Resource allocation method and apparatus, electronic device, and computer program product

Abstract

The application relates to the technical field of wireless resource allocation, and provides a resource allocation method and device, electronic equipment and a computer program product. The method comprises the following steps: determining predicted values of various traffic indexes according to cell traffic data; determining fluctuation intervals of the various traffic indexes according to the predicted values of the various traffic indexes and standard deviations; determining a first sparse matrix of outlier samples according to the predicted values of the various traffic indexes and the fluctuation intervals; and recursively adjusting weight factors of various levels of traffic data according to the first sparse matrix, so as to perform resource allocation based on the adjustment result. According to the embodiment of the application, the weight factors are adjusted in a recursive calling mode, wireless resource allocation self-convergence is realized, service user perception is guaranteed, and the allocation efficiency and utilization efficiency of wireless network resources are improved.

Description

Technical Field

[0001] This application relates to the field of wireless resource allocation technology, specifically to a resource allocation method, apparatus, electronic device, and computer program product. Background Technology

[0002] Currently, 5G networks manage various types of data services through the 5QI (5G QoS Identifier) ​​level. Services with similar rate and latency requirements share the 5QI level and are allocated similar scheduling mechanisms and wireless bandwidth resources. In the early stages of service development, 5G cells are generally in a light-load environment, and the baseline 5QI wireless parameter setting strategy can basically meet the needs of various services. However, in the later stages of 5G service development, some areas of the 5G network have experienced heavy loads, and the baseline 5QI wireless parameter setting strategy results in poor user experience for most users in hotspot 5G cells. Therefore, the current baseline 5QI wireless parameter strategy cannot meet and adapt to the differentiated service distribution between different cells. Based on this, how to achieve effective wireless resource allocation is an urgent problem to be solved. Summary of the Invention

[0003] This application provides a resource allocation method, apparatus, electronic device, and computer program product to solve the problem of wireless resource allocation.

[0004] In a first aspect, embodiments of this application provide a resource allocation method, including:

[0005] The predicted values ​​for various call statistics indicators are determined based on the call statistics data of the community;

[0006] The fluctuation range of each of the aforementioned call statistics indicators is determined based on the predicted value and standard deviation of each indicator.

[0007] The first sparse matrix of outlier samples is determined based on the predicted values ​​and fluctuation ranges of the various statistical indicators.

[0008] The weighting factors of each level of speech statistics data are recursively adjusted based on the first sparse matrix, and resources are allocated based on the adjustment results.

[0009] In one embodiment, determining the predicted values ​​of various call statistics indicators based on the call statistics data of the cell includes:

[0010] Determine the compensation factor matrix based on the aforementioned call statistics data;

[0011] The second live network indicator matrix is ​​determined based on the compensation factor matrix and the first live network indicator matrix.

[0012] Based on the second existing network indicator matrix and the fitting method, the predicted values ​​of each of the call statistics indicators are determined.

[0013] In one embodiment, the recursive adjustment of the weighting factors of each level of speech statistics data according to the first sparse matrix includes:

[0014] The adjustment range of each level of call statistics data is determined based on the element values ​​in the first sparse matrix.

[0015] The weighting factors of each level of call statistics data are recursively adjusted according to the aforementioned adjustment range.

[0016] In one embodiment, the resource allocation based on the adjustment result includes:

[0017] If it is determined based on the adjustment results that the recursive call cycle has not been reached, then resource allocation is performed based on the adjustment results.

[0018] If it is determined based on the adjustment results that the recursive call cycle has been reached, then the recursively adjusted weight value is obtained.

[0019] If the weight value reaches the adjustment threshold of the weight factor, then resource allocation is performed based on the adjustment result.

[0020] In one embodiment, determining the first sparse matrix of outlier samples based on the predicted values ​​and fluctuation ranges of each of the statistical indicators includes:

[0021] The second sparse matrix is ​​determined based on the predicted values ​​and fluctuation ranges of each of the aforementioned statistical indicators;

[0022] A one-dimensional matrix of the second sparse matrix is ​​determined, and a first sparse matrix of the outlier samples is determined based on the one-dimensional matrix and the value of the matrix elements.

[0023] In one embodiment, determining the second sparse matrix based on the predicted values ​​and fluctuation ranges of each of the aforementioned statistical indicators includes:

[0024] The element values ​​of the second sparse matrix are determined based on the predicted values ​​and fluctuation ranges of the various statistical indicators.

[0025] The second sparse matrix is ​​determined based on the element values ​​of the second sparse matrix.

[0026] In one embodiment, determining the fluctuation range of each of the call statistics indicators based on the predicted value and standard deviation of each indicator includes:

[0027] Determine the sum and difference of the predicted values ​​and standard deviations of each of the aforementioned statistical indicators;

[0028] The fluctuation range is determined based on the sum and the difference.

[0029] Secondly, embodiments of this application provide a resource allocation device, including:

[0030] The prediction module is used to determine the predicted values ​​of various call statistics indicators based on the call statistics data of the cell.

