A method and system for optimizing uplink and downlink rates based on LTE base stations

Abstract

The present invention discloses a method and system for optimizing uplink and downlink rates based on an LTE base station, and relates to the field of data processing technology. The method comprises the following steps: collecting uplink and downlink traffic received by a target LTE base station in real time to obtain initial traffic data, preprocessing the initial traffic data to obtain preprocessed data, obtaining historical traffic data, and predicting the uplink and downlink traffic of the next cycle at the current moment based on the historical traffic data and the preprocessed data to obtain first predicted traffic data; substituting the preprocessed data into a preset model to obtain second predicted traffic data; obtaining a target predicted traffic based on the first predicted traffic data and the second predicted traffic data; and optimizing the uplink and downlink rates based on the target predicted traffic. By collecting uplink and downlink traffic in real time and performing preprocessing, the changing trend of the traffic can be obtained in real time. By combining traffic prediction based on historical data and a preset model, network parameters can be dynamically adjusted, thereby improving optimization efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular relates to an uplink and downlink rate optimization method and system based on an LTE base station. Background Art

[0002] With the rapid development of the mobile internet, users' demand for data transmission speeds continues to increase, especially in 4G LTE networks. Improving uplink and downlink speeds has become a key task in network optimization. LTE networks offer significant advantages in terms of user experience, network capacity, and signal coverage. However, in high-density areas or when there are many users, network performance may be affected, resulting in unstable or substandard uplink and downlink speeds. To ensure a high-quality network experience for users, accurate rate prediction and intelligent scheduling optimization are necessary to proactively address network load fluctuations, avoid congestion, and ensure that every user receives the desired speed in different scenarios. However, current rate prediction methods still suffer from insufficient accuracy, making it difficult to fully and accurately reflect changes in complex network environments.

[0003] Patent CN112702792B discloses a method for jointly allocating uplink and downlink resources in a wireless energy-carrying network based on GFDM. Each user in the downlink uses a power-splitting receiver structure to decode the received signal and collect energy. The collected energy is then used for uplink information transmission. By allocating GFDM subcarriers and subsymbols, power allocation, and power-splitting factors in the uplink and downlink, the weighted achievable sum rate of the uplink and downlink is maximized while satisfying the constraints of collected energy and total transmitted power. To solve this non-convex optimization problem, the Lagrangian dual method, subgradient method, and greedy algorithm are used to jointly optimize the uplink and downlink power allocation and power-splitting factors. However, this solution may still lead to low optimization efficiency when the specific uplink and downlink transmission requirements are unknown. Summary of the Invention

[0004] The purpose of the present invention is to solve the above problems and to propose a method and system for optimizing uplink and downlink rates based on LTE base stations.

[0005] In a first aspect of the present invention, a method for optimizing uplink and downlink rates based on an LTE base station is first proposed, the method comprising:

[0006] Collecting uplink and downlink traffic received by the target LTE base station in real time to obtain initial traffic data, and preprocessing the initial traffic data to obtain preprocessed data;

[0007] Acquire historical traffic data, and predict the uplink and downlink traffic of the next cycle at the current moment based on the historical traffic data and the preprocessed data to obtain first predicted traffic data;

[0008] Substituting the preprocessed data into a preset model to obtain second predicted flow data;

[0009] Obtaining a target predicted flow rate according to the first predicted flow rate data and the second predicted flow rate data;

[0010] Optimize the uplink and downlink rates based on the target predicted traffic.

[0011] Optionally, predicting the uplink and downlink traffic of the next cycle at the current moment according to the historical traffic data and the preprocessed data to obtain the first predicted traffic data includes:

[0012] Acquiring the time characteristics and state characteristics of the pre-processed data, and filtering the historical traffic data according to the time characteristics and the state characteristics to obtain a target historical traffic data set;

[0013] For each target historical flow data in the target historical flow data set, performing uplink and downlink flow curve fitting on the target historical flow data to obtain an initial prediction curve;

[0014] Performing curve fitting on the preprocessed data to obtain an existing flow curve, and screening all initial prediction curves according to the existing flow curve to obtain an effective prediction curve set;

[0015] A target prediction curve is obtained according to the effective prediction curve set, and the uplink and downlink traffic of the next cycle at the current moment is obtained according to the target prediction curve and the existing traffic curve, and the uplink and downlink traffic is recorded as the first predicted traffic data.

