related to user settings and function control of the device. The output device 114 may include a display device such as a display screen.

[0338] An embodiment of the present application further provides a storage medium containing computer-executable instructions. When the computer-executable instructions are executed by a computer processor, the computer-executable instructions are used to perform a method for determining an outer loop value, including:

[0339] Determining a pre-trained outer loop initialization model based on current feature data of the user device;

[0340] An initialization outer loop value of the user equipment is determined based on a current air interface measurement value of the user equipment and the outer loop initialization model.

[0341] Of course, the storage medium containing computer-executable instructions provided in an embodiment of the present application is not limited to the operations described above, and the computer-executable instructions can also execute the relevant operations in the outer loop value determination method provided in any embodiment of the present application.

[0342] Through the above description of the implementation methods, those skilled in the art can clearly understand that the present application can be implemented with the help of software and necessary general-purpose hardware, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as a computer's floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0343] The above description is merely an exemplary embodiment of the present application and is not intended to limit the scope of protection of the present application.

[0344] It will be appreciated by those skilled in the art that the term user terminal covers any suitable type of wireless user equipment, such as a mobile phone, a portable data processing device, a portable web browser or a vehicle-mounted mobile station.

[0345] In general, various embodiments of the present application may be implemented in hardware or dedicated circuits, software, logic, or any combination thereof. For example, some aspects may be implemented in hardware, while other aspects may be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device, although the present application is not limited thereto.

[0346] Embodiments of the present application may be implemented by executing computer program instructions by a data processor of a mobile device, for example, in a processor entity, or by hardware, or by a combination of software and hardware. The computer program instructions may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages.

[0347] The block diagram of any logical flow in the accompanying drawings of the present application can represent program steps, or can represent interconnected logical circuits, modules and functions, or can represent a combination of program steps and logical circuits, modules and functions. The computer program can be stored on a memory. The memory can have any type suitable for the local technical environment and can be implemented using any suitable data storage technology, such as but not limited to read-only memory (ROM), random access memory (RAM), optical memory device and system (digital versatile disc DVD or CD optical disc) etc. Computer-readable media can include non-transient storage media. The data processor can be any type suitable for the local technical environment, such as but not limited to a general-purpose computer, a special-purpose computer, a microprocessor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a programmable logic device (FPGA) and a processor based on a multi-core processor architecture.

[0348] The above description of exemplary embodiments of the present application has been provided by way of exemplary and non-limiting examples. However, various modifications and adaptations to the above embodiments will be apparent to those skilled in the art, when considered in conjunction with the accompanying drawings and claims, without departing from the scope of the present invention. Therefore, the proper scope of the present invention will be determined by reference to the claims.

Claims

1. A method for determining an outer loop value, characterized in that: include: Determine a pre-trained outer loop initialization model based on current feature data of the user equipment, wherein the outer loop initialization model includes a first outer loop model and a second outer loop model, the first outer loop model is a corresponding relationship model between an air interface measurement value and an MCS under a set target block error rate BLER interval, and the second outer loop model is a corresponding relationship model between an air interface measurement value and an outer loop value under a set target BLER interval; An initialization outer loop value of the user equipment is determined based on a current air interface measurement value of the user equipment and the outer loop initialization model.

2. The method according to claim 1, characterized in that The determining of the pre-trained outer loop initialization model based on the current feature data of the user equipment includes: Determine the grid to which the user equipment belongs based on the current feature data of the user equipment and the grid division strategy; Obtain a first outer ring model and a second outer ring model corresponding to the grid to which the user equipment belongs.

3. The method according to claim 2, characterized in that The determining the initialization outer loop value of the user equipment based on the current air interface measurement value of the user equipment and the outer loop initialization model includes: Determine a first outer loop value based on the current air interface measurement value and the first outer loop model; Determine a second outer loop value based on the current air interface measurement value and the second outer loop model; An initialization outer loop value is determined based on the first outer loop value and the second outer loop value.

4. The method according to claim 3, characterized in that The determining the first outer loop value based on the current air interface measurement value and the first outer loop model includes: Determine a modulation and coding strategy MCS corresponding to the current air interface measurement value based on the current air interface measurement value and the first outer loop model; Acquire a system mapping SINR corresponding to the target BLER based on the determined MCS and a reference demodulation curve; The difference between the system mapping SINR and the current air interface measurement value is determined as a first outer loop value.

5. The method according to claim 3, characterized in that: The determining the second outer loop value based on the current air interface measurement value and the second outer loop model includes: Based on the current air interface measurement value, a second outer loop value is acquired through the second outer loop model.

6. The method according to claim 1, characterized in that Before determining the pre-trained outer loop initialization model based on the current feature data of the user equipment, the method further includes: An outer loop initialization model is trained based on the historical feature data of the user equipment.