[0031] The first determining module is used to determine the fluctuation range of each of the call statistics indicators based on the predicted value and standard deviation of each of the call statistics indicators;

[0032] The second determining module is used to determine the first sparse matrix of outlier samples based on the predicted values ​​and fluctuation ranges of each of the aforementioned statistical indicators.

[0033] The resource allocation module is used to recursively adjust the weight factors of each level of the sparse matrix to allocate resources based on the adjustment results.

[0034] Thirdly, embodiments of this application provide an electronic device, including a processor and a memory storing a computer program, wherein the processor executes the program to implement the steps of the resource allocation method described in the first aspect.

[0035] Fourthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the steps of the resource allocation method described in the first aspect.

[0036] The resource allocation method, apparatus, electronic device, and computer program product provided in this application determine the predicted values ​​of various call statistics indicators based on the call statistics data of the cell; determine the fluctuation range of each call statistics indicator based on the predicted values ​​and standard deviations of the indicators; determine a first sparse matrix of outlier samples based on the predicted values ​​and fluctuation ranges of the indicators; and recursively adjust the weight factors of the call statistics data at each level based on the first sparse matrix to allocate resources based on the adjustment results. This application embodiment achieves self-convergence in wireless resource allocation by adjusting the weight factors through recursive calls, ensuring user experience and improving the allocation and utilization efficiency of wireless network resources. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is one of the flowcharts illustrating the resource allocation method provided in the embodiments of this application;

[0039] Figure 2 This is a second flowchart illustrating the resource allocation method provided in the embodiments of this application;

[0040] Figure 3 This is a schematic diagram of the structure of the resource allocation device provided in the embodiments of this application;

[0041] Figure 4 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0043] Figure 1 This is one of the flowcharts illustrating the resource allocation method provided in the embodiments of this application. (Refer to...) Figure 1 This application provides a resource allocation method, which may include:

[0044] Step 100: Determine the predicted values ​​of various call statistics indicators based on the call statistics data of the community;

[0045] It should be noted that the embodiments of this application include four modules: module A is used to implement cell 5QI (5G QoSIdentifier, QoS level identifier in 5G) level service data input; module B is used to implement cell 5QI level service outlier sample calculation; module C is used to implement cell-level service automatic adaptation and adjustment; and module D is used to implement cell-level periodic recursive calls.

[0046] Module A enables the input of 5QI-level service data for the cell. Specifically, it collects hourly call statistics data at the cell level and 5QI level over a period of time (such as the most recent week) across the entire network and inputs it into the module for training and analysis.

[0047] The 5QI levels include 5QI6, 5QI7, 5QI8 and 5QI9, and the four 5QI levels are mapped to the following types of services.

[0048] 5QI6: High-priority video streaming and TCP-based services;

[0049] 5QI7: Live streaming and interactive gaming business;

[0050] 5QI8: Medium-priority video streaming and TCP-based services;

[0051] 5QI9: Low-priority video streaming and TCP-based services;

[0052] For the four 5QI-level services mentioned above, the current baseline wireless configuration weights are: 5QI6:5QI7:5QI8:5QI9 = 8:1:4:2. This weight coefficient represents the rate ratio of the four 5QI-level users when performing full-buffer services, which is 8:1:4:2. The weight coefficients can be determined based on requirements and are not limited here.

[0053] Each 5QI level includes at least three traffic statistics metrics, such as RLC (Radio Link Control) layer traffic flow, average number of active users, and average RLC packet latency.

[0054] The input module's call statistics data should include at least four types of RLC (Radio Link Control) layer traffic flow, average number of active users (user), and average RLC packet latency (t) for each of the four 5QI levels. delay These are three types of statistical indicators.

[0055] Module B enables cell-level call statistics time series prediction. Module B determines predicted values ​​for various call statistics indicators based on the cell's call statistics data. For example, it predicts the cell's 5QI-level RLC layer traffic flow, average number of active users (user), and average RLC packet latency (t) extracted over a week. delay By balancing the accuracy of prediction based on time series characteristics with the computational cost of prediction, a model is constructed and parameters are fitted to obtain the predicted values ​​of each statistical indicator.

[0056] It should be noted that the predicted values ​​of each call statistics indicator in the embodiments of this application refer to the actual predicted values.

[0057] Step 200: Determine the fluctuation range of each of the call statistics indicators based on the predicted value and standard deviation of each indicator.

[0058] After determining the predicted values ​​of each call statistics indicator, the fluctuation range of each call statistics indicator is determined based on the predicted values ​​and standard deviations of each call statistics indicator, that is, the fluctuation range of each call statistics indicator is predicted.

[0059] Step 300: Determine the first sparse matrix of outlier samples based on the predicted values ​​and fluctuation ranges of each of the aforementioned statistical indicators;

[0060] After determining the predicted values ​​of each call statistics indicator, the first sparse matrix of outlier samples is determined based on the predicted values ​​and fluctuation ranges of each call statistics indicator. For example, the sparse matrix of outlier points of the call statistics indicator is constructed by the statistical voting method of the 5QI level call statistics indicator, so as to realize the determination of outlier abnormal fluctuations at the 5QI level in the cell.