[0016] Optionally, screening all initial prediction curves according to the existing flow curve to obtain a valid prediction curve set includes:

[0017] For all initial prediction curves, the similarity value corresponding to each initial prediction curve is obtained by calculating the DTW similarity between the existing flow curve and the initial prediction curve;

[0018] The initial prediction curves with similarity values greater than a preset threshold are obtained to obtain a valid prediction curve set.

[0019] Optionally, substituting the preprocessed data into a preset model to obtain second predicted traffic data includes:

[0020] Segmenting the preprocessed data according to a preset time period, and performing mean normalization on all segmented data to obtain a normalized data set;

[0021] Performing logarithmic transformation on all data in the normalized data set and then deseasonalizing to obtain a time data set;

[0022] Substitute the time data set into the encoder to obtain time features, and substitute the time features into the TCN prediction model to obtain second predicted traffic data.

[0023] Optionally, after optimizing the uplink and downlink rates according to the target predicted traffic, the method further includes:

[0024] Obtain the optimized results and verify them to obtain the correction factor;

[0025] Calculating the predicted flow rate of the next cycle after optimization, and correcting the predicted flow rate of the next cycle after optimization according to the correction factor to obtain a target corrected flow rate;

[0026] The uplink and downlink rates are optimized according to the target corrected traffic.

[0027] In a second aspect of the present invention, a system for optimizing uplink and downlink rates based on an LTE base station is proposed, comprising:

[0028] A preprocessing module is used to collect the uplink and downlink traffic received by the target LTE base station in real time to obtain initial traffic data, and preprocess the initial traffic data to obtain preprocessed data;

[0029] A first traffic prediction data module is used to obtain historical traffic data, and predict the uplink and downlink traffic of the next cycle at the current moment based on the historical traffic data and the pre-processed data to obtain first predicted traffic data;

[0030] A second predicted traffic data module, configured to substitute the pre-processed data into a preset model to obtain second predicted traffic data;

[0031] a target predicted flow determination module, configured to obtain a target predicted flow according to the first predicted flow data and the second predicted flow data;

[0032] The first rate optimization module is used to optimize the uplink and downlink rates according to the target predicted traffic.

[0033] Optionally, the first traffic prediction data module includes:

[0034] A traffic screening module, configured to obtain the time characteristics and state characteristics of the pre-processed data, and screen the historical traffic data according to the time characteristics and the state characteristics to obtain a target historical traffic data set;

[0035] An initial prediction curve determination module is used to perform uplink and downlink flow curve fitting on each target historical flow data in the target historical flow data set to obtain an initial prediction curve;

[0036] a curve screening module, configured to perform curve fitting on the preprocessed data to obtain an existing flow curve, and screen all initial prediction curves according to the existing flow curve to obtain a valid prediction curve set;

[0037] The first predicted traffic data generation module is used to obtain a target prediction curve based on the effective prediction curve set, obtain the uplink and downlink traffic of the next cycle at the current moment based on the target prediction curve and the existing traffic curve, and record the uplink and downlink traffic as the first predicted traffic data.

[0038] Optionally, the curve screening module includes:

[0039] A curve similarity value calculation module is used to obtain a similarity value corresponding to each initial prediction curve by calculating the DTW similarity between the existing flow curve and the initial prediction curve for all initial prediction curves;

[0040] The effective prediction curve set generation module is used to obtain the initial prediction curves whose similarity values are greater than a preset threshold to obtain the effective prediction curve set.

[0041] Optionally, the second traffic prediction data module includes:

[0042] A normalization module, configured to segment the preprocessed data according to a preset time period, and perform mean normalization on all segmented data to obtain a normalized data set;

[0043] a time data set determination module, configured to perform logarithmic transformation on all data in the normalized data set and then deseasonalize the data to obtain a time data set;

[0044] The feature extraction module is used to substitute the time data set into the encoder to obtain time features, and substitute the time features into the TCN prediction model to obtain second predicted traffic data.