7. The method according to claim 6, characterized in that The training of the outer loop initialization model based on the historical feature data of the user equipment includes: Performing grid division based on the historical feature data; Determine the number of ACK / NACK messages and the outer loop value in each grid; Calculate the BLER of each grid based on the number of ACK / NACK information in each grid; An outer loop initialization model is determined based on the BLER of each grid and the outer loop value.

8. The method according to claim 7, characterized in that The grid division based on the historical feature data includes: Dividing each cell in the network into grids based on the historical feature data; Each grid is divided into grids according to the scheduled MCS and the corresponding air interface measurement values.

9. The method according to claim 7, characterized in that: The determining the number of ACK / NACK information and the outer loop value in each grid includes: Obtain ACK / NACK information corresponding to demodulation; Delivering the ACK / NACK information and the corresponding outer loop information to the corresponding grid; Count the number of ACK / NACK information and the outer loop value in each grid.

10. The method according to claim 7, characterized in that The determining of the outer loop initialization model based on the BLER and the outer loop value of each grid includes: In each grid, based on the BLER of each grid, the MCS corresponding to the air interface measurement value and the outer loop value corresponding to the air interface measurement value under the set target BLER interval are determined; Determine a first outer loop model based on a correspondence between the air interface measurement value and the MCS; A second outer loop model is determined based on the correspondence between the air interface measurement value and the outer loop value.

11. The method according to claim 10, characterized in that The determining, based on the BLER of each grid, of the MCS corresponding to the air interface measurement value and the outer loop value corresponding to the air interface measurement value in the target BLER interval includes: For an MCS, if the number of grids that meet the set target BLER interval is greater than a preset value, a weight factor of each grid is calculated according to the number of ACK / NACK information in each grid; Performing weighted average calculation on the air interface measurement values ​​corresponding to each grid according to the weight factor of each grid to obtain the air interface measurement value corresponding to the MCS; Performing weighted average calculation on the outer ring values ​​corresponding to each grid according to the weight factor of each grid to obtain the outer ring value corresponding to the MCS; The outer loop value corresponding to the air interface measurement value is determined according to the air interface measurement value corresponding to the MCS and the outer loop value corresponding to the MCS.

12. The method according to claim 10, characterized in that The determining, based on the BLER of each grid, of the MCS corresponding to the air interface measurement value and the outer loop value corresponding to the air interface measurement value in the target BLER interval includes: For an MCS, if the number of grids that meet the set target BLER interval is less than a preset value, find the air interface measurement value interval corresponding to the target BLER interval corresponding to the MCS on the reference demodulation curve; The difference between the median value in the searched air interface measurement value interval and the system default initialization outer loop value is used as the air interface measurement value corresponding to the MCS; The system default initialized outer loop value is used as the outer loop value corresponding to the air interface measurement value.

13. The method according to claim 1, characterized in that After determining the initialization outer loop value of the user equipment based on the current air interface measurement value of the user equipment and the outer loop initialization model, the method further includes: Obtaining a first network performance indicator before applying the model and a second network performance indicator after initializing the model using the outer loop; When the first network performance indicator and the second network performance indicator meet preset conditions, starting online learning to obtain an outer loop online model; The outer loop online model is used to determine a new initialized outer loop value for the user equipment.

14. The method according to claim 1 or 13, characterized in that: The method further comprises: The outer loop model of the corresponding grid is modified based on ACK / NACK information after the user in the grid uses the outer loop model for the first time; the outer loop model includes a pre-trained outer loop initialization model or an outer loop online model.

15. The method according to claim 14, characterized in that The correcting the outer loop model of the corresponding grid based on the ACK / NACK information after the user in the grid uses the outer loop model for the first time includes: For each grid, calculate the proportion of negative response NACK in the ACK / NACK information; When the proportion of NACKs is greater than a target value, reducing the outer loop value of the outer loop model according to a corresponding strategy; When the proportion of NACKs is less than the target value, the outer loop value of the outer loop model is increased according to a corresponding strategy.

16. An outer loop value determination device, characterized in that: include: A model determination module is configured to determine a pre-trained outer loop initialization model based on current feature data of a user equipment, wherein the outer loop initialization model includes a first outer loop model and a second outer loop model, the first outer loop model is a correspondence relationship model between an air interface measurement value and an MCS under a set target block error rate BLER interval, and the second outer loop model is a correspondence relationship model between an air interface measurement value and an outer loop value under a set target BLER interval; The outer loop value determination module is configured to determine the initialization outer loop value of the user equipment based on the current air interface measurement value of the user equipment and the outer loop initialization model.

17. A device, characterized in that include: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 15.

18. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 15 is implemented.

Images