[0061] Step 400: Recursively adjust the weight factors of each level of speech statistics data according to the first sparse matrix, and allocate resources based on the adjustment results.

[0062] After determining the first sparse matrix, the weight factors of each level of speech statistics data are recursively adjusted based on the first sparse matrix, and resources are allocated based on the adjustment results.

[0063] The resource allocation method provided in this application involves determining the predicted values ​​of various call statistics indicators based on the cell's call statistics data; determining the fluctuation range of each call statistics indicator based on the predicted values ​​and standard deviations of each indicator; determining a first sparse matrix of outlier samples based on the predicted values ​​and fluctuation ranges of each indicator; and recursively adjusting the weight factors of call statistics data at each level based on the first sparse matrix, thereby allocating resources based on the adjustment results. This application's recursive approach to adjusting weight factors achieves self-convergence in 5QI-level service resource allocation at the cell level, ensuring user experience for various types of 5QI services and improving the allocation and utilization efficiency of wireless network resources.

[0064] Based on the above embodiments, the step of determining the predicted values ​​of various call statistics indicators according to the call statistics data of the cell includes: determining a compensation factor matrix according to the call statistics data; determining a second current network indicator matrix according to the compensation factor matrix and the first current network indicator matrix; and determining the predicted values ​​of various call statistics indicators according to the second current network indicator matrix and the fitting method.

[0065] For example, taking a single cell as an example, the compensation factor matrix is ​​calculated based on the cell's call statistics data.

[0066] Input the 5QI-level call statistics for 24 / 7 operation per week into the four-dimensional matrix J 4*3*7*24 The first dimension represents the four 5QI dimensions, the second dimension represents the three statistical indicators, the third dimension represents the seven days of the week, and the fourth dimension represents the 24 hours of the day. Then, the four-dimensional matrix J... 4*3*7*24 The three-dimensional matrix K is obtained by taking the average of the elements in the middle with respect to the third dimension. 4*3*24 The elements in the three-dimensional matrix represent the hourly average of a specific 5QI-level call statistics indicator over a week. Finally, the three-dimensional matrix K... 4*3*24 When the index in the matrix is ​​divided by 12, the corresponding statistical index value is (i.e., matrix S). 4*3 (Identifier) ​​to obtain the compensation factor matrix K' 4*3*24 That is, the compensation factor matrix K' 4*3*24 =Matrix K 4*3*24 / Matrix S 4*3 , where the rounded S element is the 12-hour daily average index.

[0067] By calculating the compensation factor matrix mentioned above, periodic features at different times of the 24-hour period can be captured.

[0068] The first live network indicator matrix J 4*3*7*24 Divide by the compensation factor matrix K' 4*3*24 The second live network index matrix J' is obtained 4*3*7*24 That is, the second current network indicator matrix J' 4*3*7*24 =First live network indicator matrix J 4*3*7*24 / Compensation factor matrix K' 4*3*24 .

[0069] For the second live network indicator matrix J' 4*3*7*24 Each indicator dimension and each 5QI level dimension corresponds to 168 (corresponding to the second live network indicator matrix J') 4*3*7*24 The statistical indicators (in the third and fourth dimensions) form a sequence and are fitted using the following formula to achieve the prediction calculation of relevant statistical indicators:

[0070] X i =α*S i +(1-α)*S i-1

[0071] In the formula, α represents the parameter to be fitted, α∈(0,1), S i X represents the actual value of the index element at time step i. i The initial predicted value of the statistical index element is represented by time step i.

[0072] Calculate the value of parameter α under the following constraints:

[0073]

[0074] Where k represents the number of cell-level call statistics indicators within a week, i.e., k = 168;

[0075] After calculating the value of the α parameter based on the constraints, the predicted values ​​of the three call statistics indicators corresponding to each of the 5QI levels in the current time period can be calculated based on the α parameter, forming the prediction matrix S' after factor compensation. 4*3 , where matrix S' 4*3 The rows in the table represent the number of 5QI levels, and the columns represent the number of statistical indicators.

[0076] Understandably, due to X i These are the initial predicted values ​​after compensation factor transformation. Therefore, when determining the actual predicted values, it is necessary to adjust X. i To perform the conversion, that is, to convert X i Multiplying this by the compensation factor yields the actual predicted value of the statistical index. For example, using the prediction matrix S' 4*3 Multiply by the compensation factor matrix K' 4*3*24 The corresponding time period values ​​are used to obtain matrix S 4*3 Matrix S4*3 The elements in the table represent the actual predicted values ​​of various statistical indicators.

[0077] This application embodiment constructs a simple and effective time series algorithm that considers both the overall trend and periodic fluctuation characteristics of the indicators, thereby achieving accurate prediction of 5QI-level wireless indicators and improving the prediction accuracy of various call statistics indicators.