[0045] Optionally, the system further includes:

[0046] Correction factor determination module, used to obtain the optimized results and verify them to obtain the correction factor;

[0047] a target corrected flow determination module, configured to calculate the predicted flow of the next cycle after optimization, and correct the predicted flow of the next cycle after optimization according to the correction factor to obtain the target corrected flow;

[0048] The second rate optimization module is used to optimize the uplink and downlink rates according to the target corrected traffic.

[0049] Beneficial effects of the present invention:

[0050] The present invention proposes a method for optimizing uplink and downlink rates based on an LTE base station. The method collects uplink and downlink traffic received by a target LTE base station in real time to obtain initial traffic data, preprocesses the initial traffic data to obtain preprocessed data, obtains historical traffic data, and predicts the uplink and downlink traffic of the next cycle at the current moment based on the historical traffic data and the preprocessed data to obtain first predicted traffic data. The preprocessed data is substituted into a preset model to obtain second predicted traffic data. A target predicted traffic is obtained based on the first and second predicted traffic data. Uplink and downlink rate optimization is performed based on the target predicted traffic. By collecting uplink and downlink traffic in real time and performing preprocessing, traffic trends can be obtained in real time. By combining traffic predictions based on historical data and a preset model, network parameters can be dynamically adjusted, thereby improving optimization efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The present invention will be further described below with reference to the accompanying drawings.

[0052] Figure 1 A flowchart of a method for optimizing uplink and downlink rates based on an LTE base station provided in an embodiment of the present invention;

[0053] Figure 2 A framework diagram of an uplink and downlink rate optimization system based on an LTE base station provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention, in conjunction with the accompanying drawings. Obviously, the described embodiments represent only a portion of the embodiments of the present invention, not all of them. The term "and / or" herein simply describes an association relationship between associated objects, indicating that three possible relationships exist. For example, "A" and "B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, references to "first," "second," and so on in the present invention are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined as "first" or "second" may explicitly or implicitly include at least one of these features. Furthermore, the technical solutions of the various embodiments may be combined, but only if they are achievable by a person of ordinary skill in the art. If a combination of technical solutions contradicts or is unachievable, such combination shall be deemed non-existent and outside the scope of protection claimed by the present invention.

[0055] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0056] The embodiment of the present invention provides a method for optimizing uplink and downlink rates based on an LTE base station. Figure 1 , Figure 1 A flowchart of a method for optimizing uplink and downlink rates based on an LTE base station is provided in an embodiment of the present invention. The method comprises the following steps:

[0057] S101, collecting uplink and downlink traffic received by a target LTE base station in real time to obtain initial traffic data, and preprocessing the initial traffic data to obtain preprocessed data;

[0058] S102, obtaining historical traffic data, and predicting the uplink and downlink traffic of the next cycle at the current moment based on the historical traffic data and pre-processed data to obtain first predicted traffic data;

[0059] S103, substituting the pre-processed data into a preset model to obtain second predicted flow data;

[0060] S104, obtaining a target predicted flow rate according to the first predicted flow rate data and the second predicted flow rate data;

[0061] S105: Optimize the uplink and downlink rates according to the target predicted traffic.

[0062] According to an embodiment of the present invention, an uplink and downlink rate optimization method based on an LTE base station is provided. By collecting uplink and downlink traffic in real time and performing preprocessing, the changing trend of traffic can be obtained in real time. By combining traffic prediction based on historical data and a preset model, network parameters can be dynamically adjusted, thereby improving optimization efficiency.

[0063] In one implementation, by collecting and preprocessing initial traffic data in real time and combining it with predictions of historical traffic data, traffic changes in future time periods can be predicted more accurately, helping to promptly understand the changing trends of network loads. By combining multiple prediction methods (such as traffic predictions based on historical data and preset models), network parameters can be dynamically adjusted and upstream and downstream rates can be optimized, helping to ensure that the network can maintain stable service quality under different traffic loads and avoid congestion or waste of resources.