[0078] Based on the above embodiments, the step of recursively adjusting the weight factors of each level of call statistics data according to the first sparse matrix includes: determining the adjustment range of each level of call statistics data according to the element values ​​in the first sparse matrix, and then recursively adjusting the weight factors of each level of call statistics data according to the adjustment range.

[0079] The first sparse matrix M″ is used to identify outliers in the call statistics indicators corresponding to each 5QI level service at the cell level. n For elements exhibiting outliers in call statistics indicators, the corresponding 5QI wireless weight factors are appropriately adjusted. For example, the adjustment range for the 5QI weight factors is set to [-50%, 50%], i.e., the 5QI6 weight factor is [4, 12], the 5QI7 weight factor is [0.5, 1.5], the 5QI8 weight factor is [2, 6], and the 5QI9 weight factor is [1, 3]. To minimize the impact of drastic adjustments on user perception, each adjustment is limited to 10%, and the adjustments are implemented according to the following principles:

[0080] If the element e″ in the first sparse matrix n =1 or -1, but if the 5QI level wireless weight factor has been set to the upper or lower limit of [-50%, 50%], then this parameter will not be adjusted.

[0081] If the element e″ in the first sparse matrix n =1, then the weight factor for the specific 5QI level of this cell will be increased by 10%;

[0082] If the element e″ in the first sparse matrix n If the value is -1, then the weighting factor for the specific 5QI level of the cell will be reduced by 10%.

[0083] After determining the adjustment range, the weighting factors of each level of call statistics data are recursively adjusted based on the adjustment range.

[0084] This application embodiment achieves self-convergence of 5QI-level service resource allocation at the cell level through an hourly recursive call method, ensuring user experience for various types of 5QI services and improving the utilization efficiency of wireless network resources.

[0085] Based on the above embodiments, the resource allocation based on the adjustment result includes: if it is determined based on the adjustment result that the recursive call cycle has not been reached, then resource allocation is performed based on the adjustment result; if it is determined based on the adjustment result that the recursive call cycle has been reached, then the recursively adjusted weight value is obtained; if the weight value reaches the adjustment threshold of the weight factor, then resource allocation is performed based on the adjustment result.

[0086] Since user behavior patterns can change due to various factors, the configuration of 5QI-level weight factor parameters at the cell level can be self-optimized through periodic recursive calls, such as implementing recursive calls at the hourly granularity.

[0087] If the recursive call cycle is not reached, the existing wireless parameter configuration scheme continues; if the recursive call cycle is reached, it is determined whether the recursively adjusted weight value has reached the adjustment threshold. If the adjustment threshold has been reached, the existing wireless parameter configuration scheme continues; if the adjustment threshold has not been reached, the steps for inputting the cell's 5QI-level service data are recursively called.

[0088] This application embodiment achieves self-convergence of 5QI-level service resource allocation at the cell level through an hourly recursive call method, ensuring user experience for various types of 5QI services and improving the utilization efficiency of wireless network resources.

[0089] Based on the above embodiments, the step of determining the first sparse matrix of outlier samples according to the predicted values ​​and fluctuation ranges of each of the call statistics indicators includes: determining a second sparse matrix according to the predicted values ​​and fluctuation ranges of each of the call statistics indicators; then, determining a one-dimensional matrix of the second sparse matrix, and determining the first sparse matrix of outlier samples based on the one-dimensional matrix and the value method of the matrix elements.

[0090] The second sparse matrix M is determined based on the predicted values ​​and fluctuation ranges of various statistical indicators. n*o The elements e in the second sparse matrix are summed using the corresponding call statistics indicators for each of the 5QI levels in the cell. Based on the three call statistics indicators, a one-dimensional matrix M′ is obtained through a voting method. n , where the one-dimensional matrix M′ n In this context, 'n' represents the 5QI category, i.e.:

[0091]

[0092] Based on the calculation of three call statistics indicators, the matrix element values ​​are set between [-3, 3]. When the element value in the one-dimensional matrix is ​​greater than or equal to 1 or less than or equal to -1, it indicates that the service distribution corresponding to 5QI in that cell has obvious outlier characteristics (i.e., at least two of the three call statistics indicators exceed the standard range). The first sparse matrix M″ is formed according to the following principles. n :

[0093] If the element e′ in a one-dimensional matrix n When >= 1, then the element e″ in the first sparse matrix n =1;

[0094] If the element e′ in a one-dimensional matrix n When <= -1, then the element e″ in the first sparse matrix n =-1;

[0095] Other elements e″ in the first sparse matrix n =0.

[0096] This application embodiment constructs a second sparse matrix of outlier points in call statistics indicators to determine outlier anomalies at the 5QI level in a cell, thereby improving the accuracy of resource allocation.

[0097] Based on the above embodiments, the step of determining the second sparse matrix according to the predicted values ​​and fluctuation ranges of each of the call statistics indicators includes: determining the element values ​​of the second sparse matrix according to the predicted values ​​and fluctuation ranges of each of the call statistics indicators, and then determining the second sparse matrix according to the element values ​​of the second sparse matrix.

[0098] Based on the actual predicted values ​​of the 5QI-level call statistics indicators at the entire network's cell level and the fluctuation range of the call statistics indicators, a second sparse matrix M is constructed. n*o Where n represents the 5QI category (e.g., 4 by default, corresponding to 5QI6-5QI9), and o represents the dimension of the statistical indicators (e.g., 3 by default). For example, the array of three 5QI-level statistical indicators at the cell level in the current network is mapped into the second sparse matrix, where the elements in the second sparse matrix begin with e. n*o The identifier, where n represents the 5QI category and o represents the call statistics dimension. If e n*o The element value is outside the fluctuation range, i.e., greater than P. flow +STD flow Or less than P flow -STD flow At this time e n*o Assign a value of 1; if e n*o The element value is within the fluctuation range, that is, in P. flow -STD flow and P flow +STD flow Between, at this time e n*o By assigning a value of -1, and based on this, after determining the values ​​of each element, the second sparse matrix M can be obtained. n*o .

[0099] This application embodiment constructs a second sparse matrix of outlier points in call statistics indicators to determine outlier anomalies at the 5QI level in a cell, thereby improving the accuracy of resource allocation.

[0100] Based on the above embodiments, the step of determining the fluctuation range of each of the call statistics indicators according to the predicted value and standard deviation of each of the call statistics indicators includes: determining the sum and difference of the predicted value and standard deviation of each of the call statistics indicators, and then determining the fluctuation range according to the sum and difference.

[0101] For example, after determining the predicted values ​​for various call statistics indicators such as 5QI-level RLC layer traffic volume, average number of active users, and average RLC packet latency, the predicted values ​​of each call statistics indicator are multiplied by X. i Multiplying the actual predicted values ​​of each call statistics indicator by the corresponding compensation factor yields the actual predicted values ​​of the RLC layer traffic volume, such as P. flow The actual predicted value for the average number of activated users is P. user The actual predicted value of the average packet delay of RLC is P. t Then, by using the call statistics indicators over a week (168 hours), the standard deviation (STD) corresponding to each of the three call statistics indicators at the 5QI level can be obtained. flow STD user and STD t Based on the actual predicted values ​​and standard deviations of various call statistics indicators, the following predicted fluctuation ranges for the call statistics indicators are obtained:

[0102] RLC layer traffic volume prediction fluctuation range ∈ [P] flow -STD flow P flow +STD flow ];

[0103] Average active user count predicted fluctuation range ∈ [P] user -STD user P user +STD user ];

[0104] RLC average packet delay prediction fluctuation range ∈ [P] t -STD t P t +STD t ];

[0105] Specifically, when the actual predicted value of the call statistics indicator exceeds the predicted fluctuation range, it is determined that the call statistics indicator has an outlier during that period.

[0106] This application embodiment predicts the fluctuation range of various call statistics indicators based on their actual predicted values ​​and standard deviations, and identifies outlier call statistics indicators through the predicted fluctuation ranges, thereby improving the accuracy of resource allocation.

[0107] To more clearly illustrate the resource allocation method of this application, refer to Figure 2 The following specific examples will be used to illustrate this.

[0108] Step A: Taking cell A as an example, collect hourly call statistics data for the cell at the 5QI level within one week and input it into the module. The input call statistics indicators include the RLC layer traffic volume, average number of active users, and average packet latency at the RLC layer for four types of 5QI (including 5QI6, 5QI7, 5QI8, and 5QI9). The total number of input samples is:

[0109] 4 (number of 5QI levels) * 3 (number of indicator dimensions) * 168 (7 * 24) = 2016

[0110] Step B: Output the outlier samples from the 2016 input data in sparse matrix form, and input the matrix data into the next step, as shown in steps B1-B4.

[0111] Step B1: Input 2016 sample data to achieve time series prediction of statistical indicators.

[0112] For example, let's analyze the average number of activated users (user) at level 5QI6, which represents the number of activated users within 168 hours in a week at level 5QI6, using matrix J. 7*24 This means that matrix J 7*24 The mean of the columns is used to obtain the one-dimensional matrix K:

[0113] K=[1.1, 2.0, 4.3, 5.4, 4.5, 10.2, 20.4, 24.5, 30.3, 31.4, 42.4, 56.9, 42.7, 46.8, 39.5, 36, 32.3, 30, 29, 26.2, 19.4, 12.7, 6.3, 2.2]

[0114] In this one-dimensional matrix, K 12 The element 56.9 indicates that the average number of active users under 5QI6 at 12:00 within 7 days is 56.9. Using the user data at 12:00 as a benchmark, the compensation factor matrix for each time period is calculated:

[0115] K′=[1.1 / 56.9, 2.0 / 56.9, 4.3 / 56.9, 5.4 / 56.9, 4.5 / 56.9, 10.2 / 56.9, 20.4 / 56.9, 24.5 / 56.9, 30.3 / 56.9, 31.4 / 56.9, 42.4 / 56.9, 56.9 / 5 6.9, 42.7 / 56.9, 46.8 / 56.9, 39.5 / 56.9, 36 / 56.9, 32.3 / 56.9, 30 / 56. 9, 29 / 56.9, 26.2 / 56.9, 19.4 / 56.9, 12.7 / 56.9, 6.3 / 56.9, 2.2 / 56.9]

[0116] Based on the compensation factor matrix K′, using matrix J 7*24 Divide by matrix K′, i.e. matrix J 7*24 The matrix K′ yields the factor-compensated matrix J′ for each sample. The 168 elements of J′ are then fitted using a formula to determine the value of the fitting parameter α, such that... Obtain the minimum value.

[0117] Since α∈(0,1), considering computational cost and accuracy, the calculation is performed with a step size of 0.1, yielding the following results:

[0118] When α = 0.9,

[0119] When α = 0.8,

[0120] When α = 0.7,

[0121] When α = 0.6,

[0122] When α = 0.5,

[0123] When α = 0.4,

[0124] When α = 0.3,

[0125] When α = 0.2,

[0126] When α = 0.1,

[0127] The above fitting results show that when α = 0.7, To obtain the minimum value, substitute α = 0.7 into X. i =α*S i +(1-α)*S i-1 S i =189.67, S i-1 =310.00, thus obtaining

[0128] X i =0.7*189.67+(1-0.7)*310.00=225.77

[0129] X i Multiplying by the compensation factor for the corresponding time period in the compensation factor matrix yields the actual predicted value. Therefore, the actual predicted value for users at the 5QI level in the next time period is 4.36, i.e.:

[0130] Xi *K′1=225.77*1.1 / 56.9=4.36

[0131] Step B2: Calculate the standard deviation of the number of activated users and determine the range of fluctuation.

[0132] Assuming the standard deviation of the active user count statistics over 168 time periods is 3.81, and since the actual predicted value for 5QI6 level users is 4.36, the fluctuation range of the 5QI6 level active user count in the next time period is [4.36-3.81, 4.36+3.81], which is a value range of [0.55, 8.17].

[0133] The actual output of the number of activated users for this period was 12. Since 12 is greater than 8.17, the number of 5QI Level 6 activated users for this period is out of the loop and exceeds the fluctuation threshold.

[0134] Step B3: Construct a sparsity matrix for outlier samples.

[0135] Based on the calculation results of step B2, the number of 5QI6 level activated users in this period exceeds the threshold limit and is marked as 1, forming a coefficient matrix.

[0136] Step B4: Construct the sparsification matrix for outliers.

[0137] If the traffic volume and average packet latency of the RLC layer at level 5QI6 also exceed the upper limit during this period, then the predicted values ​​of the indicators will be used to construct matrix M. n*o The element corresponding to 5QI6 is 3, and so on. Using the method described above, we can obtain M′. n*o The matrix is:

[0138] M′ n*o = [[3, 0, 0], [-1, 0, 0], [0, 0, 3], [0, 0, 0]]

[0139] For matrix M′ n*o Summing by columns yields the sparse matrix of outliers that exceed the bounds:

[0140] M″ n*o = [3, -1, 3, 0]

[0141] Step C enables automated adjustment of the 5QI-level wireless weight factor.

[0142] Based on the output of step B, there are three 5QI levels where user metrics exceed limits. For matrix sparsity, 5QI6 exceeds the upper limit, 5QI7 is below the lower limit, and 5QI8 exceeds the upper limit. For these three outlier dimensions, the corresponding 5QI level weight factors need to be adjusted during this period.

[0143] Assuming the current weight factor for 5QI6 is 8, it will be increased by 10% to 8.8; the current weight factor for 5QI7 is 1, it will be decreased by 10% to 0.9; the current weight factor for 5QI8 is 4, it will be increased by 10% to 4.4; and the current weight factor for 5QI9 remains unchanged. The corresponding optimized weight factors are:

[0144] 5QI6:5QI7:5QI8:5QI9=8.8:0.9:4.4:2

[0145] After the calculation is completed, the radio weight factor of cell A1 is adjusted according to the above ratio through the automated interface.

[0146] Step D implements periodic recursive adjustments.

[0147] Step A is called again every hour to perform high-frequency adjustments to the 5QI-level wireless weight factor on an hourly basis, ensuring that the cell-level wireless 5QI-level weight parameters can be dynamically adjusted every hour. For example, it checks whether the recursive call cycle has been reached. If the recursive call cycle has not been reached, the existing wireless parameter configuration scheme continues. If the recursive call cycle has been reached, it checks whether the recursively adjusted weight value has reached the adjustment threshold. If the adjustment threshold has been reached, the existing wireless parameter configuration scheme continues; if the adjustment threshold has not been reached, step A is called recursively.