[0064] In one implementation, the target LTE base station is a base station that needs to optimize the uplink and downlink rates; the historical traffic data is the uplink and downlink traffic data collected by the current LTE base station in the past year, excluding the current day; the first predicted traffic data and the second predicted traffic data are averaged to obtain the target predicted traffic; optimizing the uplink and downlink rates according to the target predicted traffic is processed by existing methods and is not described here.

[0065] In one implementation, based on the predicted target traffic, base station resources can be allocated more efficiently, such as adjusting bandwidth, power control, etc., so that network resources can be reasonably utilized, avoiding over-allocation or idle resources, thereby improving the overall network throughput and performance; first, the predicted traffic data can understand the changing patterns of traffic from a historical period.

[0066] In one implementation, the first predicted traffic data is mainly predicted by analyzing historical traffic data, especially the periodic traffic change patterns (such as daily peak hours, weekend effects, etc.), to predict the traffic in the future period, which can reflect long-term trends and patterns and reduce long-term trend errors; the second predicted traffic data is predicted based on the traffic changes on the current date, taking into account the local changes at the current moment, and can capture the short-term change patterns at the current moment, and can adjust the deviations caused by short-term fluctuations. The combination of the two can make the prediction more accurate.

[0067] In one embodiment, predicting the uplink and downlink traffic of the next cycle at the current moment based on the historical traffic data and the pre-processed data to obtain the first predicted traffic data includes:

[0068] Obtain the time characteristics and state characteristics of the preprocessed data, and filter the historical traffic data according to the time characteristics and state characteristics to obtain the target historical traffic data set;

[0069] For each target historical traffic data in the target historical traffic data set, the target historical traffic data is fitted with an uplink and downlink traffic curve to obtain an initial prediction curve;

[0070] Perform curve fitting on the preprocessed data to obtain the existing flow curve, and screen all initial prediction curves according to the existing flow curve to obtain a valid prediction curve set;

[0071] A target prediction curve is obtained according to the effective prediction curve set, and the uplink and downlink traffic of the next cycle at the current moment is obtained according to the target prediction curve and the existing traffic curve, and the uplink and downlink traffic is recorded as the first predicted traffic data.

[0072] In one implementation method, by combining time characteristics and state characteristics to screen historical traffic data, it is possible to effectively find historical data related to traffic changes at the current moment, reduce the impact of noise, and thus improve the accuracy of the prediction results; by fitting the uplink and downlink traffic curves for each target historical traffic data, it can be flexibly adjusted according to different data characteristics, so that the prediction method can adapt to different traffic change patterns.

[0073] In one implementation method, the cycle in the next cycle at the current moment is determined by technical personnel; the time feature is the time period, for example, if the current date is Saturday, the time period is Saturday, and the data recorded on Saturday in the historical traffic data is filtered through the time feature; the status feature is the current network status, weather status and special holidays, for example, if the current network status is normal, the weather is sunny, and there are no holidays, then these data will be used to filter the data that meets the conditions in the historical traffic data.

[0074] In one implementation, the initial prediction curve and the current flow curve are both flow change curves for one day (from midnight of the current day to midnight of the next day).

[0075] In one implementation, the uplink and downlink flow curve fitting of the target historical flow data can be performed through polynomial regression, exponential smoothing, ARIMA model, etc.; the method of curve fitting for preprocessed data is the same as the method of fitting the uplink and downlink flow curve for the target historical flow data.

[0076] In one implementation, all curves in the effective prediction curve set are weightedly fused to obtain a target prediction curve, and each effective prediction curve has the same weight; the uplink and downlink traffic of the next cycle at the current moment is obtained based on the target prediction curve and the existing traffic curve. Specifically, the existing traffic curve is extrapolated to obtain the traffic change curve of the first next cycle, and the traffic change curve of the next cycle of the target prediction curve at the current moment is obtained and recorded as the historical prediction effective change curve. The average of the traffic change curve of the first next cycle and the historical prediction effective change curve is calculated to obtain the target traffic change curve for the next cycle, and the uplink and downlink traffic of the next cycle at the current moment is determined based on the target traffic change curve.