[0148] This application embodiment achieves accurate prediction of 5QI-level wireless indicators by constructing a time-series algorithm that considers both the overall trend and periodic fluctuation characteristics of the indicators. A sparse matrix of outlier points is constructed using a statistical voting method for 5QI-level indicators to identify outlier and abnormal fluctuations at the cell level. Through hourly recursive calls, the distribution of users at each 5QI level within the cell is periodically calculated, and wireless parameters are quickly adjusted based on this calculation to ensure service availability for users at different 5QI levels. Furthermore, the recursive adjustment is automated, reducing manual resource input. This application embodiment can simply and effectively achieve reasonable allocation of 5QI service resources for different types of users within the cell, ensuring service availability for all types of users in the network.

[0149] The resource allocation apparatus provided in the embodiments of this application is described below. The resource allocation apparatus described below and the resource allocation method described above can be referred to in correspondence.

[0150] refer to Figure 3 , Figure 3 This is a schematic diagram of the structure of the resource allocation device provided in the embodiments of this application. The resource allocation device provided in the embodiments of this application includes a prediction module 301, a first determination module 302, a second determination module 303, and a resource allocation module 304.

[0151] The prediction module 301 is used to determine the predicted values ​​of various call statistics indicators based on the call statistics data of the cell.

[0152] The first determining module 302 is used to determine the fluctuation range of each of the call statistics indicators based on the predicted value and standard deviation of each of the call statistics indicators;

[0153] The second determining module 303 is used to determine the first sparse matrix of outlier samples based on the predicted values ​​and fluctuation ranges of each of the aforementioned statistical indicators;

[0154] The resource allocation module 304 is used to recursively adjust the weight factors of each level of speech statistics data according to the first sparse matrix, so as to allocate resources based on the adjustment results.

[0155] The resource allocation device provided in this application determines the predicted values ​​of various call statistics indicators based on the call statistics data of the cell; determines the fluctuation range of each call statistics indicator based on the predicted values ​​and standard deviations of each call statistics indicator; determines the first sparse matrix of outlier samples based on the predicted values ​​and fluctuation ranges of each call statistics indicator; and recursively adjusts the weight factors of the call statistics data at each level based on the first sparse matrix, so as to perform resource allocation based on the adjustment results. This application embodiment achieves self-convergence of 5QI-level service resource allocation at the cell level by adjusting the weight factors through recursive calls, ensuring user experience for various types of 5QI services and improving the allocation and utilization efficiency of wireless network resources.

[0156] In one embodiment, the prediction module 301 is specifically used for:

[0157] Determine the compensation factor matrix based on the aforementioned call statistics data;

[0158] The second live network indicator matrix is ​​determined based on the compensation factor matrix and the first live network indicator matrix.

[0159] Based on the second existing network indicator matrix and the fitting method, the predicted values ​​of each of the call statistics indicators are determined.

[0160] In one embodiment, the resource allocation module 304 is specifically used for:

[0161] The adjustment range of each level of call statistics data is determined based on the element values ​​in the first sparse matrix.

[0162] The weighting factors of each level of call statistics data are recursively adjusted according to the aforementioned adjustment range.

[0163] In one embodiment, the resource allocation module 304 is specifically used for:

[0164] If it is determined based on the adjustment results that the recursive call cycle has not been reached, then resource allocation is performed based on the adjustment results.

[0165] If it is determined based on the adjustment results that the recursive call cycle has been reached, then the recursively adjusted weight value is obtained.

[0166] If the weight value reaches the adjustment threshold of the weight factor, then resource allocation is performed based on the adjustment result.

[0167] In one embodiment, the second determining module 303 is specifically used for:

[0168] The second sparse matrix is ​​determined based on the predicted values ​​and fluctuation ranges of each of the aforementioned statistical indicators;

[0169] A one-dimensional matrix of the second sparse matrix is ​​determined, and a first sparse matrix of the outlier samples is determined based on the one-dimensional matrix and the value of the matrix elements.

[0170] In one embodiment, the second determining module 303 is specifically used for:

[0171] The element values ​​of the second sparse matrix are determined based on the predicted values ​​and fluctuation ranges of the various statistical indicators.

[0172] The second sparse matrix is ​​determined based on the element values ​​of the second sparse matrix.

[0173] In one embodiment, the first determining module 302 is specifically used for:

[0174] Determine the sum and difference of the predicted values ​​and standard deviations of each of the aforementioned statistical indicators;

[0175] The fluctuation range is determined based on the sum and the difference.

[0176] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other via the communication bus 440. The processor 410 can call a computer program in the memory 430 to execute steps of a resource allocation method, such as:

[0177] The predicted values ​​for various call statistics indicators are determined based on the call statistics data of the community;

[0178] The fluctuation range of each of the aforementioned call statistics indicators is determined based on the predicted value and standard deviation of each indicator.