[0077] In one implementation, by screening the existing flow curves and the initial prediction curves, a set of valid prediction curves that best matches the actual situation can be selected, thereby improving the reliability of the prediction and reducing the risk of overfitting; after obtaining the target prediction curve, by comparing it with the existing flow curve, the prediction results can be dynamically adjusted and optimized to make the model more consistent with actual flow changes.

[0078] In one implementation, a screening method based on time characteristics can better adapt to traffic forecasting needs in different time periods, such as the differences in traffic change patterns between peak and non-peak periods, thereby optimizing the prediction of upstream and downstream traffic; by screening and optimizing the initial prediction curve, the amount of calculation can be effectively reduced, and only the most valuable prediction curve set is retained, thereby improving prediction efficiency.

[0079] In one embodiment, all initial prediction curves are screened according to existing flow curves to obtain a valid prediction curve set including:

[0080] For all initial prediction curves, the similarity value corresponding to each initial prediction curve is obtained by calculating the DTW similarity between the existing traffic curve and the initial prediction curve;

[0081] The initial prediction curves with similarity values greater than a preset threshold are obtained to obtain a valid prediction curve set.

[0082] In one implementation, by calculating the dynamic time warping (DTW) similarity between the existing traffic curve and the initial prediction curve, the degree of match between different prediction curves and the existing traffic data can be effectively measured. The effective prediction curve set selected in this way has higher accuracy and can better reflect future traffic trends.

[0083] In one implementation, the preset threshold is determined by a technician.

[0084] In one implementation, DTW similarity can measure the time alignment differences of curves, ensuring that even if there are time offsets or fluctuations in traffic changes, the curve that best matches the current traffic characteristics is selected from the set of valid prediction curves, thereby improving adaptability to different traffic patterns and enhancing robustness.

[0085] In one implementation, by setting a preset threshold and screening out initial prediction curves with a similarity greater than the threshold, prediction curves that differ greatly from existing traffic curves can be removed. These irrelevant or significantly different curves may interfere with the final prediction results. This ensures that the selected curve is more representative and avoids the impact of low-quality prediction curves on the overall prediction accuracy.

[0086] In one embodiment, substituting the preprocessed data into a preset model to obtain second predicted traffic data includes:

[0087] The preprocessed data is segmented according to a preset time period, and all segmented data are mean normalized to obtain a normalized data set;

[0088] All data in the normalized dataset are logarithmically transformed and then deseasonalized to obtain the time dataset;

[0089] Substitute the time dataset into the encoder to obtain the time feature, and substitute the time feature into the TCN prediction model to obtain the second predicted traffic data.

[0090] In one implementation, segmenting the data according to a preset time period helps capture the regular characteristics of different time periods and reduce the interference of long-term trends. Data in different time periods may have different traffic change trends. Segmented processing can better reflect these change characteristics, thereby improving the accuracy of the prediction.

[0091] In one implementation, the preset time period is the same as the period in the next period predicted by the preprocessed data at the current moment; the preprocessed data is sorted in time series; the TCN prediction model includes an input layer, 4 convolution layers (the first convolution layer: 16 convolution kernels, the convolution kernel size is 3, the dilation factor is 1, and causal convolution is used; the second convolution layer: 64 convolution kernels, the convolution kernel size is 3, the dilation factor is 2, and causal convolution is used; the third convolution layer: 128 convolution kernels, the convolution kernel size is 3, the dilation factor is 4, and causal convolution is used; the fourth convolution layer: 256 convolution kernels, the convolution kernel size is 3, the dilation factor is 8, and causal convolution is used), a batch normalization layer is added after each convolution layer, and after each convolution layer, the input and output are added through residual connection, and the last layer is a fully connected layer.

[0092] In one implementation, data is standardized through mean normalization, which eliminates the scale differences of the original data and balances the impact of different data dimensions. This helps the model learn the underlying patterns of the data more effectively, reduces the deviation between different features, and improves the stability and prediction effect of the model.

[0093] In one implementation, using an encoder to extract temporal features can help the model understand the complex patterns and underlying regularities in time series. In particular, the encoder in a deep learning model can usually automatically identify useful information in time series data, and the model can better capture the potential dependencies in traffic data. TCN can process time series data and retain the effects of longer time lags, which helps capture changes in traffic changes and thus improve the accuracy of traffic prediction.

[0094] In one implementation, performing a logarithmic transformation on the normalized data can effectively reduce data volatility, making the data more consistent with a normal distribution. This helps improve the model's learning ability. The deseasonalization process removes the impact of seasonal changes, helping the model focus more on long-term trends in the data or other non-seasonal factors, thereby more accurately predicting future traffic changes.

[0095] In one embodiment, after optimizing the uplink and downlink rates according to the target predicted traffic, the method further includes:

[0096] Obtain the optimized results and verify them to obtain the correction factor;

[0097] Calculate the predicted flow rate of the next cycle after optimization, and correct the predicted flow rate of the next cycle after optimization according to the correction factor to obtain the target corrected flow rate;

[0098] Optimize uplink and downlink rates based on target corrected traffic.

[0099] In one implementation, by obtaining the optimized results and verifying them to obtain a correction factor, the model's prediction results can be further adjusted to make them more consistent with the changing trend of actual traffic. The calculation of the correction factor is based on the verification of actual data, which can effectively correct the prediction deviation and ensure that the predicted traffic in the next cycle is more accurate.

[0100] In one implementation method, the optimized result is obtained and verified as follows: the traffic data obtained by the LET base station is obtained (the data here is the data corresponding to the next cycle of the current moment mentioned above, and the next cycle of the current moment is recorded as the second moment) and recorded as the second traffic data. The target predicted traffic is subtracted from the second traffic data and divided by the target predicted traffic to obtain the error value (correction factor). If the error value is positive, the coefficient 1 is obtained by subtracting the error value from 1, and the target corrected traffic is obtained by multiplying the predicted traffic of the next cycle after optimization (the next cycle of the second moment) by the coefficient 1; if the error value is negative, the coefficient 2 is obtained by adding the error value to 1, and the target corrected traffic is obtained by multiplying the predicted traffic of the next cycle after optimization (the next cycle of the second moment) by the coefficient 2; if the error value is zero, no change is made.

[0101] In one implementation, a correction factor is used to correct the predicted traffic for the next cycle, making the prediction more flexible and adaptable to changing network conditions. During the optimization process, the target corrected traffic is used as a new input for uplink and downlink rate optimization, enabling rate allocation to better meet actual traffic requirements.

[0102] In one implementation, the optimized uplink and downlink rates are adjusted again based on the target corrected traffic, which helps balance the uplink and downlink allocation of network resources, thereby ensuring optimal system resources while improving the user experience. Through accurate prediction and correction, over-reservation or insufficient resources can be avoided, improving overall network performance.

[0103] Based on the same inventive concept, the embodiment of the present invention also provides an uplink and downlink rate optimization system based on an LTE base station. Figure 2 , Figure 2 A framework diagram of an uplink and downlink rate optimization system based on an LTE base station provided in an embodiment of the present invention includes:

[0104] A preprocessing module is used to collect the uplink and downlink traffic received by the target LTE base station in real time to obtain initial traffic data, and preprocess the initial traffic data to obtain preprocessed data;

[0105] The first traffic prediction data module is used to obtain historical traffic data and predict the uplink and downlink traffic of the next cycle at the current moment based on the historical traffic data and pre-processed data to obtain first predicted traffic data;

[0106] A second predicted traffic data module is used to substitute the pre-processed data into a preset model to obtain second predicted traffic data;

[0107] A target predicted flow determination module, configured to obtain a target predicted flow based on the first predicted flow data and the second predicted flow data;

[0108] The first rate optimization module is used to optimize the uplink and downlink rates according to the target predicted traffic.

[0109] An uplink and downlink rate optimization system based on an LTE base station provided by an embodiment of the present invention can obtain the changing trend of traffic in real time by collecting uplink and downlink traffic in real time and performing preprocessing. By combining traffic prediction based on historical data and preset models, network parameters can be dynamically adjusted, thereby improving optimization efficiency.