[0179] The first sparse matrix of outlier samples is determined based on the predicted values ​​and fluctuation ranges of the various statistical indicators.

[0180] The weighting factors of each level of speech statistics data are recursively adjusted based on the first sparse matrix, and resources are allocated based on the adjustment results.

[0181] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0182] On the other hand, this application also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform the steps of the resource allocation method provided in the above embodiments, such as including:

[0183] The predicted values ​​for various call statistics indicators are determined based on the call statistics data of the community;

[0184] The fluctuation range of each of the aforementioned call statistics indicators is determined based on the predicted value and standard deviation of each indicator.

[0185] The first sparse matrix of outlier samples is determined based on the predicted values ​​and fluctuation ranges of the various statistical indicators.

[0186] The weighting factors of each level of speech statistics data are recursively adjusted based on the first sparse matrix, and resources are allocated based on the adjustment results.

[0187] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0188] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0189] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A resource allocation method, characterized in that, include: The predicted values ​​for various call statistics indicators are determined based on the call statistics data of the community; The fluctuation range of each of the aforementioned call statistics indicators is determined based on the predicted value and standard deviation of each indicator. Based on the predicted values ​​and fluctuation ranges of each of the aforementioned call statistics indicators, the second sparse matrix corresponding to each call statistics indicator is determined respectively; The second sparse matrix is ​​reduced in dimensionality to obtain a one-dimensional matrix; Based on whether each element in the one-dimensional matrix exceeds a predetermined threshold, a first sparse matrix of outlier samples is determined; wherein, the element values ​​in the first sparse matrix are used to identify whether the data at the corresponding level is an outlier sample. The adjustment range of each level of call statistics data is determined based on the element values ​​in the first sparse matrix. The weighting factors of each level of call statistics data are recursively adjusted according to the adjustment range, so as to allocate resources based on the adjustment results; The recursive adjustment includes: when the recursive call cycle is met, redetermining the predicted values ​​and fluctuation ranges of various call statistics indicators to update the first sparse matrix, and adjusting the weight factors again based on the updated first sparse matrix.

2. The resource allocation method according to claim 1, characterized in that, The process of determining the predicted values ​​of various call statistics indicators based on the call statistics data of the cell includes: Determine the compensation factor matrix based on the aforementioned call statistics data; The second live network indicator matrix is ​​determined based on the compensation factor matrix and the first live network indicator matrix. Based on the second existing network indicator matrix and the fitting method, the predicted values ​​of each of the call statistics indicators are determined.

3. The resource allocation method according to claim 1, characterized in that, The resource allocation based on the adjustment results includes: If it is determined based on the adjustment results that the recursive call cycle has not been reached, then resource allocation is performed based on the adjustment results. If it is determined based on the adjustment results that the recursive call cycle has been reached, then the recursively adjusted weight value is obtained. If the weight value reaches the adjustment threshold of the weight factor, then resource allocation is performed based on the adjustment result.

4. The resource allocation method according to claim 1, characterized in that, The step of determining the second sparse matrix corresponding to each of the call statistics indicators based on the predicted values ​​and fluctuation ranges of each indicator includes: The element values ​​of the second sparse matrix are determined based on the predicted values ​​and fluctuation ranges of the various statistical indicators. The second sparse matrix is ​​determined based on the element values ​​of the second sparse matrix.

5. The resource allocation method according to claim 1, characterized in that, The step of determining the fluctuation range of each of the call statistics indicators based on the predicted value and standard deviation of each indicator includes: Determine the sum and difference of the predicted values ​​and standard deviations of each of the aforementioned statistical indicators; The fluctuation range is determined based on the sum and the difference.

6. A resource allocation device, characterized in that, include: The prediction module is used to determine the predicted values ​​of various call statistics indicators based on the call statistics data of the cell. The first determining module is used to determine the fluctuation range of each of the call statistics indicators based on the predicted value and standard deviation of each of the call statistics indicators; The second determining module is used to determine the second sparse matrix corresponding to each of the call statistics indicators based on the predicted value and fluctuation range of each of the call statistics indicators. The second sparse matrix is ​​reduced in dimensionality to obtain a one-dimensional matrix; Based on whether each element in the one-dimensional matrix exceeds a predetermined threshold, a first sparse matrix of outlier samples is determined; wherein, the element values ​​in the first sparse matrix are used to identify whether the data at the corresponding level is an outlier sample. The resource allocation module is used to determine the adjustment range of each level of call statistics data based on the element values ​​in the first sparse matrix; and to recursively adjust the weight factors of each level of call statistics data according to the adjustment range, so as to allocate resources based on the adjustment results; wherein, the recursive adjustment includes: when the recursive call cycle is met, re-determining the predicted values ​​and fluctuation ranges of each call statistics indicator to update the first sparse matrix, and adjusting the weight factors again based on the updated first sparse matrix.

7. An electronic device comprising a processor and a memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the resource allocation method according to any one of claims 1 to 5.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the resource allocation method according to any one of claims 1 to 5.

Images