[0110] In one embodiment, the first traffic prediction data module includes:

[0111] The traffic screening module is used to obtain the time characteristics and state characteristics of the pre-processed data, and screen the historical traffic data according to the time characteristics and state characteristics to obtain the target historical traffic data set;

[0112] An initial prediction curve determination module is used to obtain an initial prediction curve by performing uplink and downlink flow curve fitting on each target historical flow data in the target historical flow data set;

[0113] The curve screening module is used to perform curve fitting on the pre-processed data to obtain the existing flow curve, and to screen all the initial prediction curves according to the existing flow curve to obtain the effective prediction curve set;

[0114] The first predicted traffic data generation module is used to obtain a target prediction curve based on the effective prediction curve set, obtain the uplink and downlink traffic of the next cycle at the current moment based on the target prediction curve and the existing traffic curve, and record the uplink and downlink traffic as the first predicted traffic data.

[0115] In one embodiment, the curve screening module includes:

[0116] The curve similarity value calculation module is used to obtain the similarity value corresponding to each initial prediction curve by calculating the DTW similarity between the existing traffic curve and the initial prediction curve for all initial prediction curves;

[0117] The effective prediction curve set generation module is used to obtain the initial prediction curves whose similarity values are greater than a preset threshold to obtain the effective prediction curve set.

[0118] In one embodiment, the second traffic prediction data module includes:

[0119] Normalization module, used to segment the preprocessed data according to a preset time period, and perform mean normalization on all segmented data to obtain a normalized data set;

[0120] A time dataset determination module is used to perform logarithmic transformation on all data in the normalized dataset and then deseasonalize it to obtain a time dataset;

[0121] The feature extraction module is used to substitute the time data set into the encoder to obtain time features, and substitute the time features into the TCN prediction model to obtain the second predicted traffic data.

[0122] In one embodiment, the system further comprises:

[0123] Correction factor determination module, used to obtain the optimized results and verify them to obtain the correction factor;

[0124] The target corrected flow determination module is used to calculate the predicted flow of the next cycle after optimization, and correct the predicted flow of the next cycle after optimization according to the correction factor to obtain the target corrected flow;

[0125] The second rate optimization module is used to optimize the uplink and downlink rates according to the target correction flow.

[0126] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for optimizing uplink and downlink rates based on an LTE base station, characterized in that: The method comprises: Collecting uplink and downlink traffic received by the target LTE base station in real time to obtain initial traffic data, and preprocessing the initial traffic data to obtain preprocessed data; Acquire historical traffic data, and predict the uplink and downlink traffic of the next cycle at the current moment based on the historical traffic data and the preprocessed data to obtain first predicted traffic data; Substituting the preprocessed data into a preset model to obtain second predicted flow data; Obtaining a target predicted flow rate according to the first predicted flow rate data and the second predicted flow rate data; Optimize uplink and downlink rates according to the target predicted traffic; Predicting the uplink and downlink traffic of the next cycle at the current moment according to the historical traffic data and the pre-processed data to obtain first predicted traffic data includes: Acquiring the time characteristics and state characteristics of the pre-processed data, and filtering the historical traffic data according to the time characteristics and the state characteristics to obtain a target historical traffic data set; For each target historical flow data in the target historical flow data set, performing uplink and downlink flow curve fitting on the target historical flow data to obtain an initial prediction curve; Performing curve fitting on the preprocessed data to obtain an existing flow curve, and screening all initial prediction curves according to the existing flow curve to obtain an effective prediction curve set; Obtain a target prediction curve according to the effective prediction curve set, obtain the uplink and downlink traffic of the next cycle at the current moment according to the target prediction curve and the existing traffic curve, and record the uplink and downlink traffic as first predicted traffic data; All initial prediction curves are screened according to the existing flow curves to obtain a valid prediction curve set including: For all initial prediction curves, the similarity value corresponding to each initial prediction curve is obtained by calculating the DTW similarity between the existing flow curve and the initial prediction curve; Obtaining the initial prediction curves whose similarity values are greater than a preset threshold to obtain a valid prediction curve set; Substituting the pre-processed data into a preset model to obtain second predicted flow data includes: Segmenting the preprocessed data according to a preset time period, and performing mean normalization on all segmented data to obtain a normalized data set; Performing logarithmic transformation on all data in the normalized data set and then deseasonalizing to obtain a time data set; Substitute the time data set into the encoder to obtain time features, and substitute the time features into the TCN prediction model to obtain second predicted traffic data.

2. The method for optimizing uplink and downlink rates based on an LTE base station according to claim 1, wherein: After optimizing the uplink and downlink rates according to the target predicted traffic, the following steps are further included: Obtain the optimized results and verify them to obtain the correction factor; Calculating the predicted flow rate of the next cycle after optimization, and correcting the predicted flow rate of the next cycle after optimization according to the correction factor to obtain a target corrected flow rate; The uplink and downlink rates are optimized according to the target corrected traffic.

3. The uplink and downlink rate optimization system based on an LTE base station according to the uplink and downlink rate optimization method based on an LTE base station according to claim 1 or 2, characterized in that: The system comprises: A preprocessing module is used to collect the uplink and downlink traffic received by the target LTE base station in real time to obtain initial traffic data, and preprocess the initial traffic data to obtain preprocessed data; A first traffic prediction data module is used to obtain historical traffic data, and predict the uplink and downlink traffic of the next cycle at the current moment based on the historical traffic data and the pre-processed data to obtain first predicted traffic data; A second predicted traffic data module, configured to substitute the pre-processed data into a preset model to obtain second predicted traffic data; a target predicted flow determination module, configured to obtain a target predicted flow according to the first predicted flow data and the second predicted flow data; The first rate optimization module is used to optimize the uplink and downlink rates according to the target predicted traffic.

4. The uplink and downlink rate optimization system based on LTE base station according to claim 3, characterized in that: The first traffic prediction data module includes: A traffic screening module, configured to obtain the time characteristics and state characteristics of the pre-processed data, and screen the historical traffic data according to the time characteristics and the state characteristics to obtain a target historical traffic data set; An initial prediction curve determination module is used to perform uplink and downlink flow curve fitting on each target historical flow data in the target historical flow data set to obtain an initial prediction curve; a curve screening module, configured to perform curve fitting on the preprocessed data to obtain an existing flow curve, and screen all initial prediction curves according to the existing flow curve to obtain a valid prediction curve set; The first predicted traffic data generation module is used to obtain a target prediction curve based on the effective prediction curve set, obtain the uplink and downlink traffic of the next cycle at the current moment based on the target prediction curve and the existing traffic curve, and record the uplink and downlink traffic as the first predicted traffic data.

5. The uplink and downlink rate optimization system based on LTE base station according to claim 4, characterized in that: The curve screening module includes: A curve similarity value calculation module is used to obtain a similarity value corresponding to each initial prediction curve by calculating the DTW similarity between the existing flow curve and the initial prediction curve for all initial prediction curves; The effective prediction curve set generation module is used to obtain the initial prediction curves whose similarity values are greater than a preset threshold to obtain the effective prediction curve set.

6. The uplink and downlink rate optimization system based on LTE base station according to claim 3, characterized in that: The second traffic prediction data module includes: A normalization module, configured to segment the preprocessed data according to a preset time period, and perform mean normalization on all segmented data to obtain a normalized data set; a time data set determination module, configured to perform logarithmic transformation on all data in the normalized data set and then deseasonalize the data to obtain a time data set; The feature extraction module is used to substitute the time data set into the encoder to obtain time features, and substitute the time features into the TCN prediction model to obtain second predicted traffic data.

7. The uplink and downlink rate optimization system based on LTE base station according to claim 3, characterized in that: The system further comprises: Correction factor determination module, used to obtain the optimized results and verify them to obtain the correction factor; a target corrected flow determination module, configured to calculate the predicted flow of the next cycle after optimization, and correct the predicted flow of the next cycle after optimization according to the correction factor to obtain the target corrected flow; The second rate optimization module is used to optimize the uplink and downlink rates according to the target corrected traffic.

Images